Lecture Notes in

Topics in
Path Integrals and
String Representations

10129hc_9789813143463_tp.indd 1 23/9/16 12:01 PM

 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

 Lecture Notes in
Topics in
Path Integrals and
String Representations
Luiz C L Botelho
Federal Fluminense University, Brazil

World Scientific
NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TA I P E I • CHENNAI

10129hc_9789813143463_tp.indd 2 23/9/16 12:01 PM

 Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

Library of Congress Cataloging-in-Publication Data

Names: Botelho, Luiz C. L.
Title: Lecture notes in topics in path integrals and string representations /
Luiz C.L. Botelho (Federal Fluminense University, Brazil).
Description: New Jersey : World Scientific, 2017.
Identifiers: LCCN 2016054695 | ISBN 9789813143463
Subjects: LCSH: Path integrals. | Integral representations. | Integration, Functional. |
Gauge fields (Physics)
Classification: LCC QC174.17.P27 B6745 2017 | DDC 530.12--dc23
LC record available at https://lccn.loc.gov/2016054695

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

Copyright © 2017 by World Scientific Publishing Co. Pte. Ltd.

All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance
Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy
is not required from the publisher.

Printed in Singapore

ZhangJi - Lecture Notes in Topics in Path Integrals and String Representations.indd 1 21-12-16 2:25:15 PM
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page v

To my aphrodites, Mermaids

and ninphes, including “Hanna Nelma” – my wife

v
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page vii

Preface

Since the years 1978–1984, the search for an analytical formalism for han-
dling Non Abelian Gauge Theories through String Path integrals has been
pursued with some success by strong interactions Theoretical-Mathematical
Physicists. The basic idea is very simple: write loop wave equations for
the SU (∞) quantum Wilson loop, at least in a formal Mathematical level
in relation to the rigorous mathematical meaning of the loop functional
derivatives and to the rigorous (perturbative or not) Euclidean quantum
Yang-Mill measure – and try to solve it – at least for some classe of suit-
able surfaces; through a string path integral.
These ideas have been throughly exposed in my previous monograph
entitled Methods of Bosonic and Fermionic Path Integral Representations
– Continuous Random Geometry in Quantum Field Theory. Nova Science
Publishers - Inc. NY, 2009 and references therein.
My aim in this another set of informal lecture notes on the subject
is to present author’s original research material and developments on this
problem done after 2007; Mainly based on revised and amplified on author’s
published papers on international scholar journals during 2007–2015.
The methodology used to write my set of lecture notes is the same of
our previous lecture notes sets/research monographs ([1]): My exposition
is intend to be quick and specially expected to be thought – provoking. It
is informal, with graduate students on mind and the application – oriented
style is expected also to estimulate the involviment of Mathematicians in
solving and analyzing the problem on the subject.
Cumbersome use of English and certainly types and spelling mistakes
reflects the author’s limitations on the using of advanced English grammar.
The reader’s criticism and comments are welcome.

Luiz Carlos Lobato Botelho

vii
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page viii

viii Lecture Notes in Topics in Path Integrals

CNPq- Visiting Research – Grant 164438/2015-7 – IMECC-Universidade

de Campinas, São Paulo, Brazil.

[1] Botelho, Luiz C.L., Methods of Bosonic and Fermionic Path Inte-
grals Representations, Continuum Random Geometry in Quantum
Field Theory, Nova Science Publishers, Inc., New York-USA, ISBN:
978-1-60456-068, (2009).

PS. Special thanks to Professor Waldyr Rodriguez by Sponsoring my

research stay at UNICAMP.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page ix

Contents

Preface vii

1. Bosonic Free Strings and Non-supersymetric

QCD(SU (∞))
(∞)): A constant gauge field path integral
study 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The Static Confining Potential for the Eguchi Kawai Model
on the Continuum . . . . . . . . . . . . . . . . . . . . . . 3
1.3 The Luscher correction to inter quark potential on the re-
duced model . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Some path integral dynamical aspects of the reduced QCD
as a path integral dynamics of euclidean strings . . . . . . 12
1.5 Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2. Basics Polyakov’s quantum surface theory on the

formalism of functional integrals and applications 25
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Elementary results on the Classical (Bosonic) Surface theory 26
2.3 Path integral quantization on Polyakov’s theory of surface
(or how to quantize 2D massless scalar fields in the presence
of 2D quantum gravity) . . . . . . . . . . . . . . . . . . . 32
2.4 Path-Integral quantization of the Nambu-Goto theory of
random surfaces . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

ix
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page x

x Lecture Notes in Topics in Path Integrals

2.6 Appendix A: 2D Abelian Dirac Determinant On the Formal

Evaluation of the Euclidean Dirac Functional Determinant
on Two-dimensions . . . . . . . . . . . . . . . . . . . . . . 42
2.7 Appendix B: On Atyah-Singer Index Theorem in the
Framework of Feynman Pseudo-Classical Path Integrals -
Author’s Original Remarks . . . . . . . . . . . . . . . . . 44
2.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.9 Appendix C: Path integral bosonisation for the Thirring
model in the presence of vortices . . . . . . . . . . . . . . 48
2.10 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.11 Appendix D: . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.12 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.13 Appendix E: Path-Integral Bosonization for the Abelian
Thirring Model on a Riemann Surface – The QCD(SU (N ))
String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.14 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

3. Critical String Wave Equations and the QCD(U (Nc )) String 65

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2 The Critical Area-Diffusion String Wave Equations . . . . 65
3.3 A Bilinear Fermion Coupling on a Self-Interacting Bosonic
Random Surface as Solution of QCD(U (Nc )) Migdal-
Makeenko Loop Equantion . . . . . . . . . . . . . . . . . . 69
3.4 Appendix A: A Reduced Covariant String Model for the
Extrinsic String . . . . . . . . . . . . . . . . . . . . . . . . 74
3.5 Appendix B: The Loop Space Program i the Bosonic λφ4 −
O(N )-Field Theory and the QCD Triviality for RD . D > 4 77
3.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

4. The Formalism of String Functional Integrals for the

Evaluation of the Interquark Potential and Non Critical
Strings Scattering Amplitudes 83
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.2 Basics Results on the Classical Bosonic Surface Theory and
the Nambu-Goto String Path Integral . . . . . . . . . . . 84
4.3 The Nambu-Goto Extrinsic Path String . . . . . . . . . . 91
4.4 Studies on the perturbative evaluation of closed Scattering
Amplitude in a Higher order Polyakov’s Bosonic String Model 96
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page xi

Contents xi

4.5 Appendix A: The distributional limit of the Epstein

function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.6 Appendix B: Integral Evaluation . . . . . . . . . . . . . . 104
4.7 Appendix C: On the perturbative evaluation of the
bosonic string closed scattering amplitude on Polyakov’s
framework . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

5. The D → −∞ saddle-point spectrum analysis of the

open bosonic Polyakov string in RD × SO(N ) – The
QCD(SU (∞)) string 109
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.2 The non-tachyonic spectrum and scalar amplitudes at
D → −∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

6. The Electric Charge Confining in Abelian Rank two

Tensor Field Model 117
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 117
6.2 The interquark potential evaluation . . . . . . . . . . . . . 117
6.3 Appendix A: The dynamics of the QCD(SU (∞)) tensor
fields from strings . . . . . . . . . . . . . . . . . . . . . . . 121
6.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.5 Appendix B: Path-integral bosonization for a nonrenormal-
izable axial four-dimensional fermion model . . . . . . . . 124
6.6 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.7 The model . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.8 Appendix C . . . . . . . . . . . . . . . . . . . . . . . . . . 129
6.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

7. Infinities on Quantum Field Theory: A Functional

Integral Approach 133
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 133
7.2 Infinities on Quantum Field Theory on the Functional
Integral Formalism . . . . . . . . . . . . . . . . . . . . . . 134
7.3 On the cut-off remotion on a two-dimensional Euclidean
QFT model . . . . . . . . . . . . . . . . . . . . . . . . . . 139
7.4 On the construction of the Wiener Measure . . . . . . . . 145
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page xii

xii Lecture Notes in Topics in Path Integrals

7.5 On the Geometrodynamical Path Integral . . . . . . . . . 147

7.6 Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . 150
7.7 Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . 151
7.8 Appendix C . . . . . . . . . . . . . . . . . . . . . . . . . . 153
7.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

8. Some comments on rigorous finite-volume euclidean

quantum field path integrals in the analytical regular-
ization scheme 157
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 157
8.2 Some rigorous finite-volume quantum field path integral in
the Analytical regularization scheme . . . . . . . . . . . . 158
8.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
8.4 Appendix A: Some Comments on the Support of Functional
Measures in Hilbert Space . . . . . . . . . . . . . . . . . . 168
8.5 Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . 171

9. On the Rigorous Ergodic Theorem for a Class of Non-

Linear Klein Gordon Wave Propagations 173
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 173
9.2 On the detailed mathematical proof of the R.A.G.E.
theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
9.3 On the Boltzman Ergodic Theorem in Classical Mechanics
as a result of the R.A.G.E theorem . . . . . . . . . . . . . 177
9.4 On the invariant ergodic functional measure for non-
linear Klein-Gondon wave equations with kinetic trace class
operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
9.5 An Ergodic theorem in Banach Spaces and Applications to
Stochastic-Langevin Dynamical Systems . . . . . . . . . . 184
9.6 Appendix A: The existence and uniqueness results for some
polinomial wave motions in 2D . . . . . . . . . . . . . . . 187
9.7 Appendix B: The Ergodic theorem for Quantized wave
propagations . . . . . . . . . . . . . . . . . . . . . . . . . 191
9.8 Appendix C: A Rigorous Mathematical proof of the
Ergodic theorem for Wide-Sense Stationary Stochastic
Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
9.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page xiii

Contents xiii

10. A Note on Feynman-Kac Path Integral Representations

for Scalar Wave Motions 197
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 197
10.2 On the path integral representation . . . . . . . . . . . . . 198
10.3 Appendix A: The Acoustic Case . . . . . . . . . . . . . . . 209
10.4 Appendix B: A Toy model for stable numerics on wave
propagation . . . . . . . . . . . . . . . . . . . . . . . . . . 211
10.5 Appendix C . . . . . . . . . . . . . . . . . . . . . . . . . . 213
10.6 Appendix D: The Causal Propagator – The Retarded
Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
10.7 Appendix E: The Causal Propagator – The Damped Case 216
10.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

11. A Note on the extrinsic phase Space path Integral

Method for quantization on Riemannian Manifold Par-
ticle Motions - An application of Nash Embedding Theorem 219
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 219
11.2 The Phase Space Path Integral Representation . . . . . . 220
11.3 References . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

Index 227
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 1

Chapter 1

Bosonic Free Strings and

Non-supersymetric QCD(SU (∞)) (∞)):
A constant gauge field path integral
study

We study in a reduced dynamical model for QCD(SU (∞)), defined by con-

stant gauge fields Yang-Mills path integral, several concepts on the validity
of string representations on QCD(SU (∞)) and the confinement problem.

1.1 Introduction

In the last years approaches have been pursued to reformulate non super-
symetric quantum chromodynamics as a String Theory ([1], [2], [3]) and
thus handle the compound hadron structure in the QCD model for strong
interactions ([3]). The common idea of all those attempts is to represent
the full quantum ordered non supersymetric phase factor as a string path
integral, which certainly takes into account more explicitly the geometrical
setting of the non abelian gauge theory than its usual description by gauge
potential.
Other main protocol to achieve such string representation for the wilson
loop operator in QCD is to use the still not completely understand large
number of colors of t’Hooft for non supersymetric quantum Yang-Mills
theory.
It is the purpose of this chapter to evaluate the static potential between
two static charges with opposite signal on the approach of an effective
reduced quantum dynamics of Yang-Mills constant-gauge fields ([3]) these
results surely expected to be relevant for the validity of the old conjecture
of E. Witten about a QCD(SU (∞)) dynamics dominated by a constant
gauge SU (∞) master field configuration ([4], [5], [6]). These studies are
presented in section 2 of this chapter. In section 3, we present the relevant

1
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 2

2 Lecture Notes in Topics in Path Integrals

QCD(SU (∞)) loop wave equation for our reduced model of constant –
gauge fields for QCD(SU (∞)) and suggest that, a free bosonic string as
solution for this reduced Loop Wave Equation ([7]). We continue with our
study and present also a detailed calculation of the quark-antiquark static
potential from a one-loop approximation on the Regge slope string constant
directly from the well-known Nambu-Goto string path integral ([8]). 1
Finally in section 4, we present also somes studies on the dynamical
aspects of this framework of constant field model by presenting path integral
studies on evaluation of vectorial-scalar color singlet quark currents ([6],[8]).
Before to proceed, let us firstly reproduce two enlightneen discourses on
the present day problem of handle quantitatively Yang-Millls fields outside
the lattice approximation
1 – Quoted from A. Jaffe and Eduard Witten.
Classical properties of non abelian gauge theory are within the reach of
established mathematical methods, and indeed, classical non abelian gauge
theory has played a very important role in pure mathematics in the last
twenty years, especially in the study of three- and four-dimensional (C ∞ -
differentiable) manifolds.
On the other hand, one does not yet have a mathematically complete
example of a quantum gauge theory in four-dimensional space-time, not
even a non abelian quantum gauge theory in four-dimensional.
2 – Related to the pure string holographic approach based on the Mal-
dacena conjecture and Super String Theory it appears interesting to cite
V. Rivasseu (Math-ph/0006017) about the general philosophy underlying
supersymmetric strings.
Today the main strean of theoretical physics holds the view that field
theory is only an effective (approximated!) theory and that superstring
or its variant, M -theory are the best candidate for a fundamental global
theory of nature (including QCD).
However this superstring theory has not yet received direct experimental
confirmation; it has (surely) opened up a new interface with mathemati-
cians, mostly centered around concepts and ideas of geometry and topology
(of C ∞ -manifolds), with algebra and geometry dominating over analysis
and calculational aspects.
Fortunately there is Lattice Gauge Theory, which although has re-
mained largely phenomenological, it has produced somewhat “precise” re-
sults on Experimentall Hadron mass spectroscopy as it has been pointed
1 It is worth to call attention that our constant gauge fields at SU (∞) are not the

rigorous continuum version of the one-plaquette the Eguchi-Kawai lattice model.

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 3

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 3

out by F. Wilczek (Nature 456,449, 2008). Being enough for that by just
taking some mesons (π, μ, Σ) mass inputs, even if in the context of QED
e2 1
one needs as input only the fine structure α = ∼ ·
4πc 137
As again in this chapter we try to implement the QED one universal
protocol for QED, by taking now our reduced model as the QCD effective
theory at large Nc and as universal imput parameter, the Gluonic conden-
sate 0|Tr(F 2 )|0SU(∞) ([7]).

1.2 The Static Confining Potential for the Eguchi Kawai

Model on the Continuum

The basic gauge-invariant observable on probing the non-perturbative vac-

uum ([1]) of SU (N ) Euclidean Yang-Mills bosonic field theory on R4 is the
Wilson loop quantum average

dμ[A] W [C]
W [C] =  (1.1)
dμ[A]
where the loop parallel transport SU (N )-valued matrix is given by
   
1
W [C] = TrSU(N ) P exp ig Aμ (x) dxμ (1.2)
N C
and dμ[A] denotes the Yang-Mills path-integral measure given formally by
the Feynman prescription

1
dμ[A] = (dA(x)) × exp − 2
tr(Fμν (A))(x)d4 x . (1.3)
4
4 R D
x∈R

The Gauge connection Aμ (x) and the field strenght Fμν (x) are explicitly
given by
 
Fμν (x) = ∂μ Aν − ∂ν Aμ + ig[Aμ , Aν ]− (x)
Aμ (x) = Aaμ (x)λa (1.4)
 
The SU (N ) generators λa a=1,...,N 2 −1 , are supposed to be Hemiteans and
satisfying the will known matrix relationship below
 a b
λ , λ = ifabc λc
  δab
Tr λa λb =
2
f abc f dbc = N δ ad (1.5)
In our proposal for the Euchi-Kawai model on continuum, we introduce
the space-time trajector C(R,T ) of a static quark-antiquark pair, separated
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 4

4 Lecture Notes in Topics in Path Integrals

apart a distance R with a (Euclidean) temporal evolution 0 ≤ x0 ≤ T

 
(R4 = {(x0 , x ), x ∈ R3 ).
We face thus the problem of evaluating the path integral eq. (1.1) for
constant gauge fields configurations at the large Nc limit ([5], [6]) and at
the physical space-time R4 .
Let us briefly review our previous framework.
We firstly consider the full infinite volume space-time R4 reduced to a
finite volume space-time Ω. This step has the effect to turns our “reduced”
path integrals mathematically well defined.
This finite volume space-time is supposed to be formed by the supper-
position of p four-dimensional hypercubes of caracteristic volume (“size”)
V = L4 .
The area S[C(R,T ) ] enclosed by our rectangle C(R,T ) is such that
 2
S[CR,T ) ] ∼ q L2 for large p. Obviously S[C] ≤ vol(Ω). So ours inte-
√
gers q and p should satisfy q ≤ p, an important bound to be kept on
mind on what follows.
Our improved large N -limit will be defined in such way by already
taking into account the basic phenomena of QCD-Yang-Mills dimensional
transmutation of the strong coupling constant, a fully non perturbative
phenomena similar to the Higgs mass mechanism on Weinberg-Salan the-
ory. We, thus, define the effective SU (∞) coupling constant through the
relationship for our space-time of finite volume

g2N q
lim < ∞. (1.6)
N →∞ L2 p

It is worth observe that for p → ∞ the infinite volume limit should be

taken according the underlying Nc → ∞ limit. Namely

q
≡a→0 (1.7)
p

Na
lim g 2 = (g∞ )2dim (1.8)
N →∞ L2
a→0

Here L2 is a physical finite area parameter expected to be related to the

domain area size of the famous QCD spaghetti vacuum ([7]).
After these preliminaries remarks, we must solve the invariant constant
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 5

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 5

gauge field SU (∞) matrix path integral below written:

⎛ ⎞
  +∞ N 2 −N D−1 
  1
WSU(∞) C(R,T ) = lim × ⎝ d Aaμ ⎠
N →∞ W (0) −∞ a=1 μ=0

g2  2
× ΔF p [Aμ ] exp + V Tr Aμ , Aν −
4
|g 2 S[C(R,T ) ]|2 1  
× exp · TrSU(N ) [A0 , A1 ]2 ·
2 N
(1.9)
In order to evaluate the SU (N )-invariant constant Gauge field path in-
tegral eq. (1.9), we use the Bollini-Giambiagi Cartan matrix decomposition
([5])
Aμ = Bμa Ha + Gbμ Eb (1.10)

where the Cartan basis {Ha , Ea } of the SU (N ) Lie algebra possesses the
special calculations properties ([5], [6])
a) For a, b = 1, 2, . . . , N − 1
 
Ha , Hb − = 0. (1.11)

N (N − 1)
b) For b = ±1, . . . , ±
2
 
Ha , Eb − = ra (b) Eb . (1.12)

N (N − 1)
c) For a = 1, 2, . . . , ·
2
−1
  N
Ea , E−a = rc (a) Ha . (1.13)
=1

N (N − 1)
d) For a = −b, a, b = ±1, . . . , ±
2
 
Ea , Eb − = Nab Ea+b . (1.14)

In this distinguished Lie Algebra basis, one can easily fix the Gauge
on the SU (N )-valued invariant matrix path integral by simply choosing all
the N -abelian components Bμa on the connection eq. (2.10) to be vanished.
Namely
Bμa = 0. (1.15)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 6

6 Lecture Notes in Topics in Path Integrals

It is expected thus that the Faddev-Popov term ΔF p [Aμ ] should be

quenched at the N → ∞ limit eq. (1.6)-eq. (1.8) i.e. ([5])
lim ΔF p [Aμ ] → 1 (1.16)
N →∞

Any way we take the Faddev-Popov quenched determinant to unity in

this sort of approximate evaluation of ours. SU (∞)-matrix valued invari-
ant path integral (note that procedure to evaluate degrees of Freedom re-
duced path integrals is usually implemented when one handles for instance,
fermions degrees of Freedon on SU (N ) lattice path integral ([3]). We will
adopt such procedure here).
By assembling all the above results one gets the following outcome
eq. (1.9) defined now by SU (N ) constant gauge field configurations for
a general euclidean space-time RD from now on
⎧ ⎡ ⎤⎫
  1 ⎨ +∞ N 2 −N D−1 ⎬
W C(R,T ) = × ⎣ dGaμ ⎦
W [0] ⎩ −∞ a=1 μ=0
⎭

$ %  &'
1 a b c d [g 2 S][C(R,T ) ]
exp + Gμ Gν Gμ Gν Labcd g V + δμ0 δν1
2
N  (1.17)
2 2
2

The above matrix valued path integral can be easily exactly evaluated
through re-scalings, at large N : Namelly (see Appendix A for details).
a) For μ = 0, 1: Gaμ → Gaμ [g 2 V ]− 4 .
1

b) For μ = 0, 1;
% &−1/4
[g 2 S[C(R,T ) ])2 ]
Gμ → Gμ g +
a a 2
N  (1.18)
2

And leading thus to the exactly result

N2 − N 
 −
  2
(g S[C(R,T ) ]) 2
2
W C(R,T ) = lim g2V +
N →∞ N
2
2
⎛ ⎞

N−1
Labcd = ⎝ ri (a) r (c) δi δci −d δci −b ⎠
i,=1
    
+ Nab Ncd 1 − δai −b 1 − δci −d δa+b,−(c+d)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 7

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 7

[N 2 − N )(D − 2) ( (N2 − N )D 
 − −
× g2V 4 2
(g V ) 4 (1.19)

It yields thus, the following N → ∞ limit

⎛ ⎞ −N (N − 1)
  2
  ⎜ g 2 S 2 CR,T ) ⎟
W CR,T = lim ⎝1 + ⎜ ⎟
N →∞ N ⎠
V
2
  2 
g (N − 1)L2 S 2
= lim exp − (1.20)
N →∞ L2 V
For D = 4, we have on the context of our proposed SU (∞) infinite
volume limit eqs. (1.6)-(1.8), our R4 Wilson Loop “string” behavior.
 2 
  g (N − 1) 2 q 2 L4
W C(R,T ) = exp − L
L2 pL4
 2 
(N →∞) g N q
∼ exp − 2 2
(qL )
L p
(N →∞)  
∼ exp −(g∞ )2 RT (1.21)
From eq. (1.21), one obtains the confining quark-antiquark potential
1   
V (R) = lim − g W CR,T ) = (g∞ )2 R (1.22)
T →∞ T
leading to an attractive constant force “biding” the static pair of quarks
as originally obtained by K. Wilson on his lattice gauge - modelling ([3]).

1.3 The Luscher correction to inter quark potential on the

reduced model

In this section we intend to show that our proposed SU (∞) constant gauge
field theory leads to a free string theory path-integral. We thus evaluate ex-
plicitly through the string path-integral the next non-confining corrections
to the quark-antiquark potential eq. (2-22).
In order to argument an effective low energy QCD string representation
in this model, we are going to consider the loop have equation ([1]) for
constant gauge fields already on the continuum at large N limit.

Let us thus firstly consider general loops Cx x = Xμ (σ) μ=0,1,2,3 ; 0 ≤

σ ≤ 2π on R4 . It is well-known that formally we have the functional loop
derivative
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 8

8 Lecture Notes in Topics in Path Integrals

a)
  
  1
ψ Cx,x , Aμ (x) = Tr P eig Aμ (X(σ))dXμ (σ) (1.23)
N Cxx

b)

δ   ig   dX ν (σ)
ψ Cxx , Aμ (x) = Tr P Fμν Xx (σ)
δXμ (σ) N dσ
 
× exp ig Aμ (X(σ))dX μ (σ) (1.24)
Cxx
3
c)
δ2  
ψSU(N ) Cxx , Aμ (x)
δXμ (σ)δX μ (σ)

δ
= (Fμν (X(σ)))
δXμ (σ)
 + ,- .  
1 dX ν
= ig Tr P (∇Fμν )(X (σ))
β
(σ) exp ig Aμ dXμ )
N dσ Cxx
 
(ig)2
+ Tr P (Fμν Fμν )(X β (σ))
N
 
dX ρ dX ρ
× · (σ) exp ig Aμ dXμ ) (1.25)
dσ dσ Cxx
For constant gauge fields configurations the first term of the right-hand
δ
side of eq. (1.25) ( constant Fμν ) = 0! vanishes identically. So,
δXμ (σ)
after taking the path integral average of eq. (3.25) through the path integral
of constant gauge fields configurations eq. (1.9) and considering the usual
path integral factorization of a product of gauge invariant observable at
SU (∞), together with the formation of non-vanishing value of the Yang-
Mills energy on the non-trivial QCD vacuum one gets finally the following
loop wave equation for the quantum Wilson Loop our Loop in our reduced
SU (∞) gauge theory on R4 .
2π   μ

δ2
ψSU(∞) X (σ)] dσ
0 δXμ (σ) δX μ (σ)
3 The central hypothesis about the Constant Gauge Fields QCD(S ∪ (∞)) stringy

vacuum:
ΩVac |(Fμν F μα )(x)|ΩVac  = ΩVac |(Fμν F μν )|ΩVac .
να
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 9

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 9

(≥ 0)

/   0 2π 1  1  
2
= − g∞ (0) 0|F 2 |0SU(∞) × 1Xμ (σ)12 × ψSU(∞) X μ (σ)] dσ
0
(1.26)
Here, we have the SU (∞) Euclidean Gauge Theory parameters
identification a)
 2
g∞ (0) = lim (g 2 N )
N →∞

b)
1 2 1  1 3
δ ρα 0|F 2 |0SU(∞) = lim Tr 01 dD x Fμρ (x) F μα (x) 10 <0
N →∞ N
(1.27)
By comparing the above parameters with those coupling constants of the
static case,one has the following identification for the Spaghetti QCD non
perturbative broken scale invariance vacuum effective area domain with the
QCD value condensate
1  2 3
eff
= − 0|F 2 |0 > 0. (1.28)
a
A result already expected ([7]).
At this point we point out that the reduced loop wave equation is the
same of a free Bosonic string theory with the string Regge slope identifica-
tion with the reduced Gauge theory at SU (∞)
1  2
= − g∞ (0) 0|F 2 |0 (1.29)
(2πα )2
As a consequence, one should expects the phenomenological path-
integral representation between the large Nc and extreme low energy con-
tinuum QCD(SU (∞)) (represented by constant SU (∞) gauge fields), with
a free bosonic (creation process) string path integral on the light-cone gauge
2  3
ψ Cxx , Aμ (x) SU(∞) = G(Cxx , A)
low−energy

∞  X μ (σ,A)=Cxx (σ) 
= dA DF [X μ (σ(ζ)]
0 X μ (σ,0)=0

  
1 A 2π  2 1  2
× exp − dt dσ ∂ζ X μ +  2
∂σ X μ (1.30)
2 0 0 (2πα )
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 10

10 Lecture Notes in Topics in Path Integrals

   
Gstring Cxx , 0 = Gstring Cxx , ∞ = 0 (1.31)
At this point it is worth observe that our light-cone string path integral
propagator
 
Gstring Cxx , A = DF [X μ (σ, π)]
X μ (σ,0)=Cxx (σ)

  
1 A 2π  2 1  2
× exp − dt dσ ∂ζ X μ + ∂σ X μ (1.32)
2 0 0 (2πα )2
satisfies the area-Difusion euclidean Schrodinger loop functional equation:
 
∂Gstring Cxx , A
∂A
 2π 
δ2 1 
 
= dσ − |X (σ)| 2
Gstring Cxx , A (1.33)
0 δXμ (σ)δXμ (σ) (2πα )2 μ
togheter with the boundary conditons:
   
Gstring Cxx , 0 = Gstring Cxx , ∞ = 0. (1.34)
Let us now evaluate in details the quark-antiquark potential from the
general Nambu-Goto string, path integral in RD
   
ψ C(R,T ) = DF X μ (σ, ζ)
$ % &'
1 T R 4
× exp − dζ μ
dσ det(h(X (σ, ζ)) (1.35)
2πα 0 0

Here the orthogonal dynamical string vector position is considered as

closed quantum fluctuations from the static quark-antiquark trajected by
C(R,T ) i.e.:
a)
√
X μ (σ, ζ) = ζ ê1 + σê2 + πα Y μ (σ, ζ)
b)
⎧
⎪
⎨Y (σ, ζ ± T ) = Y (σ, ζ)
⎪ μ μ

Y μ (0, ±T ) = Y μ (0, 0)
⎪
⎪
⎩μ = 2, 3, . . . , D − 2

c)
     
h00 X μ (σ, ζ = ∂ζ X μ ∂ζ Xμ (σ, ζ) = 1 + πα ∂ζ Y μ ∂ζ Y μ (σ, ζ)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 11

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 11

d)
   
h01 X μ (σ, ζ = πα ∂σ Y μ ∂ζ X μ (σ, ζ)

e)
   
h11 X μ (σ, ζ = 1 + πα ∂σ Y μ ∂σ X μ (σ, ζ) (1.36)

As a consequence, we have explicitly the following one-loop order ap-

proximation for the string path integral weight eq. (1.35):
6   
1  1  ∂Y μ ∂Y μ ∂Y μ ∂Y μ  

μ
h X (σ, ζ) = 
1 + πα + ×0 (α )2
2πα 2πα ∂ζ ∂ζ ∂σ ∂σ
(1.37)
As a result of substituting eq. (1.36) on eq. (1.37), one gets the following
closed bosonic string Gaussian path-integral to evaluate
 
ψ C(R,T )
 
= DF Y μ (σζ)
Y μ (σ,ζ±T )=Y μ (σ,ζ)
$ '
RT 1 T R     
× exp − − dζ dσ Y μ
− Δ (R,T ) Yμ (σ, ζ)
2πα 2 0 0

(D−2)  
= e− 2πα det −
RT
2 − Δ (R,T ) (1.38)

where the Laplacian −Δ(R,T ) on the rectangle C(R,±1) has Dirichlet bound-
ary conditions and considered projected out from the zero modes.
It has been evaluated fully on the literature ([8]):
 
det − − Δ(R,T )
D−2
2

%  &
R
( D−2
2
∞ / 0−2(D−2)
+ πT − 2πn T
= e 6R (D−2) × 1−e R (R) (1.39)
T n=1

At this point, one can easily verify the string result for the quark-
antiquark potential:

1   1 (D − 2)π 1
V (R) = lim − n ψ C(R,T ) = R− ·
T →∞ T 2πα 6 R
+ ,- .
Luscher correction
(1.40)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 12

12 Lecture Notes in Topics in Path Integrals

1.4 Some path integral dynamical aspects of the reduced

QCD as a path integral dynamics of euclidean strings

Let us start this section on dynamical aspects by writing firstly in details,

the operational euclidean path integral expression for the non-relativistic
Feynman propagator of a spinless particle in the presence of an external
(euclidean) abelian gauge field and an external scalar potential.
As a first step let us write the Feynman propagator above cited in the
euclidean space-time R4 ( = 1)

1   1
2 1 1  e 2 1 3
1
G(x, y, t) = x 1exp −t − i∇ − A − i e ϕ + g V 1y (1.41)
2m c 1

Then A = (A1 , A2 , A3 ) denotes the time-dependent vectorial abelian

field, ϕ the field potential and V (x, , t), the external potential, also supposed
time dependent.
The phase-space path integral is easily written as of as

X(t)−x → →
G(x, y, t) = DF (X(σ)) DF [ p (σ)]
X(0)=y
⎧ ⎛ →
⎞ ⎫
⎨ t → ⎬
× exp +i ⎝P · dA ⎠ (σ) dσ
⎩ 0 dσ ⎭

$ %
t → 2
1 e→ →
× exp − P (σ) − A(X(σ))
0 2m c

→
− i e ϕ(X(σ), σ) + g V (X(σ), σ) (1.42)

After the formal path-integral change of variable into eq. (1.42)

→ e→ → →
P (σ) − A(X(σ), σ) = Q(σ) (1.43)
e
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 13

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 13

we get the following result:

→
X(t)=x → →
G(x, y, t) = →
DF (X(σ) DF [Q(σ)]
X(0)=y
⎧ →
⎫
⎨ t →
e→ dX(σ) ⎬
exp i Q(σ) + A(X(σ), σ 
⎩ 0 c dσ ⎭
 t → 
1
exp − (Q(σ))2
2m 0
 t 
→
exp +ie ϕ(X(σ), σ)dσ
0
 t 
→
exp − V (X(σ), σ)dσ (1.44)
0
→
After realizing the Gaussian Q(σ) functional⎡integral ⎤
 t →  →
→ 1  t→
dX(σ) ⎦
Q(σ) dσ exp ⎣i
2
DF [Q(σ)] exp − Q(X(σ))
2m 0 0 dσ
⎧ ⎛ → ⎞2 ⎫
⎪
⎨ m t dX ⎪
⎬
= exp − ⎝ ⎠ (σ) dσ , (1.45)
⎪
⎩ 2 0 dσ ⎪
⎭
we get our gauge invariant path integral expression for the euclidean Feyn-
man propagator under study ⎡ ⎤
→
⎛ → ⎞2
X(t)=x → t
⎢ m ⎝ dX ⎠ (σ) dσ ⎥
G(x, y, t) = → DF [X(σ)] exp ⎣− ⎦
X(0)=y 2 0 dσ
 t 
exp −g V (X(σ), σ) dσ
⎛0 ⎞
 t→ →
→
dX ⎠
exp ie ⎝ A(X(σ) · dσ
0 dσ
t 
→
+ ie ϕ(X(σ), σ) dσ (1.46)
0
The Gauge invariance of eq. (1.46) is under all reduced periodic Gauge
≤ y ; 0 ≤ z  ≤ t)
transformation (x ≤ z ⎧
⎪  
⎨A(z, t ) = (A + ∇Λ)(z, t )
⎪
ϕ(z, t ) = (ϕ + ∂Λ 
∂t (z, t )
(1.47)
⎪
⎪
⎩Λ(x, t) = Λ(y, 0) + 2πn
e
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 14

14 Lecture Notes in Topics in Path Integrals

Namely
 t t 
dX ∂Λ
exp ie (∇Λ)(X(σ), σ) dσ + ie dσ
0 dσ 0 dσ
 
t
d  
= exp ie Λ(X(σ), σ) dσ
0 dσ
  
= exp ie Λ(x, t) − Λ(y, 0) . (1.48)
D
In the euclidean quantum field case in R one must consider the gen-
erating fermionic case
1 
det 2 D∗ (A) D(A)
Z[Aμ ] = 1 , (1.49)
det 2 ( ∂ ∗ ∂)
which can be re-write through well-known propertion loop space techniques
as a loop space D-dimensionall non-relativistic propagator

  1 ∞ dt Xμ (ζ)=xμ
g Z[A]/Z(A = 0) = − dD xμ DF [Xμ (σ)]
2 0 t RD Xμ (0)=xμ
$ '
2
1 t dX μ
exp − (σ) dσ
2 0 dσ
  / t 
ie μ ν
Pspin PSU(N ) exp ie μ
Aμ (X (σ)) + [γ , γ ] Fμν (X (σ)) dσρ
,
0 4
(1.50)
where the symbols Pspin and PSU(N ) means σ-ordered matrixes indexes of
the spin-color gauge connection phase factor (the fermionic Wilson loop).
At very low energy region, one could consider as an effective theory, all
degree of Dirac spin of the particle, non-dynamical (i.e. frozen to scalar
values), or equivalently one can disregard the spin orbit shenght field cou-
ig μ ν
pling on eq. (1.50) [γ , γ ]Fμν (X β (σ)) ∼
=0 .

Let us now apply the above well-known remarks to evaluate approxi-
mately “scalar” composite operators quark-antiquark Green functions.
The effective connected generating functional for vectorial quark cur-
rents at very low energy (the strong coupling region of the underlying
Massless Yang-Mills theory) is given by the following loop expression
/ 0 9 :
1 
ef f
g ZQCD [Jμ ] = g det 2 D∗ (igAμ + Jμ ) D(igAμ + Jμ ) (1.51)
Aμ
there  Aμ denotes the complete Yang-Mills path integral, Aμ the Yang-
Mills field μand Jμ (x) the external source of the vectorial quark currents
Jμ (x)(ψγ ψ)(x) .
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 15

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 15

On the basis of the above discusions one has the following expression
for eq. (4.50), with the QCD scale ΛQCD already bult in a large SU (∞)
limit:
/ 0  ΛQCD
ef f 1 dt
g ZQCD [Jμ ]ΛQCD = −
2 1/ΛQCD t
 2
1 t dX
× dD zμ DF (X α (σ)) exp − (σ) dσ
X α (0)=X α (t)=z α 2 0 dσ
9  t : t
dX μ dX μ
× PSU(N ) exp ig Aμ (σ)dσ exp i Jμ dσ
0 dσ Aμ 0 dσ
(1.52)
The vectorial N -point bilinear quark current is given by in momentum
space
2 3
(ψγ μ1 ψ)(x1 ) . . . (ψγ μN ψ)(xμ ) Aμ
δ2 ; / 0 1
ef f 1
= g ZQCD (Jμ ) 1 = Gμ1 ...μN (x1 , . . . , xN )
δJμ1 (x1 ), . . . , δJμN (x1 ) Jμ ≡0
(1.53)
Or equivalently, after suitable Fourier momenta transforms.
 ΛQCD
  1 N dt t t
<
GΛQCD Pμ1 , . . . , PμN = − (i) dσ1 . . . dσN dD zμ
2 1/ΛQCD t 0 0
Xμ (t)=zμ
× DF [X(σ)]
Xμ (0)=zμ
 
t 2
1 dX
× exp − (σ)dσ
2 0 dσ

dXμ1 dXμN
× (σ1 ) . . . (σN )
dσ dσ
  N
 μ
× exp i pk Xμ (σk )
h=1
t
9 ig Aμ dX μ :
× e 0 (1.54)
SU(∞)

On the basis of eq. (1.54), one could envisage to try to evaluate eq. (1.53)
through an Gaussian (euclidean) string path integral. Let us take for
granted such string representation as a workable sound hypothesis on basis
of our previous studies.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 16

16 Lecture Notes in Topics in Path Integrals

The key point is to evaluate in terms of the loop variable X μ (σ), the
following anihillation string path integral:4

∞ 
  Y μ (σ,A)=0
W Xμ (σ), 0 ≤ σ ≤ t = dA DF (Y μ (σ, s))
0 Y μ (σ,0)=X μ (σ)
$   '
1 A t  
μ 2 1  
μ 2
exp − ds dσ ∂s Y + ∂σ Y (σ, s) (1.55)
2 0 0 (πα )

In order to evaluate eq. (1.55) exactly, let us firstly consider the stndard
re-scale
 1/2
σ −→ σ πα =σ
s −→ s
μ  1/4 μ
Y μ (σ, s) −→ Y (σ, s) ≡ πα (Y (σ, s)) (1.56)

which formally turns the string velocity into a overall factor into the path
integral weight

∞  μ
  Y (σ,A)=0  μ 
W X μ (σ) = dA DF Y (σ, s)
μ
0 Y (σ,0)=X μ (σ)
⎧ ⎡ ⎛  ⎞⎤⎫
⎨ 1 ⎣ A t(πα )1/2
∂Y
μ 2
∂Y
μ 2
⎬
× exp − ds dσ ⎝ + ⎠⎦
⎩ (2πα ) 0 0 ds dσ ⎭
(1.57)

After considering the “Brownian Bridge like” background loop-surface

decomposition which has the meaning of considering a toroidal like fluc-
tuating closed string world sheet Zμ (σ, ζ) bounded by the closed quark-
antiquark trajectory X μ (σ) X μ (σ + t) = X μ (σ), t, fixed loop proper-time

A−s √
Y μ (σ, s) = X μ (σ) + πα Z μ (σ, s)
A
Z μ (σ, A) = Z μ (σ, 0) = 0
Z μ (σ + t, s) = Z μ (σ, s), (1.58)

4 If 1
the action was A
0
ds t
0
dσ[(∂s Y μ )2 + (πα )2
(∂σ Y μ )2 ](s, σ), then eq. (1.56) would
μ
takes the form σ → σ = σ(πα ) and Y μ
(σ, s) = Y (σ, s).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 17

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 17

one gets the regularized proper-time string propagator

⎛ ⎞
∞  1
(πα ) 2 ·t 2
  (A/3) ⎝ dX μ (σ)
W X μ (σ) = dA exp − dσ ⎠
ε 2πα 0 dσ
⎛ ⎞
1
(πα ) 2 t  / / 00
1 ⎝ ⎠ −D
− (X μ (σ))2
dσ × det 2 −Δ 1
2πα A 0
(πα ) 2 t,A)

(1.59)
Just for completeness, we note the following exactly expressions for the
fluctucting worl-sheet Z μ Laplacean determinant and its Green function on
σ
+; ,- . + ,- .
ξ
 12
the rectangle 0, (πα ) t × [0, A]:
 % & −D/2
/ 0 2πn
2
2πm
2
−D
det 2 −Δ 1 = 1 +
((πα) 2 t,A)
n,m ((πα ) 2 t A
 D/2
(πα ) 2 t
1
π A
= exp D
A 6 (πα ) 12 t
∞   −2D
2πnA
× 1 − exp 1 ; (1.60-a)
n=1 (πα ) 2 t

(−Δ)−1 1 (σ, σ  , s, s )
((πα ) 2 t,A)
2πin(σ−σ )

+∞ 1
1 e (πα ) 2 t
=−
2 n,m 2πn
2
 2πm 2
−∞ 1 + A
(πα ) 2 t
 
2πm  2πm 
× cos (s − s ) − cos (s + s ) (1.60-b)
A A
As a consequence we get for N -point euclidean scalar meson Green
function after disregarding the contribution of the functional determinant
eq. (1.60-a) and by considering πα = 1 from now on
< (t) (P μ , . . . , P μ )
G 1 N
 ΛQCD  ∞
1 dt t t
=− × dσ1 . . . dσN dA
2 1/ΛQCD t 0 0 ε

 μ μ 
× F (Pk · Pk ), A, T, {σ1 , . . . , σN } . (1.61)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 18

18 Lecture Notes in Topics in Path Integrals

Where the quark-antiquark harmonic oscillator form factor coming from

eqs. (1.59), eq. (1.54) (with for notation simplicity πα = 1 and by consid-
dX μ1 (σ1 ) dX μN (σN )
ering the scalar case ... → 1) is given explicitly by
dσ dσ
the result:
 
F (Pkμ Ṗkμ ), A, t, {σ1 , . . . , σμ }
⎛ 6 ⎞D/2
2  √
(3 + 2A) A 3 A
⎝ ⎠
= 6 2  × exp − √ 6 2 
6π sin h A t 2(3 + 2A) sin h A t
⎡ ⎤
 = = 
⎢ μ
N
2 2 ⎥
×⎣ (Pk · Pkμ ) sin h (t − σk ) × sin h σk ⎦ (1.62)
A A
k=1
k =1
It is very important to remark that our “toy model” given by eq. (1.62)
has the correct structure to generate a Lorentz-invariant scattering ampli-
tude, after continuation to Minkowski space, on the light of the Hall and
Wightman theorem ([9]) (2πα = 1). Namely:
   
G(Λ(QCD,t) P1μ , . . . , PNμ = F(ΛQCD ) Pμi · Pμk (1.63)
One point now worth to be called the reader atteention for is that care
should be taken in applying straightforwardly the Feynman path integral
eq. (1.50) to represent the propagator of a particle possessing fermionic
degrees in the presence of an external Gauge field ([3]). One can avoid this
operational path integral procedure by squaring the fermionic determinant
and making use now of the well-defined proper-time formalism for bosonic
coloured particles ([3]). Namely (see eq. (1.51))
1
det 2 ( D(igA + J) D(igA + J)
1 
= det 2 D∗ (igA + J)D(igA + J)
ig μ ν
× det 1 − [γ , γ ](D∗ D)−1 (igA + J) × Fμν (igA + J) (1.64)
4
Since the Klein-Gordon bosonic propagator can be written in term of
the SU (N ) normalized holonomy factor as of as
 ∗ 
D (igAμ + Jμ )D(igAμ + Jμ ) (x1 , x2 )
⎧ ⎫
⎪
⎪
t  2
⎪
⎨ ∞ X(t)=x2 −2
1
X (σ) dσ ⎪
⎬
=N dt D[X(σ)] e 0
⎪
⎪ ⎪
⎪
⎩ 0 X(0)=x1 ⎭

× ψx1 x2 [C, A] × Φx1 x2 [C, J] (1.65)

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 19

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 19

Here the non-abelian dynamical and abelian vectorial external sources

phase factors are defined explicitly by
 t 
1
ψx1 x2 [C, A] = P exp ig μ
Aμ (X(σ))dX (σ)
N 0
 t 
μ
Φx1 x2 [C, J] = exp Jμ (X(σ))dX (σ) (1.66)
0
The final expression for the generating functional eq. (1.51) at large
N , is thus easily written in the proper-time formalism, before taking the
Yang-Mills path integral average is
> 1 ?
g det 2 D∗ (igA + J) D(igA + J)
 ∞
1 dt
=− d4 x1 dμ[Cx1 x2 ]
N 0 t

 
TrSU(N ) Ψx1 x2 [C, A] Φx1 x2 [C, J]

∞ n−1 n
1 1
−
n=2
2 1/N
∞
d4 x1 . . . d4 xn (dt1 . . . dtn )
dμ[Cx1 x2 ]
0
 
dμ[Cxn x1 ] × TrDirac [γμ1 , γν1 ] . . . [γμn , γνn ]

δ
× TrSU(N ) (ψx1 x2 [C, A]Φx1 x2 [C, A])
δσμ1 ν1 (x1 )

δ
··· (ψxn x1 [C, A]Φxn x1 [C, A]) (1.67)
δσμn νn (xn )
Here the Migdal-Makeenko loop derivative is introduced ([3])
ε
δ δ2
= lim dζ.ζ (1.68)
δσμν (X(σ)) ε→0 −ε δXμ (σ + ζ2 )δXν (σ − ζ2 )
which by its turn has the geometrical meaning of dividing the path tra-
jectory Cx1 x1 , quite closely analogous to the joining and splitting picture
of the old theory of dual strings (after taking the SU (∞) limit into our
constant gauge fields model as given by eq. (1.31) or eq. (1.35) in section 3
of this chapter).
Another point worth to call attention is the expansion parameter on
eq. (1.67) is the color N , but appearing now as a Laurent power series on
+n
1
.
1/N
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 20

20 Lecture Notes in Topics in Path Integrals

Conclusion: We can see from this work that another time in QCD physics
it is raised hopes that on underlying string dynamics is in place to han-
dle correctly the mathematical – calculational aspects of Euclidean Non-
Abelian Gauge – abelian Gauge theories in theirs confining phase, signaled
here by
 a hypothesizednon-vanishing energy for the non-perturbative vac-
uum 0|tr(F 2 )|0 = 0 . (See eq. (1.27) for the analytic expression of this
hypothesis).
At this point let us remark that our string representation for the QCD-
Eguchi-Kawai reduced model is a free bosonic one. However if one considers
next non-constant full space-time variable corrections/fluctuations to the
gauge connections entering into the full Yang-Mills path integrals, one is
lead to the self-avoiding fermionic full structure of the QCD(SU (∞)) ([3])
with the extrinsic string as an effective bosonic string representation for
QCD(SU (∞)) (chapter 3).
Finally, we should roughly say that our path integral is at SU (∞),
1
but surely we are in the context of a somewhat expansion for the pure
D
quantum Yang-Mills field, with a non perturbative vacuum. Unfortunately,
1
the famous expansion of Lattice QCD has not been generalized or even
D
well-understood on the continuum. We hope that our work should be a
step in this direction.

1.5 Appendix A

Let us consider the term

  
1 a b c d (g 2 S)2
J1 = exp G0 G1 G0 G1 Labcd × g 2 V + (A.1)
4 (N/2)
After the re-scaling
 − 1
f < f 2 (g 2 S)2 4
G0,1 = G0,1 g V + (A.2)
(N/2)
It terms out to be
 
1 <a <b < c <d
I1 = exp G G G G Labcd (A.3)
4 0 1 0 1
However a “mixed” term of the form
 
1 a b c d
I2 = exp G G G G Labcd (g V ) 2
(A.4)
4 0 2 0 2
under the re-scaling
  1
Gf2 = G <f g2 V − 4 (A.5)
2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 21

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 21

becomes now
⎧ ⎫
⎨1 (g V ) × (g V )
2 2 −1/2 ⎬
<b < c <d
<a G
I2 = exp G0 2 G0 G2 Labcd  2 s)2  (A.6)
⎩4 g 2 V + (gN/2 ⎭

Note that N → ∞, we have the leading asymptotic limit

(g 2 V )1/2 (g 2 V )1/2
 
2 S)2 1/2
∼ → 1. (A.7)
g 2 V + (gN/2 (g 2 V )1/2

It is worth recall that

⎧ ⎫⎛ ⎞
N 2 −N
 a  ⎨ (g 2
S)
 (N 22−N ) ⎬ N 2 −N
2 −
dG0 dGa1 = g2V + ⎝ <a dG
dG <a ⎠
⎩ N/2 ⎭ 0 1
a=1 a=1
(A.8)
⎛ ⎞
N 2 −A   N 2 −N
(N 2 −N ) ⎜ <a ⎟
dGaμ = (g 2 V )− 4 (D−2) × ⎝ dGμ⎠ (A.9)
a=1 a=1
μ=0,1 μ=0,1

We added also the remark about the constant gauge field non-abelian
Stokes theorem
/ /  0?
W [C] = TrSU(N ) P eig c Aμ dXμ
   
1 μν
= TrSU(N ) P eig Fμν ds
N
1 > / ??
= TrSU(N ) P e(ig)ig[Aμ ,Aν ]− S (δμ0 δν1 )
N  
N →∞ 1 (g 2 S)2
∼ TrSU(N ) 1 − g 2 S[A0 , A1 ]− + [A0 , A1 ]2− + . . .
N 2
 
N →∞ (g 2
S)2  
∼ exp + TrSU(∞) [A0 , A1 ]2− (A.10)
2N
As a last point of our Wilson loop evaluations at large N at the context
of constant gauge field configurations, we point out that at D = 2 (the two
dimensional case, it is not need to consider the phenomena of the dimen-
sional transmutation coupling constant and the evaluation above displayed
leads directly to the area behaviour for the Wilson Loop. (Remark origi-
nally due Bollini-Giambiagi). However it is important to keep in mind that
such result can be obtained quite straightforwardly by using the axial gauge
Aa0 = 0, and mostly important, it shows the non-dynamical behaviour of
the pure Yang-Mills quantum (perturbative) theory at two-dimensions. At
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 22

22 Lecture Notes in Topics in Path Integrals

D = 3, our SU (∞)-constant gauge field model yields charge screening in-

stead of color charge confinement (a lenght behaviour for the Wilson Loop).
Finally for D > 4, we have a infinite volume vanishing Wilson Loop, which
by its term signals that the Yang-Mills theory in RD , D > 4 is a trivial
QF T , in place of the usual wrong, but always argued non-renormalizability
of Yang-Mills theory for D > 4.

1.6 References

[1] Luiz C.L. Botelho - “Critical String Wave Equations and the
QCD(U (Nc )) String. (Some comments)”, Int. J. Theor. Phys.
(2009) 48 - 2715-2725.

[2] Luiz C.L. Botelho - “The Electric Charge Confining in Polyakov’s

Compact QED in R4 ”, Int. J. Theor. Phys. (2009) 48 - 1695-1700.

[3] Luiz C.L. Botelho - “Methods of Bosonic and Fermionic Path Inte-
gral Representations - Continuous Random Geometry in Quantum
Field Theory”, Nova Science, New York - (2008).

[4] T. Eguchi and H. Kawai - “Reduction of Dynamical Degrees of

Freedom in the Large N Gauge Theory”, Phys. Rev. Lett., vol.
48, 1063 (1982).

[5] Luiz C.L. Botelho - “The confining behavior and asymptotic free-
dom for QCD(SU (∞)) - a constant Gauge field path integral anal-
ysis”, Eur. Phys. J. 44, 267-276 (2005).

[6] Luiz C.L. Botelho - “Triviality - Quantum Decoherence of

Fermionic Chronodynamics SU (Nc ) in the Presence of an External
Strong U (∞) Flavored Constant Noise Field”, Int. J. Theor. Phys.
(2010) 49 - 1684-1692.

[7] M.N. Chernodubat all - “On chromoeletric (super) conductivity of

the Yang-Mills vacuum”, avXiv: 1212.3168 v 1 [hep-ph], 13 Dec.
2012.

[8] Claude Itzykson & Jean-Michel Drouffe - “Statistical field theory”,

vol. 2, Cambridge Monographs on Mathematical Phisics, 1991.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 23

Bosonic Free Strings and Non-supersymetric QCD(SU (∞))

(∞)): 23

- Luiz C.L. Botelho - Research Article - “Basics Polyakov’s Quan-

tum Surface Theory on the Formalism of Functional Integrals
and Applications”, International Scholarly Research Network
ISRN High Energy Physics, vol. 2012, Article ID674985, doi:
10.5402/2012/674985.

[9] N.N. Bogoluhov, A.A. Logunov and I.T. Todorov - Introduction to

Axiomatic Quantum Field Theory - Mathematical Physics Mono-
graph series - Benjamin/Cummings Publishing Company Inc. -
USA, 1978.
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 25

Chapter 2

Basics Polyakov’s quantum surface

theory on the formalism of functional
integrals and applications

2.1 Introduction

Since its inception on seventy years ago, Non-Abelian Gauge theories have
shown as the most promissing mathematical formalism for a realistic de-
scription of strong interactives and even formulated on its supersymmetric
version it has became an attractive attempt for unify Physics.
In strong interaction the picture image of a mesonic quantum excita-
tion, is for instance, a wave quantum mechanical functional assigned to a
classical configuration of a space-time trajectory of a pair quark-antiquark
bounding a space-time non-abelian gluon flux surface (of all topological
genera) connecting both particle pair: the famous t’Hoft-Feymman planar
diagrams ([1]).
It appears thus, appealing for mathematical formulations to consider di-
rectly as dynamical variables or wave functions in this Faraday line frame-
work for non-abelian gauge theories, the famous quantum Wilson Loop
or (quantum) holonomy factor associated to a given space-time Feynman
quark-antiquark closed trajectory C in a SU (Nc ) Yang-Mills quantum field
theory. Namely
9    :
1 (c)
ψ[C] = W [C] = TrSU(Nc ) P exp ig Aμ dxμ (2.1)
Nc C

Here   is defined by the Yang-Mills quantum path-integral and Aμ (x) =

Acμ (x)λa denotes the quantized Yang-Mills (Gluon) connection.
It is thus, searched loop space dynamical equations (at least on the
formal level without considering those famous ultra-violet “renormaliza-
tion problems”) for the new wave function Eq. (2.1). However Lattice
Yang-Mills theories has indicated that “string solutions” with a symbolic

25
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 26

26 Lecture Notes in Topics in Path Integrals

   
structure form
1
ψ[C] = exp − A(S) ΦQCD [C] (2.2)
2πα
@ S,∂s=C
with denoting the Feynman-Wiener continuous sum over all surfaces
S;∂S=C
S, with all topological genera and bounded by C; α is the fundamental
constant of strong force physics called the Regge slope parameter ([1]) A(S)
is the area functional “evaluated” at the sample surface S and Φ[C] is a
functional related probably to the existence of a (neutral) non-abelian 2D
intrinsic fermions structure on the surface S, called the Elfin fermionic
functional ([1]).
It is a consequence of eq. (2.1)-eq. (21.2) that non-abelian gauge theories
(even on theirs supersymmetric versions ([2])) should be better reformulated
as a dynamics of random surfaces (strings theories).
Our aim in this chapter is to present in full technical details added
with ours original improvements, the work of A.M. Polyakov ([1]) to give
a precise path integral meaning for eq. (2.2) in a 2D-gravitational context
and only considering the case of trivial surface topology.
We give (also in details) the path integral meaning for the usual case of
Nambu-Goto for eq. (2.2) in the pure surface context, all these new results
due to ourselves ([1]).
This chapter is organized as follows: in §1.2 we survey some basic results
of classical surface theory. In §1.3, we expose the Physical toy model exactly
soluble Polyakov’s framework. In §1.4, we expose the Nambu-Goto theory
(our results).

2.2 Elementary results on the Classical (Bosonic) Surface

theory

A given surface S, with a boundary being a curve C and embedded into

a euclidean space-time RD is usually described by a vector field Xμ (ξ1 , ξ2 )
(μ = 1, . . . , D) and a given two-dimensional domain Ω, compatible with
the topology of S in the case of the surface’s hypothesis smoothness. It is
imposed also that such vector field, called from now on as surface vector po-
sition field, when restricts to the boundary of Ω should be a reparametriza-
tion of C (Xμ (ξ1 , ξ2 ))∂Ω = C).
The Nambu-Goto area functional associated to a given surface vector
position (Xμ (ξ1 , ξ2 )) is given by usual Riemann integral ((ξ1 , ξ2 ) := ξ).
 1/2 2
A(SC ) = det{hab (ξ)} d ξ (2.3)
Ω
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 27

Basics Polyakov’s quantum surface theory on the formalism of functional 27

with hab (ξ) = (∂a X μ ∂b Xμ )(ξ) denoting the metric tensor induced on the
surface Sc (rigorously induced on the surface-manifold tangent bundle) by
the parametrization {Xμ (ξ)}.
The most important property of the functional eq. (2.3) is its invariance
under the (formal) group of the reparametrizations of S. Namely:
(ξ1 , ξ2 ) = (ϕ1 (ξ1 , ξ2 ); ϕ2 (ξ1 , ξ2 )
Xμ (ξ  ) = Xμ (ϕ(ξ1 )) (2.4)
here ϕ(ξ  ) denotes a two-dimensional intrinsic C1 vector field on Ω with
everywhere non-zero Jacobian.
Formal Euler-Lagrange equations associated to the surface action func-
tional eq. (2.3) can be easily written and producing the boundary value
problem on Ω for each μ-component (μ = 1, . . . , D)
1
1 Δh X μ (ξ) = 0
1 1
1 μ (2.5)
1 X (ξ))1ξ∈∂Ω = C μ
where Δh denotes the second order elliptic operator called Laplace-Beltrami
associated to the metric (hab (ξ)) (h(ξ) = det{hab (ξ)})
1 √
Δh = √ ∂a ( h hab db ) (ξ) (2.6)
h
If one now choose the conformal gauge for the surface hab (ξ) = eϕ(ξ) δab ,
formally one obtains that the surface vector position satisfies the Dirichlet
problem in Ω.
1
1 Δh=δ X μ (ξ) = 0
1 ab
1
1 μ (2.7)
1 X (ξ))1ξ∈∂D = C μ
The solution of the above mentioned potential problem can be exactly
given by conformal complex variable theory methods ([2] - Chapter 1).
Note that eq. (2.7) produces minimal surfaces (i.e. classical surfaces
which minimize the area functional locally) as the associated classical sur-
faces actions on the Nambu-Goto theory eq. (2.3). Note that all the above
displayed discussion remains correct to the case of general topological gen-
1  √
era (i.e. χ(S) = ( h R(h))(ξ) dξ 2 = 0; here R(h) denoting the scalar
4π Ω
of curvature associated to the induced metric hab (ξ)).
However at the classical level, there is a quadratic area functional due to
Howe-Brink-Polyakov equivalent to the above result related to the classical
aspects Nambu-Goto action. It is the functional associated to a theory of
D scalar classical massless fields on Ω but interacting with intrinsic (news)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 28

28 Lecture Notes in Topics in Path Integrals

dynamical degrees of freedom given by the infinite-dimensional manifold of

two-dimensional metrical structures on Ω.
The classically equivalent area action functional is now given by the
massless fields {X μ (ξ)} gravitation content
1 √
A(SC ) := ( g g ab ∂a X μ ∂b Xμ )(ξ)d2 ξ (2.8)
2 Ω
where {gab (ξ)}a=1,2 belongs to the infinite-dimensional space-manifold of
b=1,2
all possible metrics admissible by the surface S – the famous exactly soluble
string Polyakov’ model.
That more rich (from a dynamically point of view) surface action func-
tional is now invariant (formally) under the extended group of surface
reparametrizations (the group of local diffeomorphism of the surface S).
ξa − ξa = δξb = Ea (ξ)
δXμ (ξ) = Ea (ξ)∂ a Xμ (ξ)
δgab (ξ) = (∇a Eb + ∇b Ea )(ξ). (2.9)
Here the surface’s reparametrization vector fields {Ea (ξ)} are such that
(∇a Ea )(ξ) = 0, a necessasry restriction in order to take out from the above
written reparametrizations, all those which act on as a simply metrical re-
scalings. Note that we have imposed
1 further the vector field Ea (ξ) vanishing
at the boundary of S (i.e. Ea (ξ)1∂Ω ≡ 0). The covariant derivative is usually
defined by the Christoffel-Schwarzt objects associated to the given metric
tensor {gab (ξ)}.

(∇a Eb )(ξ) = (gaa ∇a )(Eb )(ξ) = (∂a Eb − Γcab Ec )(ξ) (2.10-a)
1 cd
Γcab (ξ) = {g (∂a gbd + ∂b gad − ∂a gab )}(ξ) (2.10-b)
2
In the important case of the metric field gab (ξ) has the conformal form
gab (ξ) = eϕ(ξ) δab , the above written objects take the simple forms below
written
1  
Γcab (ξ) = δ cd e−ϕ ∂a ϕ eϕ δbd + ∂b ϕ eϕ δ − ∂d ϕ eϕ δab
2
1 1 1
= (∂a ϕ)δ cb + (∂b ϕ)δ ca − ∂c ϕ δab (2.11-a)
2 2 2
1 1 
∇a Eb = ∂a Eb − (∂c ϕ)E c δab − (∂a ϕ)Eb − (∂b ϕ)Ea (2.11-b)
2 2
Note that one has in addition to the usual reparametrization invari-
ance, one further pivotal (point dependent) new symetry called the Weyl
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 29

Basics Polyakov’s quantum surface theory on the formalism of functional 29

conformal symetry which acts solely on the new degree of freedom (with
λ(ξ) > 0), crucial for the exactly modeel solubility.
gab (ξ) → λ(ξ)gab (ξ). (2.12)
The classical motion equations associated to the Howe-Brink-Polyakov
functional are easily written down:
Δgab X μ (ξ) = 0
1
X μ (ξ)1 = C μ
∂Ω

1
Tab (ξ) = (∂a Xμ ∂b X μ − gab g cd ∂c X μ ∂d Xμ )(ξ) = 0. (2.13)
2
From the last classical motion equation (2.13) for the intrinsic metric
field, one obtains the result
gab (ξ) = λ(ξ)(∂a Xμ ∂b X μ )(ξ) (2.14)
where the unknow scale λ(ξ) can be formally fixed to be the function λ(ξ) =
1 on the interior of Ω.
   
gab (ξ) ∂Ω = gab (ξ) ∂Ω = δab . (2.15)
Another important classical surface functional, probably related to the
still not completely understood functional ΦQCD [C] on eq. (2.2) is the so
called extrinsic functional which is defined by the square of the surface
mean curvature (see eq. (2.6)))
 2
1 √
(H2 (S)) = (Δh X μ )2 = − √ ∂a ( h hab ∂b )(X μ . (2.16)
h
Namely
√  
Fextr (S) =  (Δh X μ )2 (ξ) d2 ξ. (2.17)
Ω

In the Polyakov’s version of two-dimensional Ω quantum gravity, the

above functional can be replaced by the fourth-order scalar field action
√
Fext [X μ (ξ), gab (ξ)] = ( g(Δg X μ )2 (ξ))d2 ξ
Ω
√
+ ( gλab (ξ)(gab (ξ) − hab (ξ, X μ )))(ξ) (2.18)
Ω

where λab (ξ) denote lagrange multipliers insuring that at the classical level
gab (ξ) = (∂a X μ ∂b Xμ )(ξ).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 30

30 Lecture Notes in Topics in Path Integrals

Boundary conditions to be imposed on the complete action A(S) +

Fext (S) are mathematical subtle in order to guarantee the mathematical
good property of strong elliptic of the associated boundary value problem
([3] - Chapter 5).
For instance, if one consider the case gab = δab and the associated
biharmonic operator
∂4X μ ∂4X μ ∂4X μ
(ΔX μ )2 = 4 +2 2 2 + , (2.19-a)
∂ξ1 ∂ξ1 ∂ξ2 ∂ξ24
one can associate the following system of boundary operators
1
X μ (ξ)1 = C μ (ξ)
∂Ω
(2.19-b)
∂ 1
− X μ (ξ)1∂Ω = C μ (ξ).
∂n
Here the meaning of C μ is not so clear as an independent dynamical pa-
rameter of the loop C (string).
In the following discussions we will always regard the fourth-order ac-
tions as effective actions coming from the integrating out fermionic intrinsic
degrees of freedom ([1]).
We left to our readers to prove the Green’s formula for the fourth-order
problem eq. (2.19)

(f Δ2 g)d2 ξ = (g Δ2 f )d2 ξ
Ω Ω
Δg ∂f ∂ ∂g
+ −f ∂ + Δg + g Δf − Δf ds. (2.20)
∂Ω ∂n ∂n ∂n ∂n
As a further comment and just for the reader’s curiosity, one has an ana-
logue of representing (at least locally) two-dimensional harmonic functions
on Ω by analytical complex variable functions; one still has the following
representation (not unique!) for biharmonic functions (the Goursat repre-
sentation - 1898)
V(ξ1 , ξ2 ) = Re{z ϕ(z) + ρ(z)} (2.21)
where z = ξ1 + iξ2 and (ϕ(z), ρ(z)) a pair of complex variable analytic
functions on Ω.
If the surface S is on R3 and given in “nonparametric form” X 3 =
f (X 2 , X 2 ), we can introduce as parametric coordinates ξ 1 = X 1 , ξ 2 = X 2
and obtain as a result convenient for computations (exercises for the diligent
reader), the Gauss curvature given explicitly by
 2
R(hab ) = 1 + fX2 2
1 + fX 2 · det{fab (X 1 , X 2 )}. (2.22-a)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 31

Basics Polyakov’s quantum surface theory on the formalism of functional 31

And for the mean curvature, the following result:

 2 2
−3/2
H = 1 + fX 1 + fX 2
    
× 1 + fX 2
2
fX1 X1 − 2 fX1 fX2 fX1 X2 + 1 + fX 2
1
fX2 X2 (2.22-b)
It is worth recall that in the important case of a surface S embed-
ded in R3 (useful in Surface’s Statistical Physics), but with the general
parametrization:
X 1 = X 1 (ξ1 , ξ2 ); X 2 = X 2 (ξ1 , ξ2 ); X 3 = X 3 (ξ1 , ξ2 );
the expression for the area functional and the surface extrinsic functional
can be given explicitly by (exercise for our diligent readers!)
/4 0
A(S) = EG − F 2 (ξ) d2 ξ (2.23-a)
Ω

Fext (S) = H 2 (ξ) d2 ξ (2.23-b)

Ω
where the surface objects are given by
∂X μ ∂Xμ ∂X μ ∂Xμ
E= , ; F = , ;
∂ξ1 ∂ξ1 ∂ξ1 ∂ξ2
∂X μ ∂Xμ
G= , ;
∂ξ2 ∂ξ2
1
H= (ED − 2D F + DG);
(EG − F 2 )
1 1 1
1Xξ1 ξ1 Xξ21 ξ1 Xξ31 ξ1 1
1 1 1 1
D= √ 1X Xξ21 Xξ31 11
EG − F 1 1 1 ξ1
Xξ2 Xξ22 Xξ32 1
2

1 1 1
1Xξ1 ξ2 Xξ21 ξ2 Xξ31 ξ2 1
1 1 1
D = √ 1 X1 X2 X3 1 (2.23-c)
EG − F 1 1 1 ξ1 ξ1 ξ1 1
Xξ2 Xξ22 Xξ32 1
2

1 1 1
1Xξ2 ξ2 Xξ22 ξ2 Xξ32 ξ2 1
1 1 1
D = √ 1 X1 X2 X3 1
EG − F 1 1 1 ξ1 ξ1 ξ1 1
Xξ2 Xξ22 Xξ32 1
2

Finally we remark that one could easily generalize all the above written
results to the general case of the initial ambient space RD being new a gen-
eral Riemann Manifold (M, Gμν (X γ )). In this case, the “induced” surface
metric tensor hab (X(ξ)) = ∂a X μ (ξ)∂b Xν (ξ) will be replaced by the new
“induced” tensor
hab (X(ξ), [Gμν ]) = ∂a X μ (ξ)Gμν (X γ (ξ))∂b X ν (ξ)
on the exposed formulae. The full classical and quantum theory of there
surfaces σ-models can be found in [1].
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 32

32 Lecture Notes in Topics in Path Integrals

2.3 Path integral quantization on Polyakov’s theory of sur-

face (or how to quantize 2D massless scalar fields in the
presence of 2D quantum gravity)

After exposing some basic concepts of the classical surface theory in Section
1.2, we now pass to the vital problem of quantization on the formalism of
Feynman-Wiener path integrals.
Following R.P. Feynman in his theory of path integration – sum over
histories, a mathematical meaning for the continuous sum over (now), eu-
clidean quantum (random) surfaces for eq. (1.2) should be given by follow-
ing path integrals for the Nambu-Goto case and of A.M. Polyakov respec-
tively (see eqs. (2.3)–(2.8)).
 
1
GN G (C) = 1 dμ[X (ξ)] exp −
μ
A(Sc ) (2.24-a)
Xμ (ξ)1 =Cμ 2πα
ξ∈∂Ω

 
1 1
dμ[X (ξ), gab (ξ)] exp −
μ
Xμ (ξ)1
Gp (C) = A(Sc )
=C
1ξ∈∂Ω μ 2πα
gab (ξ)1 =0
 
ξ∈∂Ω

√
× exp −μ20 g(ξ) d2 ξ (2.24-b)
Ω

Here α and μ20 are constants to be identified with certain physical pa-
rameters of the theory. Note that α has dimension of inverse if mass square.
And [μ20 ] = [α ]−1 . It is important remark that the Feynman-Wiener path
measures dμ[X μ (ξ)] and dμ[X μ (ξ), gab (ξ)] must be defined in a such way
in order to be fully invariant under the action of the local diffeomorphism
group of both descriptions (see eq. (2.4) and eq. (2.9)).
We aim now to evaluate explicitly the A.M. Polyakov’s integral eq. (1.24-
b). As a first step one recall that we have imposed the “fluctuat-
ing” zero Dirichlet
1  boundary condition for the intrinsic metric field
gab (ξ) gab (ξ)1∂Ω = 0 . That results lead us to the effective action for the
surface dynamics weight functional
1 √   
A(Sc ) = g X μ − Δg Xμ (ξ) d2 ξ. (2.24-c)
2 Ω

Since at the classical level, the metric field decouples (one can choose
it in the form gab = eϕ δab ) and considering
4 the usual classical plus quan-
CL  q
tum decomposition Xμ (ξ) = Xμ (ξ)1 + (πα ) Xμ (ξ), where the quantum
correction Xμq (ξ) is such that Xμq (ξ)1∂Ω = 0.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 33

Basics Polyakov’s quantum surface theory on the formalism of functional 33

After these preliminaries considerations one is lead to evaluate the fol-

lowing covariant path integral (after disregarding classical field contribu-
tions to the path integral)
 
1 √
I[gab ] = dμ[X μ (ξ)] exp − ( g X μ (−Δg )Xμ ) (ξ) d2 ξ . (2.24-d)
2 Ω
It is worth recall that there is a classical term (a functional depending
on the loop boundary C) coming from the partial integration to arrive at
eq. (2.24) for the case gab = δab , which clearly do not satisfy the quantum-
fluctuating nature of these non-constant gab (ξ)
 μ 
Classical Term = ∂a XCL ∂a XμCL (ξ)d2 ξ. (2.25)
Ω

The Feynman-Wiener measure dμ[X μ (ξ)) on eq. (2.24-d) must be choose

in order to be invariant under the eq. (2.9). The only local candidate is
given by the “weighted” product Feynman measure ([1])
/4 0
dμ[X μ (ξ)] := g(ξ) dX μ (ξ)
4
(2.26)
ξ∈Ω

As a consequence the path integral eq. (2.24) turns out to be Gaussian

and metric dependent. It yields the immediate result:
I[gab ] = det −D/2 (−Δg ). (2.27)
We thus, have reduced the “explicitly” evoluation of the Polyakov’s
path integral to the functional integration of the above written functional
over all fluctuating metric fields
 √ 2
dμ[gab ] det −D/2 (−Δg )e−μ0
2
gd ξ
I= Ω . (2.28)

By using the theory of invariant integration on Riemann Manifolds of

ours ([1]), one can insures automatically the preservation of the diffeomor-
phism group by defining the above mentioned metric dμ[gab ] as the (func-
tional) element of volume
 of the following functional metric (so called De
Witt metric) with = − 21 :
;√ >       ?
||δgab ||2 = g δgab g aa g bb + c g ab g a b δga b (ξ) d2 ξ (2.29)
Ω

Let us choose the conformal gauge fixing to evaluate the above metric
path integral sucessfully ([1]), but already considering the metric positivity
gab (ξ) = eϕ(ξ) δab . (2.30)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 34

34 Lecture Notes in Topics in Path Integrals

As a result the functional infinitesimal displacements δgab (ξ) can be

written in the following general form on the functional manifold of the
metric [gab (ξ)] (on the formal infinite-dimensional functional tangent bun-
dle)
δgab (ξ) = δϕ(ξ)g ab (ξ) + (∇a Eb + ∇b Ea )(ξ). (2.31)
Let us re-write the functional De-Witt metric in the form
 
||δgab ||2 = (δgab (ξ)) γ (ab,a b ) [gcd (ξ)](δga b (ξ)) (2.32-a)
Ω

 
/√    
0
γ (ab,a b ) [gcd (ξ)] = g(g aa g bb + c g ab g a b ) (ξ) (2.32-b)

Since there is only three independent components of the metric tensor

gab (ξ) in two-dimensions (g12 = g21 ), the metric coeficients {γ ij } 1≤i≤3 are
1≤j≤3
effectively a 3 × 3 matrix on the functional manifold of the intrinsic 2D
metric fields
⎧ ⎛ ⎞⎫
⎨ δ g11 ⎬
||δgab ||2 = (δg11 , δg12 , δg22 )[γ] ⎝δ g12 ⎠ (ξ) d2 ξ
Ω ⎩ ⎭
δ g22

= (δg11 )(γ 11 )(δg11 ) + (δg12 )(γ 22 )(δg22 )

+(δg22 )(γ 33 )(δg22 ) − (δg11 )(γ 12 + γ 21 )(δg12 )


+(δg11 )(γ 13 + γ 31 )(δg22 ) + (δg12 )(γ 23 + γ 32 )(δg22 ) (ξ) × d2 ξ (2.33)

Here the explicitly expressions of the effective [γ]-matrix is related to

these of eq. (2.32-b) by the results
γ 11 = γ (11,11) = eϕ (e−ϕ δ 11 e−ϕ δ 11 + ce−ϕ δ 11 e−ϕ δ 11 ) = (1 + c)e−ϕ ;
γ 12 = γ (11,12) = 0; γ 21 = γ (12,11) = 0; γ 13 = γ (11,22) = c e−ϕ
γ 31 = γ (22,11) = c e−ϕ ; γ 23 = γ (12,22) ; γ 33 = γ (22,12) = (1 + c)e−ϕ
(2.34)
As a consequence
  (1 + c)2 c2
det[γ (ab,a b ) ] = det[γ iδ ] = 3ϕ
− 3ϕ = (1 + x)e−3ϕ . (2.35)
e e
So the De Witt metric on 2D is non-singular only if (C = − 12 ).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 35

Basics Polyakov’s quantum surface theory on the formalism of functional 35

Let us substitute the general funcitonal displacement eq. (1.31) on

eq. (2.3-c).
Firstly we get (one could choose from the begining on eq. (2.32-b).

||δgab ||2 = d2 ξ eϕ(ξ) {(δϕgba + ∇a Eb + ∇b E a )

Ω
· (δϕgab ) + ∇b Ea + ∇a E b ) + 4c(∇c E c + δϕ)2 } (2.36)
Let us analyze those terms involving the conformal factor.
 
By noting that ∇c gab = ∇c g ab = 0 and (∇a Eb )g aa g bb (∇b Ea ) =
  
∇a Eb ∇b Ea , added with ∇a E a = 0; ∇a (g bb Eb ) = g bb ∇a Eb = ∇a E b ;
one obtains the chaim of results below written
||δgab ||2 = d2 ξ eϕ(ξ) {(δϕgba + ∇a Eb + ∇b E a )
Ω
· (δϕgab ) + ∇b Ea + ∇a E b ) + 4c(∇c E c + δϕ)2 }(ξ)

= d2 ξ eϕ(ξ) {(∇a Eb + ∇b E a )(∇b Ea + ∇a E b )

Ω
− 2(∇d E d )2 + 2(1 + 2c)(∇d E d + δϕ)2 }(ξ)

= 2(1 + 2c) d2 ξ eϕ(ξ) (∇d E d + δϕ)2 (ξ)

Ω

−2 d2 ξ eϕ(ξ) {E a (∇d ∇d + [∇b , ∇a ])Eb } (2.37)

Ω
1
where we have used the covariant by parts integration rule (g 1∂Ω = 0)
√ √
d2 ξ g f (∇d h) = − d2 ξ g(∇d f )h. (2.38)
Ω Ω
As a consequence we have the full result
4
||δgab ||2 = (1 + 2c) d2 ξ g(ξ)(δV(ξ))2
Ω
1 1
4 1 1
+ 2 A E1 (ξ)
d ξ g(ξ) 11(E1 (ξ), E2 (ξ))[L] 1 (2.39)
E2 (ξ) 1
Ω

where
=0
+ ,- .
δV(ξ) = δϕ(ξ) + (∇c Ec )(ξ) = δϕ(ξ) (2.40)
and the Faddev-Popov operator acting on the two-dimensional vector field
generators of the local group of coordinate transformations is explicitly
given by
 
A a = − (∇d ∇d δ b + [∇b , ∇a ])Eb (ξ)
(LE) (2.41)
a
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 36

36 Lecture Notes in Topics in Path Integrals

As a consequence the functional volume element writes as of as

dμ[gab (ξ)] = dcov [ϕ(ξ)]dcov [E1 (ξ)]dcov [E2 (ξ)] det1/2 [L]. (2.42)

Here
ϕ(ξ)
/ ϕ(ξ) 0 ; ϕ(ξ) 
dcov (ϕ(ξ)) = e 2 δϕ(ξ) = δ 2e 2 = DF e 2
(ξ∈Ω) ξ∈Ω

d cov
[Ea (ξ)] = d Ea (ξ) = DF [Ea (ξ)] (2.43)
ξ∈Ω

Just for pedagogical purpose let us evaluate in details the relevant term for
the Fadeev-Popov operator in the above formulae.
We have that, for instance, the following sample calculation
√  
d2 ξ g(∇a Eb + ∇b Ea )g aa g bb (∇a Eb + ∇b Ea )
Ω
√  
= d2 ξ{ g(∇a Eb )g aa g bb (∇a Eb )}(ξ) + similar terms
 
gbb ∇a Eb =∇a (gbb Eb )=∇a E b
 + ,- .
√
= d2 ξ g (∇a Eb , ∇a E b )

+ ∇a E ∇b E + ∇b E ∇a E + ∇b E ∇ Ea (ξ)
b a a b a b


√
= d2 ξ g Eb (−∇a ∇a )E b − E b (∇a ∇b )E a

− E (∇b ∇a )E − E (∇b ∇ )Ea
a b a b

 2∇a ∇b =2∇b ∇a +2[∇,∇ ]− b

+ ,- .
2 √
=− d ξ g 2Eb (−∇a ∇a )E − E (∇a ∇b + ∇a ∇b )E a
b b

 (∇a E a =0)
√ + ,- . 
= −2 d2 ξ g Ea (∇ ∇c )δab + [∇b , ∇a ]− Eb
c
(2.44)

As an exercise to our diligent readers, the form of the above elliptic

operator in the conformal gauge are given by (action on two-component
objects)

Δ(g=eϕ δab ) = e−ϕ(ξ) ∂z ∂z

LA(g=eϕ δab ) = e−2ϕ(ξ) ∂z (eϕ(ξ) ∂z ) (2.45)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 37

Basics Polyakov’s quantum surface theory on the formalism of functional 37

where we have introduced the complex plane notation and the above dis-
played operators now acting on complex functions (isomorphic to two-
component objects = vector fields)
∂ ∂ ∂ √
= −i (i = −1)
∂z ∂ξ1 ∂ξ2
∂ ∂ ∂
= +i (2.46)
∂z ∂ξ1 ∂ξ2
In ref[3], we present the explicit evaluation of elliptic operators of the
form (for j being a positive integer) below considered
Lj = e−(j+1)ϕ(ξ) ∂z (ejϕ)ξ) ∂z ).
The result is
>
1 + 6j(j + 1)) 1
detF (Lj ) = exp − d2 ξ (∂a ϕ)2 (ξ)
12π Ω 2
1 ?
− lim d2 ξ eϕ(ξ) (2.47)
ε→0−1 2πε Ω

Let us point out the validity of the relationship below

−D/2 −D/4
detF (Δg=eϕ δab ) = detF (Lj=0 )
det + 12 A g=eϕ δ ) = detF
(L + 12
(Lj=1 ) (2.48)
ab

As our final result of this covariant Polyakov path integral, we get the
2D-induced quantum gravity Liouville model
; ϕ   
(26 − D) 1
I= D e F 2 (ξ)
exp − 2 2
(∂a ϕ) (ξ) d ξ
48π Ω 2
 
−μ2ren
× e e ϕ(ξ) 2
d ξ . (2.49)
Ω

(2 − D)
Here μ2r = μ20 + lim+ .
8πε
ε→0
Note that the quantum measure on the above displayed Liouville model
B
is not the usual flat Feynman measure dϕ(ξ) = DF [ϕ(ξ)] as initially
ξ∈Ω
wrongly supposed by A.M. Polyakov.
As a consequence one should
1 re-write
1 it in terms of the canonical Gold-
1 ϕ(ξ) 1/2 1
stone boson field ρ(ξ) = 1 e 1 which lead us to a kind of “non-
compact” σ-model with a “renormalized” mass term
 
(26 − D) (∂a ρ)2 2
I = DF [ρ(ξ)] exp − d ξ − μ 2
r ρ 2
(ξ) d2
ξ
12π Ω ρ2 Ω
(2.50)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 38

38 Lecture Notes in Topics in Path Integrals

and signaling the dynamical breaking of the conformal group ([1]).

Sometimes one can formally consider the variable change in the path-
measure of eq. (2.49)
Functional Jacobian
; ϕ  +  ϕ ,-  .
DF e 2 (ξ) = detF e 2 δab δ(ξ − ξ  ) DF [ϕ(ξ)]
; / ϕ 0 
= lim exp T r g e 2 δab e−EΔ g = eϕ × DF [ϕ(ξ)]
ε→0+
  
2 ϕ(ξ) ϕ(ξ) 1 1
= lim+ exp d ξe + R(ϕ)
ε→0 Ω 2 hπε 12π
× DF [ϕ(ξ)]
>  
1
= lim exp eϕ(ξ) ϕ(ξ)d2 ξ
ε→0+ 8πε Ω
 ?
1
× exp − (ϕ(−Δ)ϕ(ξ)))ξ)d2 ξ
24π Ω
× DF [ϕ(ξ)] (2.51)

which would lead to the same expression eq. (2.50) but with the “reduced”
anomaly conformal coefficient 26-D to 25-D, if one could disregard the in-
finite piece on eq. (2.51) and if the “Fujikawa-like” evaluation of the func-
tional jacobian could be done more invariant. At this point it appears
that the above 2D-Liouville-Polyakov path integral only makes sense at the
classical level which is formally equuivalent to evaluate all the observables
in the ϕ(ξ)-theory at the limit of D → −∞ ([1]). One can see this after
considering the re-scale
=
48π
ϕ→ ϕ
26 − D
ϕ
; √ 48π  D→−∞
DF [e 2 ] → DF e 26−D ϕ −→ DF [e0 ] ∼ classical.

However we should point out that all there questions on two-dimensional

quantum gravity as a well-defined problem still are not well-understood
since tis inception on 1981 ([1]).
Finally it is very important on applications to the Dual model theory
for strong interactions (off-shell Scattering Amplitudes) to have a covari-
ant regularized form for the formal Green function of the Beltrami-Laplace
operator in the conformal gauge gab (ξ) = eϕ(ξ) δab . It is well know from
Hadamard theory of parametrix singular solutions that one has the follow-
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 39

Basics Polyakov’s quantum surface theory on the formalism of functional 39

ing covariant behavior for the Beltrami-Laplace operator for small distances
 −1   1 1  
Δg=eϕ δab (ξ, ξ  )ε=(ξ−ξ )→0 ∼ − g eϕ(ξ) |ξ − ξ  |
2πε 2π
1 1 1 1
∼ − ϕ(ξ) − g|ξ − ξ  |.
2π ε 2π 2π
(2.52)
This lead us to consider the regularized from of the Green function used on
dual models based on the Polyakov’s path integral
⎧
/ 1 1 0 ⎪⎨− 1 g|ξ − ξ  | ξ = ξ
1 −1 −EΔg=eϕ δab 1  2π
ξ 1Δg=eϕ δab e 1ξ =
⎪− 1   1  + 1 ϕ(ξ)
⎩ ξ = ξ
g ε 4π
4π
(2.53)
Note that we have used the following formulae to write explicitly the func-
tional trace of the general two-dimensional (strongly) elliptic operators ([3])
∂2 ∂2 ∂ ∂
A = − g11 2 + g22 2 12×2 − (A1 ) − (A2 ) − A0 (2.54-a)
∂ξ1 ∂ξ2 ∂x1 ∂x2

1
lim tr(e−tA )t→0 d2 ξ (g11 g22 )− 2 (ξ)
+ 1
=
t→0+ 4πt

1 > 1
d2 ξ(g11 g22 )− 2 (− R(g −1 ))
1
+
4π 6

1 ∂
d2 ξ(g11 g22 )− 2 (g11 g22 )− 2 A1
1 1
−
2 ∂ξ 1

∂ 1 2 1 2 ?
− 12
+ ((g 11 g 22 ) A2 ) − (A1 /g 11 ) − (A2 /g 22 ) + A0 + O(t) (2.54-b)
∂ξ 2 4 4

2.4 Path-Integral quantization of the Nambu-Goto theory

of random surfaces

In order to given a path integral meaning for the symbolic Feynman con-
tinuum sum over surfaces histories eq. (2.24-a), we start re-writting it in
the following Polyakov’s form, but in the presence of a constraint ([1])
 
1 2 √
 
GN G (C) = dμ[xμ (ξ), gab (ξ)] exp − d ξ g g ab
∂a X μ
∂b X μ (ξ)
2πα Ω
 
× δcov
(F )
gab − ∂a X μ ∂b Xμ (2.55)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 40

40 Lecture Notes in Topics in Path Integrals

Note that our constraint is re-written as a covariant (diffeomorphism) in-

variant delta functional
 √
(F )
δcov (gab − ∂a X μ ∂b Xμ ) = Dcov [λab ]ei{ Ω
d2 ξ( gλab (∂ a X μ ∂ b X μ −gab ))(ξ)}

(2.56)
where the covariant Feynman-Wiener measure on eq. (1.56) is the element
of volume of the functional metric
√  
ds2 = ||δλab ||2 = ( g δ λab (g aa g bb )δ λa b )(ξ)d2 ξ (2.57)

After introducing again the complex-euclidean light-cone coordinates on

the domain Ω
∂+ = (∂ξ1 − i ∂ξ2 )/2 = ∂z
(2.58)
∂− = (∂ξ1 + i ∂ξ2 )/2 = ∂z
and imposing the surface conformal gauge gab (ξ) = ρ(ξ)δab , which means
that g11 = g22 ; g12 = g21 = 0, or equivalently using light-cone coordinates
again:
$
∂+ X μ ∂+ Xμ = ∂− X μ ∂− Xμ (⇔ ∂ξ1 X μ ∂ξ2 Xμ = g12 = g21 = 0)
√
g = ∂+ X μ ∂− X μ (g11 = ∂+ X μ ∂− Xμ = g22 = 0)
(2.59)
Now by taking into account the Jacobian relationship between the co-
variant functional measures and the associated pure Feynman-Wiener-Kac
path measures, one obtains the following results:
A carefull discussion presented in ref[1]-chapter 12, lead us to the fol-
lowing result on the surface light-cone gauge eq. (2.59)

GN G (C) = DF [ρ(ξ)]DF [X μ (ξ)]

$ % & '
2
(26 − D) 1 ∂a ρ (2 − D)
× exp − d2 ξ + lim d2 ξ ρ2
12π Ω 2 ρ ε→0 2πε Ω
 
1
× exp − d2 ξ(∂a X μ )2 (ξ)
2πα Ω
⎡ ⎤
=∂+ X μ ∂− Xμ
⎢ + ,- . ⎥
× δ (F ) ⎣(ρ)2 (ξ)δab − hab (X μ (ξ) ⎦ (2.60)

Note the usual non-covariant function “flat” constraint relating the sur-
face vector position to the auxiliary metric field gab (ξ) on the conformal
gauge.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 41

Basics Polyakov’s quantum surface theory on the formalism of functional 41

After realizing the immediate ρ(ξ) functional integral, one gets the final
result, after a renormalization of the Regge slope constant (∂+ = ∂z , ∂− =
∂z ) (*) 1
⎡ ⎤
4
GN (C) = ⎣ dX μ (z, z) h(X μ (z, z))⎦
4

(μ,ξ,ξ  )

× δ (F ) (∂+ X μ ∂+ Xμ )δ (F ) (∂− X μ ∂− X μ )∂X μ (ξ)eμ (ξ)

 
1
exp − μ
dzdz(∂+ X )(∂− Xμ )(z, z)
2παren Ω
 1  
(26 − D) μ
2 ∂+ (∂+ X ∂− Xμ ) ∂− (∂+ X μ ∂− Xμ )
exp − dz dz × (z, z)
48π Ω (∂+ X μ ∂− Xμ ) (∂+ X μ ∂− Xμ )
(2.61)
Now we realize that one most choose D = 26 or introduce intrinsic
Majorance fermions living on those Nambu-Goto random surfaces in order
to vanish the non-renormalizable Liouville surface term. The number of
such Neutral Majorana 2D fields must be such that 26 = 4 + m on D = 4.
Note that complex pairing such neutral Majorance fermion one gets eleven
complex fermion Dirac fields living on the random surface.
It is worth (all the reader attention that in this framework of Nambu-
Goto path integrals on surface light-cone gauge, the vertexs are given by
object:
>√ ?
V = h h[a1 ,a2 ] . . . ha2m−1 ,a2n (X μ (z.z)
⎧ ⎫
⎪ + ,- . ⎪
spin polarization tensors
⎨ ⎬
× ∂a1 X μ . . . ∂a2n X μ2μ · Eμ1 ...μ2m · eiKμ X
μ
(z, z)
⎪
⎩ ⎪
⎭

which are free from tachions, etc...

A complete study of the spectrum of such Nambu-Goto strings with
those tecnions-free “geometrical” vertexs is still missing on literature and
left to our diligent readers.
Similar analysis can be straightforwardly implemented for the extrinsic
surface path-integral weight functional eq. (2.17)–eq. (2.18).

1 (*) It is worth to remark that at a Feynman diagrammatic perturbative level, the

−δ(2) (0)
measure “tad-pole” factor in eq. (2.61), exp g(h(X μ ( z))) × dzdz , can
4 Ω
be assigned to exp{0} = 1 in the dimensional regularization scheme.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 42

42 Lecture Notes in Topics in Path Integrals

2.5 References

[1] Luiz C.L. Botelho - Methods of Bosonic and Fermionic Path Inte-
grals Representation - Continuous Random Geometry in Quantum
Field Theory - Nova Science - (2007)
[2] Luiz C.L. Botelho - Lecture Notes in Applied Differential Equations
of Mathematical Physics - World Scientific - 2008.
[3] J.T. Oden and J.N. Reddy - An Introduction to the Mathematical
Theory of finite elements - John Willey & Sons - 1976.

2.6 Appendix A: 2D Abelian Dirac Determinant On the

Formal Evaluation of the Euclidean Dirac Functional
Determinant on Two-dimensions

Let us start by this appendix by considering the euclidean Dirac partition

functional on a tw-dimensional space-time
 

1 − (ψ D(A)ψ)(ξ)d 2
Z[Aμ ] =: × Dψ Dψ e / ξ
Z[Aμ = 0]
D(A)) = det1/2 ((/
=: det(/ D(A))2 ) (A-1)
where the (non-self adjoint) euclidean Dirac operator is explicitly given by
(including the massive case!)
D(A)
/ = γμ (∂μ + g Aμ ) + im. (A-2)
The two-dimensional euclidean matrixes on the Dirac operator eq. (A-2)
satisfy the relationship below
{γμ , γν } = γμ γν + γν γμ = 2δμν
γμ γ5 = i Eμν γν , γ 5 = i γ0 γ1
E01 = −E10 = 1 (A-3)
Let us note the more suitable re-definition of the euclidean Dirac oper-
ator
=i γμ ∂μ +ig ∂μ Aμ
+ ,- .
/<
D(A) = iD(A)
/ = eigγ5D
/
(i γμ ∂μ )eigγ5D
/
(A-4)
where the back-ground abelian gauge field configurations are choosen on
the landau gauge ∂μ Aμ = 0
Aμ = Eμν ∂ν ϕ (A-5)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 43

Basics Polyakov’s quantum surface theory on the formalism of functional 43

Let us note that (with a undetermined infinite phase!)

/<
detF (D(A)) /<
detF (D(A))
= (A-6a)
/<
detF (D(0)) /<
detF (D(0))
1 12 1 12
1 det (D(A))
< 1 1 det (D(A))
< 1
1 F / 1 1 F / 1
1 1 =1 1 (A-6b)
<
1 detF (D(0))
/ 1 1 <
detF (D(0))
/ 1
Since we have the result:
/<
(D(A))2
= − ∂ 2 − 2g Aμ (ξ)∂μ − g 2 A2μ (ξ)
[γ μ ,γ ν ]
g +,-.
− σ μν ·Fμν (A(ξ)) − g(∂μ Aμ )(ξ) (A-7)
4
thus
C1> ?1 D
1 1 1
lim 1 /<
exp(−t(D(A)) 2
) 1ξ ∼ 1 2×2
t→0+ 4πt
 
1 g 1 1
+ g 2 A2μ + σ μν Fμν + g ∂μ Aμ − (2gAμ )2 − ∂μ (2gAμ ) , (A-8)
4π 4 4 2
one can straightforwardly apply the Romanov-Schwartz theorem ([1]) to
write the following (formal) differential equation for the functional deter-
minant under analysis
d
2 D(Aσ ))
g det(/
⎧ dσ ⎡ ⎤⎫
⎪
⎨
=[γ μ ,γ ν ]
+,-. ⎪
⎬
2 5 ⎢ 1 g μν ⎥
= lim+ 4 d ξ TrDirac (igγ φ) ⎣ + σ σ Fμν (A)⎦ (A-9)
ε→0 ⎪
⎩ 4πε 16π ⎪
⎭

Here
A(σ) = σ Eμν ∂ν ϕ (A-10)
Since TrDirac (γ5 ) = 0, one has the immediate result
det D(A)
/ g2  g2 
= e− 2π (∂φ)2 d2 ξ
= e− 2π A2μ (ξ)d2 ξ
× δ (F ) (∂μ Aμ ) (A-11)
det ∂
In the general non gauge fixed case, one has the result:
det(/
D(A)) g2 
= DF [ω(x)]e− 2π (Aμ +∂μ ω)2 d2 ξ
det( ∂)
2    ∂μ ∂ν
 
g
− 2π d2 ξ Aμ δμν − (−∂ 2 ) Aν (ξ)
=e (A-12)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 44

44 Lecture Notes in Topics in Path Integrals

Let us show in detail the above calculation

1
D(A(σ) ))2 )dσ
(g det(/
0
    1
1 g μ,ν
= lim+ 4 2 5
d ξ tr ig γ ϕ + σ Fμν (A) × dσ σ .
ε→0 4πε 16 0

So
 
det(/
D(A)) g2
g = d2 ξTrDirac (iγ 5 (−2i Eμν )γ5 φ × Fμν (A)
det( ∂) 16π
g2
= d2 ξ φ Eμν Fμν (A)
4π
g2 g2 g2
= d2 ξ φ ∂ 2 φ = − d2 ξ(∂φ)2 = − d2 ξ(Gμ )2
2π 2π 2π
(A-13)
It is important to remark that if one had used the following self-adjoint
Dirac euclidean operator
/A
D(A) = eeγ5 φ (iγμ ∂μ )egγ5 φ (A-14)
one would obtain a “Tachion” (imaginary mass) term for the dynamically
generated mass term for the abelian gauge field
 2 
/A
detFD(A) g
= exp + d2 ξ(Gμ )2 (A-15)
det ∂ 2π
At this point it is worth recall the Seeley asymptotic expansion
lim ξ| exp(−tA)|ξ
t→0+
% &
1 1 Vμ2 ∂μ Vμ
= 1 2×2 + V0 − − (A-16)
4πt 4π 4 2

for a formal second order operator of the form on R2

A = −∂ 2 − Vμ ∂μ − V0 (A-17)

2.7 Appendix B: On Atyah-Singer Index Theorem in the

Framework of Feynman Pseudo-Classical Path Integrals
- Author’s Original Remarks

One of the most celebrated (pure) mathematical theorem on modern geom-

etry and topology of compact orientable manifolds on RN is the so called
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 45

Basics Polyakov’s quantum surface theory on the formalism of functional 45

index theorem ([1]). By the other side, it is widely known that the tech-
niques to obtain such results are extremelly intricated and based on the
most difficult aspects of the theory of elliptic operators on closed compact
orientable N -dimensional manifolds (see [2]-chapter 11). Our aim in this
section is to somewhat simplify a very important index theorem on the
Laplacian operator acting on forms ([3]).
As in the last reference [3], one can argue that the index of the Laplacian
operator Δf acting on differential forms on a N -dimensional Riemannian
compact orientable manifold M is given by the object called super-trace
which by its turn can be wirtten as a Grasnmanian (supersymmetric) eu-
clidean path integral ([1]):
⎧ ⎛ ⎞⎫
⎪ ≡Dcov
F
[X μ (s)] ⎪
⎪
⎨ ⎜ + ,- . ⎪
⎟⎬
index(Δf ) =: lim ⎜ det
1
2 [G (X μ
(s))]dX μ
(s)⎟
→0 ⎪ ⎝ μν ⎠⎪
⎪
⎩ 0≤s≤1 ⎪
⎭
1≤n≤N

X (0) = X μ (1) = xμ
μ
⎛ ⎞
≡Dcov
F i
(ξ(α) (ξ)]
⎜+ ,- .⎟
⎜ 1 √ ⎟
× ⎜ det 2 [ G(X (s))]dξ(α) (s)dξ(α) (α)⎟
μ i ∗i
⎜ ⎟
⎝ 0≤s≤1 ⎠
α=1,2
1≤i≤2N
 1 
1 1 dX μ dX μ
exp − dsGμν (X α (s)) (s) (s)

0 2 ds ds
i d i
+ Gμν (X α (s)) ξ(α i
(s) ξ(α)
2 ds

1 α i k j 
+ Rijk (X (s))(ξ(1) (s)ξ(2) (s) )(ξ(1) (s)ξ(2) (s)) (B-1)
8
An important remark to be done now originally due to ourselves - [4])
is about the “bosonic” behavior of the Grassmanian composite products
i j
ξ(1) (s) ξ(2) = σ ij (s) as can seen easily verified from the computation be-
low ([a, b]− = ab − ba; [a, b]+ = ab + ba) −[AB, CD]− = A[B, C]+ D −
AC[B, D]+ + [A.C]+ DB − C[A, D]+ B
 i j p q   p q j   p q  j
ξ(1) ξ(2) , ξ(1) ξ(2) −
i
= ξ(1) ξ(1) ξ(2) , ξ(2) +
− ξ(1) i
ξ(2) , ξ(1) ξ
+ (2)
>  ?
p q j  p q j
= ξ(j) i
ξ(1) ξ(2) ξ(2) − ξ(1) ξ(2) i
ξ(1) ξ(2)
> ?
j p q p q j
+ ξ(1) i
ξ(2) ξ(1) ξ(2) − ξ(1)
i
ξ(1) ξ(2) ξ(2)
=0+0=0 (B-2)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 46

46 Lecture Notes in Topics in Path Integrals

On the basis of the above made remark eq. (B-2) one can re-write the
Grasmanian path integral eq. (B-1) in the terms of “composite bosonic
operators” ([4])
F
index(Δf ) = lim Dcov [X μ (s)] DF [ξαi (s)]
t→0 X μ (0)=X μ (1)=X
μ i (0)=−ξ i (1)
ξα α
Bosonic Variable Bosonic Variable
+ ,- . + ,- .
× F
Dcov [σik (s)] D F
[λik (s)]
σik (0)=σik (1) λik (0)=λik (1)
 
i
× exp ds λik (ξ(1)
i k
(s) ξ(2) (s) − σik (s))
h 0
 1 
1 1 j
× exp − Rijk (X α (s))σ(s)
ik
σ(s) (B-3)
4h 0 2
After integrating the Grasmanian variables {ξ(1) i i
(s), ξ(2) }1≤i≤N , one
gets the below written functional determinant as outcome (prove it with
the definition θ(0) = 0, here θ(s) denotes the heaviside step function-
distribution)
 
d
detcov Gij (X α (s)) δij + λik (s) = 1 (B-4)
ds
Now the σik functional integral is Gaussian and yielding the result
$  −1 '
1 1
1  1 
exp − ds ds (λik (s)) α
Rpqrs (X(s) ) δ(s − s )(λj (s))
2 0 0 2 (ik).(j)
(B-5)
At the classical limit  → 0, the closed periodic quantum trajectories
X μ (0) = X μ (s) = xμ , all reduces to theirs initial common point xμ . As a
consequence one gets that at  → 0.
X μ (s) → x−μ (B-6)

dN (x) · det 2 [Gμν (x−α ]

1
F
Dcov (X μ (s)) → (B-7)
M
 1 1
1
F
Dcov (λpq (s)] exp − ds (λ ik (s))
ds
 0 -.+,
0

A

 x
 −1  
R(A,B) (X (s) λ j (s ))
-.+,
B
1
→ det RAB (x )] =: Pf (RAB (x−μ ))
2 −μ
(B-8)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 47

Basics Polyakov’s quantum surface theory on the formalism of functional 47

By group together eqs. (B-6)–eq. (B-8), one gets the index of the
Laplacean operator acting on differential forms defined on a compact ori-
entable N -dimensional is universally proportional to the famous Chern in-
dex of the endowed riemannian structure of M. Namelly (see ref [3] for
topological details). Plus analogous induced term on the Manifold Bound-
ary if it is non empty (an open manifold).

1
index(Δf ) = ξChern (M) = e dN (x)det 2 [Gμν (xμ ]
M
Pfaffinian form of the curvature two-form
+ ,- . 
× Pf (RAB (x μ)
(B-9)

The universal constant is adjusted by defining as unity the Chern in-

duces or the Euler-Poincar characteristic) of spheres S N −1 with the usual
RN embedding metric (exercise to our differential geometric oriented read-
ers!). Now it is well known that χChern (M) are integers numbers.
In two dimensions, eq. (B-9) reduces to the famous Gauss theorem for
Riemann surfaces and its conections to the Euler topological surface genera

√
index(Δf ) = χChern (Σ) = ( gR(g))(xμ )d2 x = 2π(2 − 2g), (B-10)
Σ

where g is the number of “holes” of Σ.

2.8 References

[1] M. Nakahara - “Geometry, Topology and Physics” - Graduate Stu-

dent Series in Physics - Institute of Physics, 1990, UK.
[2] Luiz C.L. Botelho - Lecture Notes in Applied Differential Equations
of Mathematical Physics - World Scientific - 2008 (Singapure).
[3] P. Cartier and Ccile De Witt-Morette - Functional Integration -
Action and Symmetries - Cambridge Monogrphs on Mathematical
Physics - 2006, UK.
[4] Luiz C.L. Botelho - String Theory in Embeddings Manifolds - IJTP,
vol. 49, Issue 8, 1886-1893, DOI = 10.1007/s10773-010-0371-9,
(2010).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 48

48 Lecture Notes in Topics in Path Integrals

2.9 Appendix C: Path integral bosonisation for the

Thirring model in the presence of vortices

Recently the bosonisation of two-dimensional fermionic models in the

framework of path integrals has been shown to be a powerful non-
perturbative technique for exactly solving these models in the absence of
fermion zero modes [1,2].
It is the purpose of this Appendix to implement the path integral boson-
isation framework for the case of non-trivial fermion zero modes by boson-
isation of the Abelian Thirring model in the presence of vortex field con-
figurations.
Let us start our analysis by considering the Abelian Thirring model
interacting with a vortex field with topological charge n. Its Euclidean
Lagrangian is given by
1 2
L0 (ψ, ψ̄, A(n) 2 (n)
μ ) = ψ̄iγμ ∂)μψ + g (ψ̄γμ ψ) + eψ̄γμ Aμ Ψ (C-1)
2
where (g, e) are positive model coupling constants. The Euclidean Hermi-
tian γμ matrices we are using satisfy the relations
{γμ , γν } = 2δμν γμ γ5 = eεμν γν γ5 = iγ0 γ1 ε01 = −ε10 = 1 (C-2)
(n)
and the vortex field Aμ (x) with topological charge n (Chern number) and
length’ R is
1
A(n)
μ (x) = n 2 ελμ xλ . (C-3)
R + x2
The generating functional associated with the theory (1) is, thus, given
by

Z[η, η̄, A(n)

μ ]= D[ψ]D[ψ̄] exp − d2 x(L0 (ψ, ψ̄, A(n)
μ ) + η̄ψ + ψ̄η)(x) .

(C-4)
In order to implement the path integral bosonisation gauge invariant
technique [1,2], we rewrite the fermion interaction term in the Hubbard-
Stratonovitch form:
1
Z[η, η̄, A(n)
μ ]= D[ψ]D[ψ̄] exp − d2 xBμ2 (x)
2

× exp − d2 x[ψ̄iγμ (∂μ + Bμ − ieA(n)

μ )ψ](x)

× exp − d2 x(ψ̄η + η̄ψ)(x) . (C-5)

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 49

Basics Polyakov’s quantum surface theory on the formalism of functional 49

Let us now proceed as in [1,2] by making the (partial) decoupling field

change in (C-5)

ψ(x) = [exp(igγs u)x](x) ψ̄(x) = [χ exp(igγs u)](x)

Bμ (x) = (gεμν ∂ν u)(x). (C-6)

As has been shown by Fujikawa [3], the transformations of (C-6) are not
free of cost in the fermionic sector since the functional measure D[ψ]D[ψ̄]
is defined in terms of the normalised eigenvectors of the Dirac operator
(n)
/ μ , Aμ in the presence of the external (auxiliary) Abelian field Bμ and
D(B
(n)
of the vortex topological field Aμ .
At this point we note that after the chiral change takes place, the new
quantum fermion vacuum is defined by the fermionic theory in the pres-
ence of the topological vortex; i.e. D[χ]D[χ̃] is now defined in terms of the
(n)
eigenvectors of the Dirac operator D(A/ μ ) which in turn has precisely n
fermionic zero modes with definite chiriality [4]. Their explicit expressions
are
1/2 l−1 (n)
1 x1 + ix2 h− (x, Aμ
ψ(0),l (x) = (C-7a)
2πR R 0

γ5 ψ(0),l (x) = ψ(0),l (x) (C-7b)

h± (x, A) = exp ie d2 zΔm=0 (x − z)(∂μ A(n)

μ ± iεμν ∂μ Aν )(z)
(n)
(C-7c)

with Δm=0 (z) the (infrared regularised) massless Klein-Gordon propagator.

The associate Jacobians are given by [1,2]
(n)
/ μ , Aμ )
det D(B
D[ψ]D[ψ̄] = D[χ]D[χ̄] (n)
. (C-8a)
det D(B
/ μ = 0, Aμ )

D[Bμ ] = D[u] × det[−∂ 2 ]. (C-8b)

So, we face the problem of the evaluation of the ratio of the two Dirac
determinants with zero modes.
By following the procedure of [1,2], we first introduce a one-parameter
continuous family of Dirac operators itnerpolating the pure vortex Dirac op-
(n) (n)
/ μ , Bμ ) = iγμ ((∂μ + gBμ + ieAμ defined by the expression
erator and D(A

/ (σ) (Bμ , A(n)

D μ ) = exp(igγ5 σu)D(Aμ ) exp(igγ5 σu) (0 ≤ σ ≤ 1).
/ (n) (C-9)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 50

50 Lecture Notes in Topics in Path Integrals

By using a proper-time prescription to define the functional determinant

(n)
/ (σ) (Bμ , Aμ ) (after making the analytic exrension g = −iḡ [2]). We
of D
2
have the result log det D/ (σ) = 12 log det(D
/ (σ) )|g=ig
1/ε
dζ 2
= − lim / (σ) )(1 − P(σ) )]
T r[exp(−ζ D (C-10)
ε→0+ ε ζ
with

n
P(σ) |ζ = ζ|β(0),l
σ
β(0),l
σ

l=1
denoting the projection over the zero modes of the Dirac operator
(n)
/ (σ) (Bμ , Aμ ). Let us remark that by the Atyah-Singer theorem, this op-
D
σ
erator still has n zero modes β(0),l (x) which are related to those of (C-7a)
by an analytically continued chiral rotation (C-6):
σ
β(0),l (x) = exp(−ḡγ5 σu)ψ(0),l (x). (C-11)
The functional determinant (C-10) satisfies the following differential
equation:
d
/ (σ) (Bμ , A(n)
log det D μ )
dσ
11/ε
/ (σ) )(1 − P(σ) )]1
2
= −2 lim T r[ḡγ5 u exp(−ζ D ε
ε→0+
1/ε  
dζ 2 d (σ)
+ lim+ T r exp(−ζ D/ (σ) ) P . (C-12)
ε→0ε ζ dσ
The second term in the right-hand side of (C-12) is easily evaluated by
noting that
d (σ) m 
n
P |ζ = ζ| − ḡγ5 uβ(0),l
σ
+ ζ|β(0),l
σ
 − ḡγ5 uβ(0),l
σ
(C-13)
dσ
l=1 l=1
which yields the zero-mode contribution for the determinant (12):
1/ε
dζ 2 d
lim T r exp(−ζ D/ (σ) ) P(σ)
ε→0+ ε ζ dσ
% &

n
= lim 4log(ε)ḡ β(0),l
σ σ
uβ(0),l  . (C-14)
ε→0+
l=1
2
Since D / (σ) is a self-adjoint invertible operator in the manifold orthog-
onal to its zero modes, wed can use the usual Seeley-De Witt technique to
evaluate the first term in (C-12), which produces the usual result
11/ε
/ (σ) )(1 − P(σ) ))1ε
2
T r(ḡγ5 u exp(−ζ D
1 / εμν 0
= σ T r (−iḡ)u(−iḡ)∂ 2 u + F |μν (A(n)
μ ) . (C-15)
π 2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 51

Basics Polyakov’s quantum surface theory on the formalism of functional 51

By combining (C-15) with (C-14) and coming back to the real coupling
constant g, we obtain the final expression for the Jacobian (8a) after inte-
grating the differential equation (C-12)
log det D(B μ ) − log det D(B
/ μ , A(n) / μ = 0, A(n)
μ )
g2 1 g
= d2 x (∂u)2 + d2 xεμν Fμν (A(n)
μ )u
π 2 π

n
− igC(ε) +
d2 xψ(0),l (x)u(x)ψ(0),l (x) (C-16)
l=1

where we have used the unitarity of the matrix exp{iḡγ5 u} to evaluate (C-
16) and C(ε) is the usual infrared divergence contribution for the zero-mode
term, which can be made finite by a multiplicative renormalisation of the
Thirring coupling constant g.
The generating functional thus takes the simple form
Z[η, η̄, A(n)
μ ]
+ ,-. > 1 ;/ g2 0 
= D[u] exp − d2 x 1 − (∂u)2 (x)
2 π
g
− d2 x(εμν Fμν (A(n)
μ )u)(x)
π
/ 0?
× exp − ig d2 xψ(0),l +
(x)u(x)ψ(0),l (x) Z(0) [η, η̄, A(n)
μ , u]

(C-17)
(n)
with Z(0) [η, η̄, Aμ , u] being the generating functional for the fermions in
the pure vortex field configuration

Z(0) [η, η̄, A(n)

μ , u] = D[χ]D[χ̄]

× exp − d2 x(χ̄D(A
/ (n)
μ )χ + χ̄ exp igγ5 uη + η̄ exp igγ5 uχ)(x) .

(C-18)
Let us exemplify our approach by calculating the two-point fermion
correlation function for an external vortex field with topological charge 1.
(n)
By functional differentiation of Z(0) [η, η̄, Aμ , u] twice, we get
(χ̄ exp igγ5 u)α (x)(exp igγ5 uχ)β (y)
= [exp(igγ5 u(x))S (1) (x − y) exp(igγ5 u(y))]αβ × det(D(A
/ (n)
μ ))
(C-19)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 52

52 Lecture Notes in Topics in Path Integrals

where S (1) (x − y) is the Euclidean Green function of the Dirac operator

(n)
/ μ ) for n = 1 which is given explicitly by [4]
D(A
⎡ / 01/2 ⎤
1+y 2 x1 −ix2 −(y1 −iy2 ) (y1 −iy2 )
0 +
1 ⎢ (1+x2 )1/2 (1+y 2 )1/2 ⎥
1+x2 (x−y)2
⎢ ⎥
S (1) (x−y) = − ⎢ ⎥.
2π ⎣ / 01/2 ⎦
2 x +ix −(y +iy )
− 1+x1+y 2
1 2
(x−y)2
1 2
+ (1+x2 )x1/2
1 +ix2
(1+y )2 1/2 0
(C-20)
By evaluation of the u average of (C-19) we finally have the complete
expression for ψ̄(x)ψ|y):
; /1 1
ψ̄(x)ψ(y) = exp d2 zd2 z̄Jμ (z; [x, y])
2 1 − g 2 /π
0 
× Δm=0 (z − z̄)Jμ (z̄; [x, y]) S (1) (x, y) (C-21)

where
/g
μ (z)) − igψ(0),0 (z)ψ(0),0 (z)
+
Jμ (z; [x, y]) = εμν Fμν (A(1)
π 0
(x) (y)
+ (igγ5 δ(z − x) + igγ5 δ(z − y)) (C-22)

with
1 1
ψ(0),0 (z) = √ √ ,0
2π 1 + z2
(1)
begin the only zero mode of D(A
/ μ ).

2.10 References

[1] Botelho L.C.L. (1985) Phys. Rev. D 31 1503; 1986 Phys. Rev. D
33 1195, 34 3250.
[2] Botelho L.C.L. (1989) Phys. Rev. D 39 3051.
[3] Fujikawa K. (1981) Phys. Rev. D 21 2848.
[4] Rothe K.D. and Schroer B. (1979) Phys. Rev. D 20 3203.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 53

Basics Polyakov’s quantum surface theory on the formalism of functional 53

2.11 Appendix D:

Several attempts have been made to find a complete solution of two-

dimensional quantum chromodynamics with massless fermions ([1], [2], [3]),
shortly denoted by (QCD)2 . All these approaches were based on integrat-
ing out the fermion fields of the model and considering an effective gluonic
theory where some interesting phenomena can be easily seen ([2], [3]).
In this appendix we intend to make some clarifying remarks on this
effective action for massless (QCD)2 .
We start our analysis by considering the generating functional for the
model in a Euclidean space-time R2 with local gauge group SU (2) (the
generalization for the case SU (N ) is straightforward)
 
− 4g12 d2 xT r (c) (Fμν
2
)(x) d 2 x Jμ ·Gμ
Z[Jμ , η, η̄] = D[Gμ ]e e

Dψ · Dψ̄e−( d2 x(ψ̄ D(G
/ μ )ψ+η̄ψ+ψ̄η)(x))
(D-1)

/ μ ) = iγμ (∂μ − iGμ ) denotes the self-adjoint Dirac operator in

where D(G
the presence of the external gauge field Gμ and the tensor field strength is
given by Fμν = ∂μ Gν − ∂νGμ + i[Gμ , Gν ].
The hermitean γ-matrices we are using satisfy the (Euclidean) relations.

{γμ , γν } = 2δμν ; γμ γ5 = iεμν γν ; γ5 = iγ0 γ1

ε01 = −ε10 (μ, ν = 0, 1) (D-2)

The functional measures in (D-1) are normalized to unity and the

fermion measure DψDψ̄ is defined in terms of the eigenvalues of the self-
adjoInt Dirac operator D(G/ μ ) which insures automatically its gauge invari-
ance.
Our plan to study (D-1) is to implement a convenient change of vari-
ables in the fermionic sector of (D-1) in order to get an effective generating
functional were the fermion fields are decoupled from the gauge field Gμ
([1], [2]). For this analysis, we are going to use a general decomposition of
the gauge field Gμ due to Roskies ([1]) and this will be explained in the
following.
Roskies in Ref. [1], has shown that for any gauge field configuration
Gμ (x), there is a unique unitary matrix Ω(x) taking values in SU (2) and
a hermitean matrix V (x) = e−γ5 ψ(x)J taking value over the axial gauge
group SU (2) (whose Lie algebra is generated by the hermitean generators
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 54

54 Lecture Notes in Topics in Path Integrals

γ5 J, with J denoting the usual SU (2)-generators) such that

iγμGμ (x)J = γμ ∂μ (V −1 Ω−1 )(x)(ΩV )(x)
= −γμ (V −1 Ω−1 )(x)∂μ (ΩV )(x). (D-3)
The proof of the validity of the decomposition (D-3) can be accom-
plished by considering the j = −iγ5 complexification of space-time, which
is denoted by C (C = {z = (x0 , x1 )|z = x0 + jx1 ; z̄ = x0 − jx1 and
(x0 , x1 ) ⊂ R2 }). In this C space-time, we can re-write the partial differen-
tial equation (D-3) into a single ordinary differential equation ([1])
Ḡ · τ = −2i(∂z̄ W )W −1 (D-4)
where Ḡ is the j-complexification of Gμ (Ḡ = G0 +jG1 ) and W is an element
of SL(2, C) (the associated j-complexification of SU (2)).
The equation (D-4) is just the equation for a holomorphic. Principal
bundle over C, and, as is well known from differential topology, all such
bundles are trivial, which means that a unique global solution W for (D-4)
exists.
In order to determine explicitly this solution W , we note that (D-4) can
be easily integrated, leading to the result
> z ?
W ((x0 , x1 )) = P e−2i −∞ (C̄ τ )dz̄ (D-5)

where dz̄ = dx0 − jdx1 and the SU (2) path-ordered integral in (D-5) is
taken over the (infinite) straight segment joining the (−∞) point to the
z = (x0 , x1 ) point.
By introducing the axial gauge field
∗Gμ · τ = (εμα Gα ) · τ (D-6)
we can re-write Eq. (D-5) in the more transparent form
  (x0 ,x1 )  (x0 ,x1 ) 
(−2i (−∞) (Gμ ·τ dxμ −2γ5 (−∞) (∗Gμ ·τ )dxμ )
W ((x0 , x1 )) = P e (D-7)

where, again, the SU (2) path ordered integral in (D-7) is taken over the
straight segment joining the (−∞) point to the (x0 , x1 ) point.
Continuing our study, we can see that the Dirac operator D(G / μ ) can
be re-written in the suitable form ([3], [5])
/ μ ) = (ΩV −1 )(x)(iγμ ∂μ )(V −1 Ω−1 )(x).
D(G (D-8)
Here, the matrices Ω(x) and V (x) are respectively the unitary and her-
mitean factors of the SL(2, C) Wu-Yang factor (D-7).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 55

Basics Polyakov’s quantum surface theory on the formalism of functional 55

In order to decouple the fermion fields from the gauge field, we follow
Ref. ([1]) by making the variable change.
ψ(x) = (Ω · V )(x)χ(x)
(D-9)
ψ̄(x) = χ̄(x)(V · Ω−1 )(x)
which yields the fermionic generating functional

Z̃(η, η̄] = D[x]D[x̄]J[Gμ ]e−( d2 x(χ̄(iγμ ∂μ )χ

+ η̄(Ω · V )χ + χ̄V Ω−1 η)(x) (D-10)

where the quantum aspect of the variable change (D-9) is taken into account
by considering the associated jabobian J[Gμ ] ([1], [2]).
Now, it is important to note that the jacobian J[Gμ ] is given by the
ration
/ μ ))
det(D(G
J[Gμ ] = . (D-11)
det(iγμ ∂μ )
In order to evaluate the functional fermionic determinant in (D-11), we
(σ)
introduce a family of gauge fields Gμ (0 ≤ σ ≤ 1) interpolating continu-
(σ=0)
ously the zero field configuration Gμ ≡ 0 to the considered configuration
(σ=1)
Gμ = Gμ in (D-11) and defined by the relation (see Eq. (D-3))
−σγ5 φτ −1
μ = −γμ (e
iγμ G(σ) Ω )∂μ (Ωeσγ5 φτ ). (D-12)
We note that we are assuming implicitly that we are computing the
Jacobian J[Gμ ] in the trivial topological sector of the manifold of the gauge
fields configurations, since Gμ is in the same homotopical class of the zero
field configuration. As a consequence of this fact, we do not taken into
account the zero-modes of the operator D(G / μ ) in what follows.
(σ)
Now, it seems important to remark that to evaluate det(D(G / μ )) we
can consider solely the “reduced” operator
/̃ (σ)
D(Gμ )= e
σγ5 φJ
(iγμ ∂μ )eσγ5 φJ (D-13)
(σ) (σ)
since /̃ μ )
D(G is related to D(G
/ μ )by a similarity transformation defined
by the unitary matrix Ω(x) (see Eq. (D-8)). This result is directly related
to the gauge invariance of the Jacobian J[Gμ ], i.e. only the axial SU (2)
matrix V (x) contributes to J[Gμ ].
(σ)
In order to evaluate det(D(G
/ μ )) we proceed as in ([3], [5]). Using the
proper-time method to define the functional determinant and making use
of the relation
d
/ (σ)
D(G 5
/ (σ)
μ ) = γ φτ D(G / (σ)
μ ) + D(G
5
μ )γ φτ (D-14)
dσ
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 56

56 Lecture Notes in Topics in Path Integrals

we get the following ordinary differential equation

d
/ (σ)
(log det(D(G 2
μ )) ) = lim+ 4 d2 xT r(e,D)
dσ ε→0

(γ 5 (φτ )(x)x|e−ε( D(G

(σ) 2
/ μ ) )
|x (D-15)
where T r(C,D) denotes the trace over the Dirac and the color indices.
The asymptotic expansion of the operator
x|e−ε(D(G |x = |e−ε{(−(∂μ −iGμ
(σ)
))2 (σ) 2
) + 2i εμν γ5 Fμν (−iG(σ)
/ μ μ )} |x
is tabulated ([6])

lim x|e−ε(D(G
(σ) 2
/ μ ) )
|x
ε→0+
1 iGμν γ5
= lim (1 − ε( Fμν (−iG(σ)
μ )))(x). (D-16)
ε→0+ 4πε 2
Substituting (D-16) into (D-15), we get the result
1 
J[Gμ ] = e− 2π εμν { dσ( d2 xT r (c) (φτ Fμν (−iG(σ)
i
0 μ ))(x))} (D-17)
We remark that the result (D-17) coincides with the result obtained by
Roskies ([1]), and so the σ integration can be done explicitly producing the
expression
εμν
   
d2 x φFμν (Gμ ) 1
− 1
(x)
J[Gμ ] = e π |φ|tanh|φ| sinh2 |φ| (D-18)
We also note that by considering the vector and axial components of the
(σ) (σ) (σ) (σ)
gauge field Gμ (Gμ = iVμ +εμν Aν ), we re-obtain the result established
in Refs. ([3]) and ([5]).
Finally, the effective generating functional for the model where the
fermion fields are decoupled from the gauge fields can be written

− 4g12 T r (c) (Fμν
2
)(x)d2 x
Z[Jμ , β, β] = D[Gμ ]e DHaar [Ω]J[GΩ
μ]
 
(J μ G Ω
DχDχ̄ e−{
2
μ )(x)d x d2 x(χ̄(iγμ ∂μ )χ
e (

+ η̄(ΩV )χ + χ̄(V Ω−1 )η)(x)} (D-19)

We remark that we have to fix a gauge in (D-19). This gauge is not
necessarily the Roskic’s gauge ([1], [2]), which choice, will imply to consider
Ω(x) = π in (D-19).
From (D-18) we see that the analysis of the fermionic correlation func-
tions are reduced to the computation of the interaction among the SL(2, C)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 57

Basics Polyakov’s quantum surface theory on the formalism of functional 57

Wu-Yang factors (D-7) with the quantum average defined by the local ef-
fective gluonic action (in the Roskies gauge Ω ≡ 1)2
1
S EF F [Gμ ] = − d2 xT r(c) (Fμν
2
)(x) + gJ[Gμ ]. (D-20)
4g 2
For instance, the two point fermionic correlation function is given by
1 xμ − yμ
ψ(x)ψ̄(y) = γμ W (x)W −1 (y)EF F (D-21)
2π |x − y|2
where  EF F is the quantum average defined by the action (D-20);
1 xμ −yμ
W ((z)) = W ((z0 , z1 )) is given by Eq. (D-7) and 2π γμ |x−y| 2 is the free

fermion propagator.

2.12 References

[1] R. Roskies, in Festschrift for Feza Gursey’s 60th Birthday, (1982)

(unpublished).
[2] R.E. Gamboa Saravi, F.A. Schaposnik and J.E. Solomin – Phys.
Rev. D – vol. 30, no 6 (1984).
[3] Luiz C.L. Botelho and M.A. Rego Monteiro, Phys. Rev. D - vol.
30, no 10 (1984); 2242 Luiz C.L. Botelho, PHD thesis, Centro
Brasileiro de Pesquisas Fsicas (1984).
[4] H. Gravert, Math. Annalen 135, 266 (1958).
[5] Orlando Alvares - Nucl. Phys. B238, 61 (1984).
[6] V.N. Romanov and A.S. Schwartz - Teor. Mat. Fiz. 41 (1975),
190.
[7] Robert J. Kares and Myron Bander-Phys. Lett. 96B no 3,4 (1980),
326.

2 It appears to be an open problem in the subject the full evaluation of the Non-Abelian

Dirac Determinant defined on a Riemman Surface (the Hodge theorem for non abelian
connections).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 58

58 Lecture Notes in Topics in Path Integrals

2.13 Appendix E: Path-Integral Bosonization for the

Abelian Thirring Model on a Riemann Surface – The
QCD(SU (N )) String

We study the path integral chiral bosonization of the Abelian-Thirring

model defined on a boundaryless Riemann surface as a two-dimensional
space-time. This short of 2D fermionic models on random surfaces are
expected to be the QCD(SU (N )) string – see Chapter 5.
Analysis of quantum field models defined on Riemann surface as two-
dimensional space-time is a fundamental issue for strings field theory in
Polyakov’s approach [1,2].
It is the purpose of the letter to solve exactly the Abelian-Thirring model
defined on a Riemann surface in the Framework of chiral path integrals [3].
We start our analysis by considering the Abelian-Thirring model asso-
ciated to a complex spin field associated to a complex structure (θi , φi ) of
a genus g Riemann surface D(g)
c2
L(ψ, ψ̄))(θi , φi ) = ψ̄iγ μ Dμ ψ + (ψ̄γ μ ψ)2 . (E-1)
2
Here thd Dirac operator is given by
iγ μ Dμ = iγêμa ∂μ (E-2)
μ
where êa are fixed background two-being satisfying the topological genus
constraint
4
ĝ R(ĝ) = 2π(2 − 2g). (E-3)
D(g)
R(ĝ) is the scalar of curvature associated to ĝμν and ωμab (ĝ) is the spin
connection defined by the relation ∇μ ∂ν = 0.
The γ μ = êμa γa Eucidean (curved) Dirac matrices are defined by the
relationship below (ξ ∈⎧D(g) )
⎨{γμ , γν ) + (ξ) = 2ĝμν (ξ),
/ μν 0 (E-4)
⎩γ μ (ξ)γ5 = i ε √ γν (ξ),
ĝ
where γa are the usual flat-space Dirac matrices.
In the framework of path integrals, the generating functional of the
Green’s function of the (mathematical) quantum field theory associated
with the Lagrangian eq. (E-1) is defined by the following covariant func-
tional integration [2]  
4
Z[ρ, ρ̄] = d [ψ]d [ψ̄] × exp −
c c
d ξ( ĝ L(ψ, ψ̄))(ξ)
2
D(g)
 
4
× exp − 2
d ξ( ĝ (ρ̄ψ + ψ̄ρ))(ξ) . (E-5)
D(g)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 59

Basics Polyakov’s quantum surface theory on the formalism of functional 59

It is worth pointing out that the classical action in eq. (E-5) s invariant
under the local diffeomorphism group and the global Abelian and Abelian-
chiral groups acting on the spin field restrict to any local region R of D(g) .
These symmetries have the associated Noether conserved currents
∇μ (ψ̄γ 5 γ μ ψ) = 0; ∇μ (ψ̄γ μ ψ) = 0. (E-6)
In order to implement the path-integral gauge and local diffeomor-
phism invariant bosonization, we rewrite the fermion interaction term in
the Hubbard-Stratonovitch form by using an ausiliary vector field Aμ (ξ)

Z[ρ, ρ̄] = dc [ψ]dc [ψ̄]dc [Aμ ]

; 4 ; 1 
× exp − d2 ξ( ĝ ψ̄iγ μ (dμ + cAμ )ψ + Aμ Aμ ](ξ)
D(g) 2
 
4
× exp − d2 ξ ĝ (ρ̄ψ + ψ̄ρ)(ξ) . (E-7)
D(g)

Let us now proceed as in [4–6] by making the local field change in

eq. (E-7)
εμν
Aμ (ξ) = − √ ∂ν η (ξ) + AH
μ (ξ), (E-8)
ĝ

ψ(ξ) = (exp[iγ5 η(ξ)]) · χ(ξ), (E-9)

ψ̄(ξ) = χ̄(ξ) · exp[iγ5 η(ξ)], (E-10)

where ∇μ (Aμ − AH μ ) ≡ 0 and Aμ (ξ) is the Hodge topological vector field
H

which is explicitly given in terms of canonical Abelian differentials ωi nd

their complex conjugates ω̄i [7]

g
AH
μ (ξ) = 2π (pi · αiμ (ξ) + ri βμi (ξ)), (E-11)
l=1

αiμ (ξ) = −Ω̄ik (Ω − Ω̄)−1

kj (ξ) + c.c., (E-12)

βμi (ξ) = (Ω − Ω̄−1 i

ij ωμ (ξ) + c.c. (E-13)
The period matrix Ω is defined by

ai = δij ; aj = Ωij , (E-14)

bi

where ai and bi are (canonical) homology cycles on D(g) .

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 60

60 Lecture Notes in Topics in Path Integrals

As it has been shown by Fujikawa [5], the transformation of

eqs. (E-9)–(E-10) are not free of cost, since the functional measures
dc [ψ]dc [ψ̄] are defined in terms of the normalized eigenvectors of the co-
variant and U (1) gauge invariant Dirac operator eq. (E-2) in the presence
of the auxiliary vector field Aμ .
The associated Jacobian of eqs. (E-9), (E-10) is given by [6]
det[iγ μ (Dμ + cAμ )]
dc [ψ]dc [ψ] = dc [χ]dc [χ̄] × . (E-15)
det[iγ μ (Dμ + cAH
μ )]

At this point we note that after the chiral change takes place the new
quantum fermionic vacuum is defined by the fermionic field χ(ξ) (with the
same spin structure of ψ(ξ)) in the presence solely of the Hodge topological
field AH
μ (eq. (E-1)).
The Jacobian associated to eq. (E-8) is [7]
 g 1/2 αi , αi  αi , β i 
d [Aμ ] = d [η] (2π)2g
c c
dpi dri × det μ μ μ μ
, (E-16)
βμi , αiμ  βμi , βμi 
l=1

where the covariant scalar product in the space of vector fields in D(g) is
defined by
4
Σμ , θμ  = d2 ξ( ĝ ĝ αβ Σα θβ )(ξ). (E-17)
D(g)

Let us remark that with this definition we have

ωμi , ωμj  = 2ImΩij . (E-18)

So, we face the problem of the evaluation of the ration of two Dirac
determinants related themselves by a chiral rotation
det[exp[icγ5 η]iγ μ (Dμ + cAμ ) exp[icγ5 η]]
J[Aμ ] = . (E-19)
det[iγ μ (Dμ + cAHμ )]

Byfollowing the procedure of ref.[6] we, at first, introduce a One-

parameter family of Dirac operators interpolating the Dirac operator
iγ μ (Dμ + cAH μ ) = D(Aμ ) and the chirally rotated exp[icγ5 η)D(Aμ ) ·
H H

exp[icγ5 η]

μ ) · exp[icγ5 ζη],
D(ζ) (Aμ ) = exp[icγ5 ζη]D(AH (0 ≤ ζ ≤ 1). (E-20)

By using a proper-time prescription to define the functional determinant

of D(ζ) (after making the analytic extension c = −ic̄), we have the following
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 61

Basics Polyakov’s quantum surface theory on the formalism of functional 61

differential equation for log det D(ζ)

d 2
log det D(ζ) = −2 lim T r[c̄γ5 η exp[−σD(ζ) ] × (1 − P(ζ) )]1/ε
ε
dζ ε→0 +

1/ε
dσ 2 d
+ lim+ T r exp[−σD(ζ) ] P(ζ)
ε→0 ε σ dζ
(ζ) (ζ)
= I(1) [Aμ ] + I(0) [Aμ ], (E-21)
@ (0),(ζ) (0),(ζ)
where P(ζ) =  , φn φn denotes the projection over the zero
n
(0),(ζ)
modes φn of the Dirac interpolating operator D(ζ) . These zero modes
are related by an analytically continued chiral rotation to those of D(AH
μ )

φ(0),(ζ)
n = exp[−c̄γ5 ζη] · φ̄(0)
n (E-22)
and
μ ) · φ̄n = 0.
D(AH (0)
(E-23)
(σ)2
Since D (Aμ ) is a self-adjoint invertible operator in the manifold or-
thogonal to the subspace generated by the zero modes, we can use the
Seeley-De Witt technique to evaluate the first term in eq. (E-21) which
yields
(ζ) 2
I(1) [Aμ ] = lim+ T r[c̄γ5 η exp[−σD(ζ) ](1 − P(ζ) )]|1/ε
ε
ε→0
 
2 1 εμν μν H
= − ζT r −ic̄ η √ ∂α (ĝ ∂β )η +
αβ
F (A ) . (E-24)
π ĝ 2
The second term on the left side of eq. (E-21) is easily evaluated giving the
result
1/ε
(ζ) dσ 2 d
I(1) [Aμ ] = lim+ T r exp[−σD(ζ) ] P(ζ)
ε→0 ε σ dζ
 4 (0) (0)
= lim+ (4 log ε · c̄) d2 ξ ĝ φ̄n · η φ̄n (ξ). (E-25)
ε→0 D(g)
n
The final result for the functional determinants ratio eq. (E-19) is thus
given by
(C (R) )2 1 /4 0
J[Aμ ] = d2 ξ ĝ ∂α ηĝ αβ ∂β η (ξ)
π D(g) 2
C (R) ; 4 
+ d2 ξ (εμν F μν (AH μ ) · η) ĝ (ξ)
π D(g)
 4 (0) (0)
− icR d2 ξ ĝ φ̄n · η φ̄n (ξ), (E-26)
n D(g)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 62

62 Lecture Notes in Topics in Path Integrals

here c(R) is the usual multiplicative infrared coupling constant renormal-

ization due to zero-mode terms.
This result generalizes and produces a proof for the conjectured func-
tional determinant of ref. [7] which was deduced in the particular case of
nonexistence of zero modes.
The generating functional thus takes the more invariant form

Z[ρ, ρ̄] = dm(pi , ri )Z (0) [ρ, ρ̄, (pi , ri )], (E-27)

where the measure over the (pi , ri ) parameters is given by [7]

g
αiμ , αiμ  αiμ , βμj
dm(pi , ri ) = (2π)2g dpl · drl × det
βμi , αjμ  βμi , βμi 
l=1
; >4
× exp − 2π 2 d2 ξ ĝ [(pk Ω̄ki − ri )
D(g)
? 
× (Ω)−1
ij (Ω jl p l − ri )] (ξ) . (E-28)

The (bosonized) generating functional is explicitly given by

 (0) 
Z ()) [ρ, ρ̄] = dc [η] exp iW [φ̂(0) H
n , φ̂n , Aμ ]

; 1 c(R)
2
4 αβ
× d [χ]d [χ̄] exp −
c c
1− d2 ξ ĝ[ĝ ∂α η∂β η)(ξ)
2 π D(g)
μ (R)
+ (χ̄iγ (Dμ + c AH
μ )χ)(ξ)

+ (χ̄ exp[ic(R) γ5 η]ρ + ρ̄ exp[ic(R) γ5 η]χ)(ξ) , (E-29)

(0)
(0)
where the functional W [φ̂n , φ̂n , AH
μ ] is defined by the interaction with
(0)
(0)
the (external) zero-mode fermion fields φ̂n , φ̂n

(0) 4 ;/ c(R) 0
W [φ̂(0) H
n , φ̂n , Aμ ] = ĝ − i
d2 ξ εμν F μν (AH )η
D(g) π
(0) 
+ (−c(R) φ̂(0)
n η φ̂n ) (ξ). (E-30)

We remark that the fermions χ(ξ) still interact with the Hodge topolog-
μ by the minimal gauge invariant interaction D(Aμ ) and with
ical field AH H

the η(ξ) field by the coupling with the source term.

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 63

Basics Polyakov’s quantum surface theory on the formalism of functional 63

Let us exemplify our main result, eq. (E-29), by displaying the general
structure of the two-point fermion correlation function
+∞
1
ψα (ξ1 )ψ̄β (ξ2 ) = dm(pi , ri ) det[iγ μ (Dμ + c(R) AH
μ )]
Z[ρ, ρ̃] −∞
% 2
&
1 c(R)
× exp − Δ (ξ1 , ξ2 ) D−1 (AH
−1
μ ),
2 1 − (c(R)2 /π)
(E-31)

where Δ−1 (ξ1 , ξ2 ) is the Green’s function of the Laplace operator on the
Riemann surface D(g) and D−1 (AH μ
μ ) = (iγ (Dμ + c
(R) H −1
Aμ )) (ξ1 , ξ2 ) is the
Gree’s function of the Dirac operator with spin structure (θi , φi ) in the
presence of the topological Hodge vector field AH μ (ξ)[1].
The determinant in eq. (E-31) was exactly evaluated ref.[1] and
expressed in terms of ϑ-functions
1 1  1
1 i 1
det iγ μ (Dμ + c(R) AH ) = |l(Ω)| 2 1
· ϑ 2 +θ (0|Ω)1. (E-32)
μ 1 1 − φi 1
2

The Green’s function of the Laplace operator may be expressed in terms

of the theta-functions
1 (ξ1 − ξ2 )2
Δ−1 (ξ1 , ξ2 ) = − log |ϑ[(ξ1 |Ω| − ϑ[(ξ2 |Ω)]| + . (E-33)
4π det(Ω)

Finally a formal expression for the Green’s function of the Dirac oper-
ator is given by [3]
% &
c(R)
exp −i (A + γ5 εμν A )dξ × (iγ μ Dμ (AH ))−1
H ν,H μ
(φi ,θ i ) (ξ1 , ξ2 )
2 Cξ1 ,ξ2 μ
% &
c(R)
× exp −i H ν,H
(A + γ5 εμν A )dξ μ
(E-34)
2 Cξ1 ,ξ2 μ

where Cξ1 ,ξ2 is an arbitrary contour on the Riemann surface D(g) which
has a nonempty intersection with each canonical homology cycles on D(g)
and connecting the points ξ1 and ξ2 .
As we have shown, chiral changes in path integrals even for fermion
model on a Riemann surface provid a quick, mathematically and concep-
tually simple way to analyse these models.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 64

64 Lecture Notes in Topics in Path Integrals

2.14 References

[1] Alvares-Guam L., Moore G. and Vafa C., Commun. Math. Phys.,
106 (1980) 1.
[2] Belavin A.A. and Knizhnic V.G., Sov. Phys. JETP , 64 (1986) 214.
[3] Botelho Luiz C.L., Phys. Rev. D, 34 (1986) 3250.
[4] Botelho Luiz C.L., Phys. Rev. D, 35 (1987) 1515.
[5] Fujikawa K., Phys. Rev., D, 21 (1980) 2848.
[6] Botelho Luiz C.L., Phys. Rev. D, 39 (1989) 3051.
[7] Freedman D.Z. and Pilch K., Phys. Lett. B, 213 (1988) 331.

Remark: For a random surfaqce {X μ (ξ)} background on the proper-time

(light-cone) gauge
(êaμ êbν )(ξ) = ρ(ξ)δμν
with
ρ(ξ) = (∂+ X μ )(∂− Xμ )(ξ) = 0
and
(∂+ X μ )2 = (∂− X μ )2 = 0
as in Chapter 2, §2.4 and perhaps added with a Gross-Neven “mass” term
(ψ̄, ψ)2 , one expects tht the resulting generalized “Elfin” string path inte-
gral is the correct QFT definition for QCD (see also ref. [19] on contents
page xi).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 65

Chapter 3

Critical String Wave Equations and

the QCD(U (Nc)) String

We present a simple non mathematical rigorous proof that self-avoiding

fermionic strings solutions solve formally (in a Quantum Mechanical Frame-
work) the QCD(U (Uc )) Loop Wave Equations written in terms of random
loops for a certain class of self-avoiding surfaces.

3.1 Introduction

We aim, in this chapter to present a formal interacting string solution for

the Migdal-Makeenko Loop Wave Equation for the colour group (U (Nc ))
(Ref. [1] and references therein) for a special class of self-avoiding surfaces.
Our main tool to solve the Migdal-Makeenko Loop Wave Equation is
based on the remark made in the Sect. 3.2 of this chapter, where we address
tshe problem of solving critical string wave equations by string functional
integral by applying simple rules of the operatorial calculus of Quantum
Mechanics. We thus apply the results of Sect. 3.2 to present a string
functional integral solution for the Migdal-Makeenko Loop Wave Equation
for the colour group U (Nc ) at this class of special self-intersecting surfaces.

3.2 The Critical Area-Diffusion String Wave Equations

Let us start this section by briefly reviewing our general procedure to write
diffusion string wave equations for Bosonic non-critical string [2]. The first
step is by considering the following fixed area string propagator in 2D in-
duced quantum gravity string quantization framework.
4
G[C out , C in , A] = Dc [gab ]Dc [Xμ ] × δ dσdτ g(σ, τ ) − A
D
× exp(−I0 (gab , Xμ , μ2 = 0)). (3.1)

65
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 66

66 Lecture Notes in Topics in Path Integrals

Here the string surface parameter domain is taken to be the rectangle D =

{(σ, τ ), −π ≤ a ≤ π, 0 ≤ τ ≤ T }. The action I0 (gab , Xμ , .μ2 = 0) is
the Brink-Di Vecchia-Howe covariant action with a zero cosmological term
and the covariant functional measures Dc [gab ]Dc [Xμ ] are defined over all
cylindrical string world sheets without holes and handles with the initial and
finaql string configurations as unique non-trivial boundaries: i.e. Xμ (c, 0) =
C in , Xμ (σ, T ) = C out .
In order to write an area diffusion wave equation for (3.1), we exploit an
identity which relates its area variation (the Mandelstam area derivative for
strings) to functional variations on the conformal factor measure when one
fixes the string diffeomorphism group in (3.1) by imposing the conformal
gauge gab (sigma, τ ) = ρ(σ, τ )δab (see Refs. [1-3]). This procedure yields,
thusm the following area diffusion string Euclidean wave equation

π
∂ δ2 1
G[C out , C in , A = dσ − + C in (σ)2
∂A −π 2e2in (σ)δCμin (σ)δCμin (σ) 2 μ

26 − D
+ lim [R(ρ(σ, τ )) + C∞ ] × G[C out , C in , A].
24π r→0+
(3.2)

At this point a subtle difficulty appears when the theory describged by

(3.1) is at its critical dimension D = 26 since the conformal field ρ(σ, τ )
decouples from the theory, making it subtle to implement the fixed area
constrint in (3.1). It is instructive to point out that for a cylinder surface
without holes and handles with non trivial boundaries, the argument that
the fixed area constrint is simply fixing the modulus λ of the (torus) con-
formal gauge gab (σ, τ ) = ρ(σ, τ )((dσ)2 + λ2 (dτ )2 ) is insufficient to cover the
case of “string creation” from the vacuum as we will need in Sect. 3. This
is because in this case λ = 0 and the string world sheet still has a non-zero
area. Note that the topology of this string world sheet creation process is
now a hemisphere which again makes impossible the use of the modulus λ
as an area parameter.
However, it makes sense to consider the limit of the parameter D = 26
directly in our string diffusion (3.2) which reproduces the usual critical
string wave equations ((3.2) with D = 26 and ρ(σ, τ ) = 1).
In this short section we intend to show that the following critical string
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 67

Critical String Wave Equations and the QCD(U (Nc )) String 67

propagator

G[C out , C in , A] = DF [X μ (σ, τ )]

X μ (σ, 0) = Cμin (σ), X μ (σ, A) = Cμout (σ)

$ % & '
2 2
1 A π
∂X μ ∂X μ
exp − dτ dσ + (σ, τ )
2 0 −π ∂σ ∂τ
(3.3)
where the intrinsic string time parameter A is identified with the area dif-
fusion variable, satisfies the string critical diffusion wave equation.
To show this simple result we evaluate the A-derivative of (3.3) by means
of Leibnitz’s rule
∂
G[C out , C in , A]
∂A $ E % & F'
2 2
−1 π
∂X μ ∂X μ
= lim dσ + (σ, τ ) (3.4)
2 τ →A− −π ∂σ ∂τ

where the surface average  s is defined by Bosonic path-integral in (3.3).

In order to translate the path integral relation (3.4) into a operator
statement, we use the usual Heisenberg Commutation Relations for two-
dimensional (2D) free fields on D (with the Bidimensional Planck constant
= Regge slope parameter set to the value one)
[Πμ (σ, τ ), Xν (σ  , τ )] = iδ(σ − σ  )δμν (3.5)
and its associated Schrdinger representation for τ = A (that are the quan-
tum mechanical definition of the lopp derivatives operators [1]).
9 :
∂ μ δ
Πμ (σ, A) = lim X (σ, τ ) = +i out (3.6)
τ →A ∂τ s Cμ (σ)
1 out 1
1 dCμ (σ) 12 μ 2
1 1 = lim ∂X (σ, τ ) . (3.7)
1 dσ 1 τ →A ∂τ
After substituting (3.6)–(3.7) into (3.4) we obtain the desired result
π 
∂ δ2 1 out
G[C out , C in , A] = − dσ − + |C (σ)| 2
.
∂A −π 2δCμout (σ)δCμout (σ) 2 μ
(3.8)
Let us point out that general string wave functionals (the Schrdinger
representation for the theory’s quantum states) may be formally expanded
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 68

68 Lecture Notes in Topics in Path Integrals

in terms of the eigenvunctions of the quantum string Hamiltonian (the

string wave operator in (3.8))
 π 
δ2 1 
−Δc ψE [c] = − dσ − |C (σ)|2 ψE (c) = EψE [c]
−π 2δCμ (σ)δCμ (σ) 2 μ
(3.9)

ψ[c] = ρ(E)ψe [c]. (3.10)
{E}

The functionals endowed with the (formal) inner product given by

ψ[c]|Ω[c] = DF [c] · ψ ∗ [c] · Ω[c] (3.11)

constitute a Hilbert space where the string Laplacian −Δc is formally a

Hermitian operator.
It is worth remarking that an explicit expression for the Green’s Func-
tion

(−Δc )−1 (C out , C in ) = ∗
ψE [C out ]ψE [C in ]/E
{E}

of the string Laplacian in terms of the cylindrical string propagator (3) may
be easily obtained.
In order to deduce this expression we integrate both side4 of (3.8) with
respect to the A-variable. Considering now the Asymptotic Behaviors.

lim G[C out , C in , A] = 0 (3.12)

A→∞

lim G[C out , C in , A] = δ F (C out − C in ) (3.13)

A→0

we obtain the relationship

∞
ΔF (C out − C in ) = −Δc dAG[C out , C in , A] (3.14)
0

leading thus to the following identify

∞
(−Δc )−1 [C out , C in ] = dAG[C out , C in , A] (3.15)
0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 69

Critical String Wave Equations and the QCD(U (Nc )) String 69

3.3 A Bilinear Fermion Coupling on a Self-Interacting

Bosonic Random Surface as Solution of QCD(U (Nc ))
Migdal-Makeenko Loop Equantion

Let us start this section by considering the (non-renormalized) Migdal-

makeenko Loop Equation satisfied by the Quantum Wilson Loop in the
form of Ref. [4] for the colour group U (Nc ).
π
dX μ (σ) dX μ (σ  )
− Δc Wl [CX(−π)X(π)  = (g 2 Nc ) dσ  ·
−π dσ dσ 
× δ (D) (Xμ (σ) − Xμ (σ  )Wkp [CX(−π)X(σ) ]Wp [CX(σ)X(π) ]. (3.16)
The Quantum Wilson Loop is given by
Wk [CX(−π)X(π) , Aμ (x)]
9 π :
1 (c) 
= TR exp − dσ(Aμ (Xμ (σ)) · X (σ)
μ
. (3.17)
Nc −π k
As usual, Aμ (x) denotes the usual U (N ) colour Yang-mills field which
possesses an additional, not yet specified intrinsic global “Flavor” group
O(M ) represented by matrix indices (k, ). The average   is given by the
U (N )-colour Yang-Mills field theory.
Let us consider the following critical non-linear interacting Fermionic
String theory first considered in Ref. [5]
% &
2 2
1 A π
∂X μ ∂X μ
S[Xμ (σ, τ ), ψ(k) (σ, τ )] = dτ dσ + (σ, τ )
2 0 −π ∂τ ∂σ
A π
+ dτ dσ[ψ̄(k) (γ a ∂a )ψ(k) ](σ, τ )
0 −π
A A π π
β
+ dτ dτ  dτ  dσ 
2 0 0 −π −π

× (ψ(k) ψ̄(k) )(σ, τ ) × T μν

(Xα (σ, τ ))δ (D)
× (Xα (σ, τ ) − Xα (σ  , τ  ))Tμν (Xα (σ  , τ  )).
(3.18a)
The notation is as follows: the string vector position is described by
the 2D-fields Xμ (σ, ζ) with the Drichlet boundary condition Xμ (σ, A) =
CX(−π),X(π) ; i.e., our special class of self-intersecting surfaces S =
{Xμ (σ, ζ), −π ≤ σ ≤ π, 0 ≤ ζ ≤ A) has a unique boundary the fixed Loop
CX(−π),X(π) of (3.16). The surface orientation tensor where it is defined is
given by
εab ∂a X μ ∂b X ν
Tμν (Xμ (σ, τ )) = √ √ (3.18b)
2 h
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 70

70 Lecture Notes in Topics in Path Integrals

with h = det{hab } and hab (σ, τ ) = ∂a X μ ∂b X μ . Note that S possesses

self-intersecting lines such that Xμ (σ, τ ) = Xμ (σ  , τ  ) with 0 < {τ, τ  } <
A has non-trivial self-intersecting lines solutions. For τ = A = τ  ,
Xμ (σ, A) = Cμout (σ) posseses solely simple isolated self-intersections points
(eights loops), however with T μν (Xα (σ, A))Tμν (Xβ (σ  , A)) = 0 for σ = σ 
(the Fermion Exclusion Pauli Principle). Additionally we have introduced
a set of single-valued intrinsic Majorana 2D-spinors on the surface domain
parameter D = {(σ, τ ), 0 ≤ τ ≤ A; −π ≤ 0 ≤ τ }. They are chosen to be-
long to a real representation of the flavor group O(22) since for this group
we have cancelled exactly the theory’s conformal anomaly (26 = 4 + 22),
which in turn leads to the vanishing of the kinetic term associated to the
conformal factor ρ(σ, ζ) (see Ref. [1]). We further impose as a boundary
condition on these Fermions the vanishing of the Fermion energy-tensor
projected on the Loop CX(−π),X(π) . Let us point out that the Weill sym-
metry makes sense to speak in conformal anomaly in our theory (3.16)
which preservation at quantum level by its turn will determine the string
flavor group to be the “String” Weinberg-Salam group O(22) (see Ref. [1]).
Associated to the non-linear strings theory (3.18) we consider the fol-
lowing Fermionic propagator for a fixed string world sheet {Xμ (σ, τ )}
Z̄k [CX(−π),X(π) ; Xμ (σ, ζ), A]

= DF [ψk (σ, τ )](ψ(k) (−π, A)ψ̄() (π, A)

$ '
A π
× exp − dτ a
dσ(ψ̄(k) (γ ∂a )ψ(k) )(σ, τ )
0 −π
> β A π A π
× exp − dτ dσ dτ  dσ  (ψ̄(k) ψ(k) )(σ, τ )
2 0 −π 0 −π
?
×T μν
(Xα (σ, τ ))δ (D)
(Xα (σ, τ ) − Xα (σ  , τ  ))Tμν (Xα (σ, τ  )) . (3.19)
The basic idea of our string solution for QCD(U (Nc )) is a technical
improvement of Ref. [1] and consists in showing that the surface averaged
propagator (3.19)
Z̄k [CX(−π),X(π) , Xμ (σ, τ ), A]s = Gk (CX(−π),X(π) , A)
when integrated with respect to the A-parameter as in (3.15), Now satisfies
the full U (Nc ) non-linear Migdal-Makeenko Loop Equation (3.16) instead
of the factorized Loop equations associated to the T’Hooft limit Nε → ∞.
The surface average  , is defined by the free Bosonic action piece of
(3.18) as in Sect. 2. In this context we consider Gk (CX(−π),X(π) , A) as
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 71

Critical String Wave Equations and the QCD(U (Nc )) String 71

the non-linear string propagator describing the “creation” of the Loop

CX(−π),X(π) = C out from the string vacuum, which is represented here
by a “collapsed” point-like string initial configuration C in ≡ (x) (x denotes
an arbitrary point of the surface which may be considered as such initial
string configuration).
Let us thus, evaluate the A-derivative of G(CX(−π),X(π) , A)

∂
Zk (CX(−π),X(π) , As
∂A
= DF [X μ (σ, τ )] DF [ψ(k)(σ, τ )]

× exp[(−S[Xμ (σ, τ ), ψ(k) (σ, τ ))] × ψ(k) (−π, A)ψ̄() (+π, A)

 
∂ > A π
1
× (−1) × dτ μ 2 a
dσ (∂X ) + ψ̄(k) (γ ∂s )ψ(k) (σ, τ )
∂A 0 −π 2
a π A π
β
+ dτ dσ dτ 
dσ  (ψ(k) ψ̄(k) (σ, τ )T μν (Xα (σ, τ ))
20 −π 0 −π
?
× δ (Xα (σ, τ ) − Xα (σ , τ )Tμν (Xα (σ  , τ  ) .
(D)  
(3.20)

The free Bosonic term in the right-hand side of (3.20) leads to the string
Laplacian as in (3.4) of Sect. 2. The free Fermion term

lim ψ̄(k) (σ, τ )(γ a ∂a )ψ(k) (σ, τ )

τ →A−

 π 
 (D) 
F μ
D [X (σ, A)] D [ψk (σ, τ )] F
dσ δ (Xμ (σ) − Xμ (σ )
−π
% 22 &
dX μ (σ) dX μ (σ  ) 
× · ψk (−π, A) · ψ̄ (+π, A) × (ψp (σ, A) · ψ̄p (σ, A))
da da p=1

× exp(−S[Xμ (σ, τ ), ψk (σ, τ )])

π π
dX μ (σ) dX μ (σ  )
=β dσ dσ  δ (D) (Xμ (σ) − Xμ (σ  ) ·
−π −π dσ dσ 
× (Z̄kp [CX(−π)X(σ) , Xμ (σ, τ )] × Z̄p [CX(σ)X(π) , Xμ (σ, τ ), A]), (3.21)

vanishes as a consequence of our imposed vanished energy-momentum ten-

sor boundary conditions on the intrinsic Fermion field. The evaluation of
the boundary limit on β-term requires explicitly that the surface {Xμ (σ, τ )}
does not possesses self-intersections of the type Xμ (σ, A) = Cμout (σ) =
Xμ (σ  , τ  ). The result of this boundary limit evaluation is given explicitly
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 72

72 Lecture Notes in Topics in Path Integrals

1
by the expression
⎛ below (Ref.
⎞ [1]-Appendix B )
$ '
22
⎜ ⎟ A π
⎜ dψ(ξ, τ )⎟ a
⎝ ⎠ exp dτ dξ(ψ̄k (γ ∂a )ψk )(ξ, τ )
p=1 −π≤ξ≤π 0 −π
0≤τ ≤A
> β A A π π
× exp − dτ dτ 
dξ  (ψ̄k ψk )(ξ, τ )Tμν (ξ, τ ))
dξ
2 0 0 −π −π
?
× δ (D) (Xα (ξ, τ ) − Xα (ξ  , τ  ))T μν (X(ξ  , τ  ))
× (ψk (−π, A)ψ̄ (π, A)ψp (σ, A)ψ̄p (σ, A)
⎛ ⎞
; ⎜
22
⎟
= ⎜ dψ(ξ, τ )⎟
⎝ ⎠
p=1 −π≤ξ≤σ
0≤τ ≤A
$ '
A π
× exp dτ a
dξ(ψ̄k (γ ∂a )ψk )(ξ, τ ) (ψk (−π, A)ψ̄p (σ, A))
0 −π
> β A A π π
× exp − dτ dτ 
dξ  (ψ̄k ψk )(ξ, τ )Tμν (ξ, τ ))
dξ
2 0 0 −π −π
?
× δ (D) (X(ξ, τ ) − X(ξ  , τ  ))T μν (X(ξ  , τ  ))
⎛ ⎞
$ '
; ⎜ ⎟ A π
× ⎜ ⎟
dψ(ξ, τ )⎠ exp dτ a
dξ(ψ̄k (γ ∂a )ψk )(ξ, τ )
⎝
σ<ξ<π 0 σ
0≤τ ≤A
> A A π π
× exp −β dτ dτ 
dξ  (ψ̄k ψk )(ξ, τ )Tμν (X(ξ, τ ))
dξ
0 0
? σ σ

× δ (D) (X(ξ, τ ) − X(ξ  , τ  ))T μν (X(ξ  , τ  )) × (ψp (σ, 0)ψ̄ (π, 0))
(3.22)
1 Note that either crucial result below, due to the special class of surfaces chosen
A A σ−ε π
lim dτ dτ  dξ dξ  [(ψ̄k ψk )(ξ, τ )]Tμν (X(ξ, τ ))
ε→0 0 0 −π σ+ε
ε>0

× δ(D) (X(ξ, τ ) − X(ξ  , tau ))T μν (X(ξ  , τ  )) = 0 (3.21-a)
since our orientation tensor strings world-sheet {X(ξ, τ )} is such that for ξ = ξ  and
ζ, ζ  ∈ [0, A]
T μν (X(ξ, τ )) · Tμν (X(ξ  , τ  )) = 0 (21-b)
(ζ̂)
(these “string fixed-time” loop Xμ (ξ, ζ̂) ≡ eμ (ξ) possesses solely simple iso-
lated self-intersections points (“eights” loops) where the non-trivial tangent lines
at theses self-intersect points are supposed to be always orthogonal to each other:
Tμν (X(ξ, ζ))T μν (X(ξ  , ζ)) = 0 as a remnant of the Fermion Exclusion Pauli Principle
still acting for these Bosonic Pieces Cμ (ξ) of the full fermionic “quark trajectories”).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 73

Critical String Wave Equations and the QCD(U (Nc )) String 73

and its unity normalization condition

Z̄pp [CX(σ)X(σ) , Xμ (σ, ξ), A] = 1. (3.23)
By imposing the identification
g 2 Nc = β
between the QCD(U (Nc )) gauge coupling constant and our non linear
string theory described by (3.3) we obtain the identification between the
QCD(U (Nc )) Wilson Loop (3.17) and the surface averaged Fermion Prop-
agator (3.19)
Wk [CX(−π)X(π) , Aμ (x)]Yang Mills U(Nc )
∞
= dAZ̄k [CX(−π )X(π) , Xμ (σ, ξ), A]s . (3.24)
0
The above equation is the main result of this note and generalizes to
the case of U (Nc ) colour group our previous studies made for the t’Hooft
limit of Ref. [1].
Finally we remark that by considering an ultra-violet cut-off on the
space-time, ΔX μ (σ, τ ) ≥ 1/Λ, our proposed self-avoiding string theory
(3.18) in the case of non dynamical 2D-Fermions ((ψk ψ̄k  = μ = constant)
produces the extrinsic string with the topological invariant of string worl-
sheet self-intersections number as an effective string theory for the proposed
QCD string as conjectured in the first Ref. [6-8] (see Ref. [9] for this study
and the enclosed Appendix).
Finally it is worth re-write (3.19) in a form where appears an interaction
with an external white-noise Gaussian auxiliary anti-symetric tensor field
as suggested in Ref. [1]. Namely:
$ '
β A A

π
exp − dτ dτ dτ (ψ̄(k) (ξ, τ )ψ(k) (ξ, τ )) × T (Xα (ξ, τ ))
μν
2 0 0 −π

× δ (D) (X(ξ, τ ) − X(ξ  , τ  ))Tμν (Xβ (ξ  , τ  ))

   
1
= D Bμν (x) exp −
F
Bμν (x)d x × exp i
2 D
Bμν (x)J μν
(x, S)
2
(3.25)
with the dynamical string world-sheet current
; π A /
22 01/2
J μν (x, S) = β dξ dτ T μν (X(ξ, τ )) ψ̄(p) ψ(p)
−π 0 p=1

×δ (D)
(x − X(ξ, τ )) . (3.26)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 74

74 Lecture Notes in Topics in Path Integrals

3.4 Appendix A: A Reduced Covariant String Model for

the Extrinsic String

Our aim in this appendix is very modest: we write a covariant action for
the elastic string and quantize in the Polyakov’s path integral framework a
truncated version of the covariant written theory.
Let us start our study by considering the classical action for the elastic
string in the conformal gauge.
% &
2
1 1 2 λab
S0 = d zρ + γ d zρ − ∂ X + i
2 2
(∂a X∂b X − gab ) .
2πα ρ ρ
(A-1)
The string surface is described by X = X(z), where X is the surface
vector position in D Euclidean dimensions; za (a = 1, 2) are the coor-
dinates of the world sheet. The first term in (A-1) is the Nambu term
with the string tension equal to 1/2πα. The second term is the square of
the extrinsic curvature with the rigidity coupling constant denoted by γ
(γ = limNc →∞ (g 2 (Nc )!) and λab (z) is a Lagrange multiplier which insures
that the metric (gab ) coincides with the intrinsic metrics (∂a X∂b X).
Let us consider a covariant version of action (A1) by promoting ρ(z) =
gab (z) to be a dynamical field. This procedure yields the following action
1 √
S1 [X(z), gab (z), λab (z)] = d2 z g
2πα
√
+ d2 z g [γ(−Δg X)2 + iλab (gab − ∂a X∂b X)].
(A-2)
4
Here g(z) = det(gab (z)) and −Δg = − √1g ∂a (g ab ∂b ) is the Laplace-
Beltrami operator associated to the intrinsic metric gab (z).
In the Polyakov’s path integral quantization effective framework the
partition functional for the theory (A-1) should be given by

Z= Dc [gab ]Dc [X]Dc [λab ] × exp −S1 [X(z), gab (z), λab (z)] (A-3)
where the functional measures are the DeWitt covariant functional mea-
sures [1].
Let us suppose that the constraint field in approximated by the intrinsic
metric λab (z) = λgab (z). (The covariant version of the usual mean field
approximation λab (z) = i(λ)δab with (λ) a positive fixed value.) As a
consequence of this hypothesis we get the truncated theory.
Z(T ) =∈ Dc [gab ]Dc [X] exp −S (T ) [gab (z), X(z)] (A-4)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 75

Critical String Wave Equations and the QCD(U (Nc )) String 75

where the truncated action theory is written as

1 4
S (T ) [gab , X] = 
d2 z g(z) + γ d2 z[(−Δg X)2 ]
2πα
√ 4
+ λ d2 z g g ab ∂a X∂b X + λ d2 z g(z). (A-5)

For the evaluation of the X-functional integral in (A-4) we consider the

non-local variable change
Xμ (z) = (−i(Δg )1/2 ϑμ )(z), μ = 1, . . . , D.
−1/2
Here −i(Δg ) is a well defined self-adjoint (pseudo-differential) operator.
The truncated action takes the following form similar to a massive scalar
field in the z domain
1 √ 1
S (T ) [gab , ϑ] = 
− λ d2 z g + 2λ ∈ d2 z ϑ2
2πα 2
1 √
+ 2γ d2 z ( g ϑ(−Δg )ϑ)(z). (A-6)
2
The change in the (covariant) functional measure Dc [x] is given by
Dc [x] = (det(−Δg )−1 )D/2 × Dc [ϑ]. (A-7)
The main step in our calculation is to define the above written function
determinant as det−D/2 (−Δg ). By choosing the conformal gauge gab =
eϕ δab and evaluating the covariant Gaussian ϑ-functional intetral we obtain
the partial result [1]
 
26 − D 1
Z(T ) = D[ϑ] exp − 2 2 ϕ 2
(∂a ϕ) + μR e d z
48π 2
× det −D/2 (−2γΔg + 2λ) (A-8)
where
2−D 1
μ2R = lim
+ = λ
ε→0 4πε 2πα
may be though as a renormalization of the bare string thension 1/2πα .
We analyze now the unrenormalized functional determinant
λ
exp −SEF F [ϕ] = det −D/2 −Δg + .
γ
By defining it by a propertime prescription we obtain the counterterms of
the above written action. Explicitly
∞
D dT λ
SEF F [ϕ] = lim − T r exp −T −Δg + . (A-9)
ε→0 2 ε T γ
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 76

76 Lecture Notes in Topics in Path Integrals

Now it is well known that the counterterms of SEF F [ϕ] are determined
by the asymptotic expansion of the diagonal part of massive Laplace-
Beltrami operator which is tabulated

λ
lim T r exp −T −Δg +
T →0 γ
 ϕ 
e 1 1 1 ϕ λ
= d2 z lim+ − Δϕ + e · (z). (A-10)
2π T →0 T 2π 2π γ

By substituting (A-10) in to (A-9) we get straightforwardly the following

counterterms associated to the two-dimensional intrinsic “mas” λ/γ

D 1 λ 1
· · lg d2 zeϕ(z) . (A-11)
2 2π γ ε

So, on the basis of the counter term (A-11) we have the following renor-
malization law for the inverse of the rigidity β = 1/γ (by choosing λ = 1)

1 1 D 1
= − · lg(ε). (A-12)
βR β0 2 2π

Equation (A-12) yields the intrinsic two-dimensional momentum depen-

dence of the running coupling constant β

D β0 ε
βR (p2 ) = β0 (p2 = 0)/1 − · lg . (A-13)
2 2π p2

It is instructive point out the D/2 factor in (A-13) which appears in a

natural way in our calculations.
Since it is naively expected that the string perturbative pase p2 -small
(p ∈ R2 ) would corresponds to the underlying QCD field theory at its
non-perturbative phase k 2 → +∞ (k ∈ R4 ), one can see that (A13) sug-
gests a natural explanation from the QCD’s String Representation for the
“strange” QCD field theory description of the asymptotic behavior for the
coupling constant at large Nc , namely

lim lim (g 2 Nc )ren (k 2 ) = 0. (A-14)

k2 →∞ Nc →∞

As a general conclusion, one can see that still exists a great deal of
not completely understood phenomena in QCD out of non-analytical field
theoretic continuum approaches-lattice approximations.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 77

Critical String Wave Equations and the QCD(U (Nc )) String 77

3.5 Appendix B: The Loop Space Program i the Bosonic

λφ4 − O(N )-Field Theory and the QCD Triviality for
RD . D > 4

Let us start our study by considering the (bare) generating functional of

the Green’s functions of the O(N ) (symmetric phase) λφ4 field theory in a
D-dimensional Euclidean space-time
N 2
N > λ 
0
a
Z[J (x)] = dμ[Φ (x)] × exp −
a D
d x a
Φ (x)2
(x)
a=1
4 a=1
N
 ?
− d x D
J a (x)Φa (x) (B-1)
a=1
a
where Φ (x) denotes a N -component real scalar O(N ) field, (μ0 , λ0 ) the
(bare) mass and coupling parameters and the Gaussian functional measure
in (B-1) is
 N 2
N N > 1 
a
dμ[Φ (x)] = dΦ (x) exp −
a D
d x a
(∂μ Φ ) (x)
a=1 a=1 D
2 a=1
x∈R


N ?
+ μ20 (Φa )2 (x) . (B-2)
a=1

Now, in order to get an effective expression for the functional integrand

(B-2), where we can evaluate the Φa functional integrations, we write the
intersection λφ4 term in the following form
⎧ N ⎫
⎨ λ 
2
⎬
0
exp − dD x (Φa (x))2
⎩ 4 ⎭
a=1
$ N '

= dμ[σ] · exp −i d xσ(x) D a
(φ (x))2
(B-3)
a=1

where σ(x) is an ausiliary scalar field and the σ functional measure in (B-3)
is given by
  
1 D 2
dμ[σ] = dσ(x) exp − 2
d x σ (x) (B-4)
D
2 λ0
x⊂R

with covariance
λ0 (D)
σ(x1 )σ(x2 )σ = dμ[σ]σ(x1 )σ(x2 ) = δ (x1 − x2 ). (B-5)
2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 78

78 Lecture Notes in Topics in Path Integrals

The last result allows us to consider the δ(x) field as a random Gaussian
potential with noise’s strength λ20 .
After substitution of (B-4) into (B-2), we can evaluate explicitly the
Φ-functional integrations since they are of Gaussian type. We, thism get
the result

Z[J a (x)] = dμ[σ] det −N/2 (−Δ + μ20 − 2iσ)

 
1
× exp d xd yJ (x)(−Δ + μ0 − 2iσ)δab J (y) . (B-6)
D D a 2 B
2

At this point of our study we implement the main idea: by following

symanzik’s analysis, we express the σ-functionals integrands in (B-6) as
functional defined in the Feynman-Kac-Wiener space of Random paths by
making use of the well known random path respesentation for the non-
realativistic euclidean propagator of a particle of mass μ0 in the presence
of the external random Gaussian potential σ(x)
∞
(−Δ + μ20 − 2iσ)−1 (x, y) = dζG(x, y, σ)(ζ) (B-7)
0

∞
dζ
log det(−Δ + μ20 − 2iσ) = − dD xF (x, x, σ)(ζ) (B-8)
0 ζ

where the non-relativistic propagator is given by


dD σ(z)j(wwy
(ζ)
G(x, y, σ)(ζ) = dμ{wxy }e
(ζ) i (z))
(B-9)

with the Feynman-Kac-Wiener path measure

⎛ ⎞
⎜ ⎟ $ '
⎜ ⎟ ζ 2
⎜ ⎟ 1 dw 1
(ζ)
dμ[wxy ] =⎜ dw[α]⎟ exp − − μ20 ζ (B-10)
⎜ ⎟ 2 dα 2
⎝ 0<α<ζ ⎠ 0
w(0)=x
w(0)=y

and the (random) world-line currents defined by

ζ
(ζ)
j(wxu )(z) = δ D (z − wxu
(ζ)
(α)) dα. (B-11)
0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 79

Critical String Wave Equations and the QCD(U (Nc )) String 79

So, we obtain the proposed reformulation of λφ4 O(N )-theory as a theory

(ζ)
of random paths {wxy (α)} in the presence of a random Gaussian potential
>N ∞
dζ
Z[J a (x)] = dμ[σ] × exp dD x (ζ)
dμ[wxx ]
2 0 ζ
?
+ exp i dD zσ(z)j(wxx
(ζ)
)(z)

>1 
N ; ∞
× exp dD xD y ja (x) dζ (ζ)
dζ[wxy ]
2 s=1 0
 ?
× exp i dD zσ(z)j(wwy
(ζ)
δab · Jb (y) . (B-12)

We shall use ther andom path formulation (B-12) to analysie the cor-
relation functions of the λφ4 theory. As a useful remark, we note by using
(B-12) that the general k-point (bare) correlation function possesses the
general structure for free fields

Φil (x1 ) . . . Φik (xk )Φ

⎧
⎪
⎪ 0 if k = 2j + 1
⎪
⎨ @
= Φil (x1 )Φi2 (x2 )Φ . . . Φik−1 (x2j−1 )Φik (x2j )Φ
⎪
⎪ (2j+1)
⎪
⎩-pairings if k = 2j
(B-13)

where the quantum averages  Φ in (B-13) are defined by the λφ4 partition
functional Z[0] (see (B-1) with J a (x) ≡ 0).
Because of this result, we have solely to study the properties of the
2-point correlation function
C ∞
Φil (x1 )Φi2 (x2 )Φ = δi1 i2 dζdμ[wx(ζ)
1 x1
]
0
> ?
× exp i dD zσ(z) · j(wx(ζ)
x
1 2
)(z)
>N ∞
dζ
× exp dD x dμ[x(ζ)
xx ]
2 0 ζ
; ?D
× exp i dD zσ(z)j(x(ζ)
xx (z)) . (B-14)
0

Let us evaluate the σ-functional averages  σ in (B-4) and (B-5)). For

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 80

80 Lecture Notes in Topics in Path Integrals

this task we expand the “close path term”in powers of N . Explicitly

N >
∞ k ∞
dζ
Φi1 (x)Φi2 (y) = δi1 i2 dD x
2 0 ζ
k=0 =1
∞ ?
× dμ[wx(ζ) x
 
] × dζ × dμ[w (ζ)
xy ]
0
C > 
× exp i k dD z σ(z ) × j(wx(ζx) )(z )
=1
?D
+i dD σ(z)j(wxy
(ζ)
)(z) (B-15)

and since the σ-average in (B-15) is of the Gaussian type we can perform
it exactly. The result reads
N >
∞
 k ∞
dζ
Φi1 (x)Φi2 (y)) = δi1 i2 dD x dD x dμ[wx(ζx) ]
2 0 ζ
k=0 =1
∞ > λ ;/  k ζ ζ
× dζ × (ζ)
dμ[wxy ] × exp − 2× dα × dα
0 4  0 0
=
0
× δ (D) (wx(ζx) α ) − wx(ζx) (α ))
/
k ζ ζ 0
+ dα δ (D) (wx(ζx) (α ) − wx(ζx) (α ))
= 0 0

/ 
k ζ ζ 0
+ 2× dα × dαδ (D) (wx(ζx) (α ) − wxy
(ζ)
(α) − wxy
(ζ)
(α))
=1 0 0
/ ζ ζ 0?
+ dα dα δ (D) (wxy
(ζ)
(α) − wxy
(ζ)
(α1)) . (B-16)
0 0
The above expression is the two-point correlation function of the λΦ4 −
O(N )-theory expressed as a sustem of interacting random paths with a
repulsive self-interaction at these points where they crosses themselves.
Now we can offer a topological explanation for the theory triviality
phenomenon for D > 4. At first, we note that the correlation function
(B-17) will differ from the free one, namely
∞
Φi1 (x)Φi2 (y)FREE = δi1 i2 (ζ)
dζdμ[wxy (B-17)
0
if the path intersections implied by the delta functions in (B-16) are non-
empty sets in the RD space-time. We intend to argument that those in-
tersection sets are empty for space-time with dimensionality greater than
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 81

Critical String Wave Equations and the QCD(U (Nc )) String 81

four. At first we recall some well-known concerpts of topology the topo-

logical Hausdorff dimension of a set A embedded in RD is d (with d being
a real number) if the inimum number of D-dimensional spheres of radius
γ needed to cover it, grow like γ −d when r → 0. The rule for (generical)
intersections for sets A and B (both are embeddeds in RD ) is given by

d(A ∩ B) = d(A) + d(B) − D (B-18)

where a negative Hausdorff dimension means no (generical) intersection or
equivalently the set A ∩ B is empty.
As is well known the Hausdorff dimension of the random paths in (B-16)
is 2. A direct application of the rule (B-18) gives us that the intersection
sets in (B-16) possesses a Hausdorff dimension 4 − D. So, for D > 4 these
sets are empty and leading to the triviality phenomenon (see (B-17)).
Finally we make some comments on the analyses of the divergencies in
the random path expression (B-16) for D ≤ 4. As a first observation we
note that all the path integrals involved in (B-16) can be exactly evaluated
by making a power series in λ0 . The resulting proper-times ζ integrals will
in general be divergents. Bu using a regularization (such as a cut off for
small proper-times) one can show that the divergencies can be absorved by
a renormalization of the bare mass μ0 and the action path term in (B-16)
(or equivalently, a wave-function and λ0 -coupling renormalization in the
field formulation (B-1)).
At this point of our remarks and comments, it is worth to point out
that there is no simple relation between our random loop space approach
for QCD where the loop defining the string world-sheet boundary is anon-
differentiable path and representing rigorously the functional determinant
associated to the matter ontent E [this means that there is no pure Yang-
Mills quantum theory without matters source in our approach (no rings
of Gluons!)]; and others approach based on suibable3 supersymmetric σ-
models formulations for conformal superstrings moving in non quantum
back-grounds (see [10] and [11]). Note that in this case there is still no true
derivation of this string/gauge field duality from first principles.
In our string representation for Bosonic QCD as we have proposed in
this note, one can see that the Hausforff dimension of the continuous man-
ifolds sampled by the (euclidean) quantum string vector position is four
(a very rough Brownion Bosonic Surface filling u any four-volume in R4 ).
However, it is expected that the Hausdorff dimension of the manifold sam-
pled by the 2D-Fermion Field should be minus two. Combining these re-
sults one can see that the effective Hausdorff dimension of the QCD string
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 82

82 Lecture Notes in Topics in Path Integrals

world-sheet is two, so allowing one assplies all concepts of classical smooth

Differential Topology and Geometry. Ilf all these results turn out to be
rigorous, one can see that our self-avoiding fermionic string representa-
tion gives a “proof” that QCD(U (Nc )) should be expected to be a trivial
quantum field theory (with on infrared cut off!) for space-times dimension
greater than four.
Finally, we should remark that our proposal for string representations
in QCD has no apparent overlap with those proposal relying heavily in
the existence of the string Liouville field theory as a bonafide 2D Field
Theory as proposed in [12] even if they can be interpreted as an extra
(umphysical) five dimension coordinate after some conformal impositions
in the non-critical string theory.
It appears interesting to remark that these Kaluza-Klein string repre-
sentations for N = 4 Supersymmetric QCD may be considered as “mod-
ern/geometrical/topological” version of the old beautiful result in String
Theory that Strings with U (N ) Chan-Paton factors leads formally to Mass-
less and Massive Yang-Mills scattering amplitude in its low energy limit of
vanishing Regge Sloppe limit [13].

3.6 References

[1] Botelho, L.C.L.: Methods of Bosonic and Fermionic Path Inte-

gral, Representations – Continuum Random Geometry in Quantum
Field Theory. Nova Science, New York (2008).
[2] Botelho, L.C.L.: Phys. Rev. D 40, 660 (1989).
[3] Botelho, L.C.L.: Phys Rev. D 41, 3283 (1990).
[4] Botelho, L.C.L.: Phys. Lett. B 152, 358 (1985).
[5] Botelho, L.C.L.: Rev. Bras. Fis. 16, 279 (1986).
[6] Polyakov, A.: Nucl. Phys. B 268, 406 (1986).
[7] Peliti, L., Leibler, S.: Phys. Rev. Lett. 54, 1690 (1980).
[8] Duplanter, B.: Commun. Math. Phys. 85, 221 (1982).
[9] Karanikas, A.I., Ktorides, C.H.: Phys. Lett. B 235, 90 (1990).
[10] Fradkin, E.S., Tseytlin, A.A.: PLB ISS, 316 (1981).
[11] Maldacena, M.: Phys. Rev. Lett. 80, 4859 (1988).
[12] Gibson, S.S.: Klebanov, I.R., Plyakov, A.M.: Gauge theory corre-
lators from non-critical string theory. arXiv:hep-th/19802109.
[13] Botelho, L.C.L.: PRD 35(4), 1515–1518 (1987).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 83

Chapter 4

The Formalism of String Functional

Integrals for the Evaluation of the
Interquark Potential and Non Critical
Strings Scattering Amplitudes1

We present news path integral studies on the Polyakov Non-Critical and

Nambu-Goto critical strings theories.
We also evaluate the long distance asymptotic behavior of the interquark
potential on the Nambu-Goto string theory with an extrinsic term. We also
propose an alternative and new view to covariant Polyakov’s string path
integral with a fourth-order two-dimensional quantum gravity.

4.1 Introduction

One of the most promising mathematical formalism for a physically sensible

description of strong interactions is quantum chromodynamics. In strong
interaction physics the image of an physically detectable mesonic quantum
excitation is, for instance, the quantum mechanical color invariant proba-
bility of the appearance of a pair quark-antiquark bounding a space-time
non abelian gluon surface connecting both the pair’s particle.
It appears tantalizing for mathematical formulations to consider as fun-
damental gauge invariant dynamical variable, the famous quantum Wilson
loop, with the loop C (defining the non-abelian holonomy factor); being
given by the quark-antiquark (space-time) Feynman trajectory ([1], [2])
(see also chapters 1 and 2).
It is thus searched loop space dynamical equations (at least on the
Lattice on the formal grounds) for the quantum Wilson loop wich supports
hopes for a complete string solution for Q.C.D., at least at the large number
of colors ([1]) (see also chapter 3).
1 Complements to Chapter 2.

83
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 84

84 Lecture Notes in Topics in Path Integrals

It is natural thus to study path integral geometrodynamical string prop-

agators in order to make such connection between string and QCD mathe-
matically more precise ([3]) (also chapter 2 and 3).
Chapter 4 is organized as follows: In section 2, we present a detailed
stuty of the Nambu-Goto string path integral. In section 3 and appendixes
A and B, we extend the results of section 2 to the case of the presence of
extrinsic geometry. In the section 3 also, we present a new result on the
large distance asymptotic behavior of the interquark potential by supposing
QCD represented by the Nambu-Goto extrinsic string at large distance.
In section 4 and appendix C, we present a new proposal for the
Polyakov’s Non-Critical String.

4.2 Basics Results on the Classical Bosonic Surface Theory

and the Nambu-Goto String Path Integral

Let us start our considerations by considering a given continuously differ-

entiable globally orientable; compact surface S, with a boundary given by
a non self-intersecting smooth curve C and fully immerse on the space-time
RD . Its mathematical description is described by a C 1 (Ω) two-dimensional
vector field Xμ (ξ1 , ξ2 ), (μ = 1, . . . , D) on a two-dimensional domain Ω
(compatible with the fixed-prescribed topology of S).
The Nambu-Goto-Buff-Lovett-Stillinger ([1]) area is given by the usual
geometric integral ((ξ1 , ξ2 ) := ξ)

A(Sc ) = (det{hab (ξ)})1/2 d2 ξ (4.1)

Ω
with hab (ξ) = (∂a X μ ∂b Xμ )(ξ) denotes the metric tensor induced on the
surface Sc .
Note that the scalar of curvature
R(X μ ) = {hab (Ric)ab }(X μ (ξ));
; 1 / ∂ 2 hac ∂hbc ∂2
Ricab (X μ (ξ)) = Rabcd = − −
2 ∂xb ∂xd ∂xa ∂xd ∂xb ∂xc
2
∂ gcd 0
+ a c + grs (Γrac Γsbd − Γrad Γsbc ) c
Racb (X μ (ξ))
∂x ∂x
= (∂c Γcba − ∂b Γcca + Γdba Γccd − Γdca Γcbd )(X μ (ξ))
with naturally

1 ar ∂ ∂ ∂
Γabc = h hrc + c hrb − r hbc
2 ∂ξ b ∂ξ ∂ξ
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 85

The Formalism of String Functional Integrals for the Evaluation 85

must satisfy the topological constraint that originates from the Gauss the-
orem. For the case of boundaryless compact surfaces without “handles”
1 √
( h R)(X μ (ξ))d2 ξ = 2 − 2g, (4.2)
2π Ω

where g denotes the number of “holes” of Sc .

Note again that the domain Ω must be compatible with the surface
topology ([4]).
The most important property of the above written functionals on the
Riemman surface Sc {(we point out that under the above mathematical
imposed conditions on the surface Sc added with an additional assumption
of C ∞ -differentiability, it is a mathematical consequence that Sc may be
endowed now with a complex structure, turning it a Riemman surface,
paramtrized by holomorphic vector fields on Ω ⊂ R2 ; X μ (ξ1 , ξ2 ) = X μ (z)
√
with z = ξ1 + iξ2 , i = −1)} is the local invariance under the group of
local diffeomorphism of the surface S
ξa − ξa = δξa =: Ea (ξ)
(4.3)
δXμ (ξ) = εa (ξ)∂ a Xμ (ξ).
It is worth that the above pointed out invariance under the diffeomor-
phism local group of the surface Sc , can be extended to the global diffeomor-
phism case, only for trivial topological surfaces with g = 0 (no “handles”)
(see eq. (4.2)).
Let us thus follow R.P. Feynman in his theory of path integration sum
over “classical-random” histories of a quantum system (with a classical me-
chanical system counter-part as in our case: The Correspondence Principle
of Quantum Mechanics in action), in order to quantize our string theory
(the curve C can be considered as our classical string and the surface Sc
denotes its euclidean quantum trajectory in the space-time RD ), through
an anihillation string process).
∞
$  
 1
G(C) = dh μ[Xμ (ξ)] × exp − A(Sc )
g=0 Xμ (ξ)|ξ∈∂Ω =C μ (σ) 2πα
  
1 √
×δ (F )
Kh (σ)dσ + ( hR)(X (ξ)) − (2 − 2g)
μ
.
C 2π
(4.4)
Here α denotes the Regge slope parameter which has the dimension of
inverse of mass square (in universal units  = c = 1). Here Kh (σ) is the
geodsica curvature of the surfaces boundary C.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 86

86 Lecture Notes in Topics in Path Integrals

The above written non-polinomial 2D quantum field theory is somewhat

complicated in its methematical perturbative calculational structure, since
dμh (X μ (ξ)) is not the usual Feynman product measure, but the “weighted”
product Feynman measure ([chapter 2]), in order to preserve the invariance
of the geometrodynamical string propagator eq. (4.4) under the action of
the main theory’s symmetry eq. (4.3). Explicitly
 1/4
dh μ[X μ (ξ)] := [ h(X(ξ)) dXμ (ξ)]. (4.5)
ξ∈Ω

The above written local diffeomorphism invariant functional measure

on the string vector position is obtained form the local diffeomorphism
invariant functional Riemann metric

||δXμ ||2 = (h(X μ (ξ))1/2 (δXμ )(ξ)(δXμ )(ξ)d2 ξ. (4.6)

Ω

However the first step to evaluate the so called Nambu-Goto string

path integral eq. (4.4) is to consider the quantization process as “quantum
fluctuations” around the classical system motion (R.P. Feynman):
/√ 0
Xμ (ξ) = XμCL (ξ) + πα X μ (ξ) ( = 1). (4.7)

The classical dynamics is given by Euoer-Lagrange equations associated

to the surface area functional under the topological constraint eq. (4.2).
And before proceeding, we remark that is only in this step and on the
topological form of Ω where the topological constraint is taken into account.
As a consequence we can disregard explicitly the functional topological
constraint on eq. (4.4). We have thus, the classical motion equations (a
Dirichlet nonlinear problem)
Δh XμCL (ξ) = 0
(4.8)
XμCL (ξ)|ξ∈∂Ω = C μ .
Here Δh denotes the second-order elliptic operator called Laplace-
Beltrami associated to the metric

hab (ξ) = hab (Xμ (ξ)) = (∂a X μ ∂b Xμ )(ξ).

1 √
Δh = √ ∂a ( h hab ∂b ) (ξ). (4.9)
h
At this point, one most use formally, at least for trivial topological sur-
faces Sc , the hypothesis of the global extension of the local diffeomorphism
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 87

The Formalism of String Functional Integrals for the Evaluation 87

group of the surface Sc , in order to use globally on Sc the conformal coor-

dinate system
ξ1 = ξ1 (ξ1 , ξ2 ); ξ2 = ξ2 (ξ1 , ξ2 ).
 
hab (X μ (ξ  )) = ρ(X μ (ξ  ))δab . (4.10)
The new coordinates are given by the global (or local) diffeomorphic
√
solutions of the Laplace-Beltrami: equations given below (i = −1)
/6
dξ1 + idξ2 = λ(ξ1 , ξ2 ) h11 (Xμ (ξ)) dξ1
4 0
(h12 (Xμ (ξ)) + i h(Xμ (ξ)))
+ 4 dξ2 (4.11-a)
h11 (X μ (ξ))
/6
dξ1 − idξ2 = μ(ξ1 , ξ2 ) h11 (Xμ (ξ)) dξ1
/ (h (X (ξ)) − i4h(X (ξ))) 0 0
12 μ μ
+ 4 dξ2 . (4.11-b)
h11 (X μ (ξ))
In this new coordinate system on Sc ([4])
1
hab (Xμ (ξ  )) = ρ2 (ξ  )δab = δab . (4.12)
|λμ|(ξ, ξ  )
If one now choose the conformal gauge for the surface Sc hab (ξ) =
ϕ(ξ)
e δab (ϕ(ξ) = 2 ln ρ(ξ)), one reduces the string non Linear elliptic problem
eq. (4.8) to the well-studied Dirichlet problem in Ω
$
Δh=δab Xμ (ξ  ) = 0
 (4.13)
Xμ (ξ  )|ξ ∈∂Ω = C μ .
Note that the problem full solution is given by
Xμ (ξ) = Xμ (ξ  (ξ)). (4.14)
The solution of eq. (4.13) can be always be analyzed by methods of
conformal complex variable methods ([4], [5]) specially for the trivial topo-
logical case (connected planar Ω ) ([4]).
Unfortunatelly, it appears that the resulting quantum theory (path in-
tegral) for eq. (4.4)–eq. (4.5) still remains as an open problem, from a
purely perturbative approach around the loop-expansions in an α – power
expansion.
However in the simply case of the domain Ω(R,T ) being a rectangle of
sizes 0 ≤ ξ1 ≤ R; 0 ≤ ξ2 ≤ T , an exact one-loop result can be obtained
(Note that C = ∂Ω).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 88

88 Lecture Notes in Topics in Path Integrals

It is straightforward to see that

μ 1
S[X (ξ1 , ξ2 )] = (RT )
2πα
1 T R
μ 
+ dξ2 dξ1 (∂X (ξ))2 + O(α 2 ). (4.15-a)
2 0 0

μ μ
dh [X (ξ1 , ξ2 )] = DF [X (ξ1 , ξ2 )] + O(α ). (4.15-b)
This leads to the result
 
1
G(C[R,T ] ) = exp − RT
2πα
−( (D−2)
2 )
× det (−Δ) . (4.16)
Ω(R,T )

By noting the explicit (non-trivial) evaluation of the functional deter-

minant of the Laplacean on the Torus Ω(R,T ) with Dirichlet conditions ([7],
vol II):
−1/2
det (−Δ)
Ω(R,T )

1 ; / − 4πT 0−1/24
= T 1/2
e R
(R )
∞

(1 − e− )−2
2πnT
× R , (4.17)
n=1

one can see that the quantum strong ground state has the “confining be-
havior” with the Coulomb-Lscher term as its energy on this one-loop ap-
proximation
 
1 R π(D − 2) 1
EVacuum (R) = lim − ln G(C(R,T ) ) = − · .
T →+∞ T 2πα 6 R
(4.18)
Unfortunatelly ths string scattering amplitudes were never evaluated
in a undisputable form in this Nambu-Goto strong theory, unless on the
light-cone gauge by S. Mandelstam ([3], [6]).
As a consequence of the above mathematical aspects on the Nambu-
Goto quantum string theory, A.M. Polyakov has proposed a new functional
integral approach to overcame some of the above difficulties.
The complete mathematical exposition of the A.M. Polyakov propose
will be exposed (in details) on next section (see also chapter 2).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 89

The Formalism of String Functional Integrals for the Evaluation 89

As a final comment let us try to evaluate the Nambu-Goto string path

integral through the use of the conformal gauge as given by eq. (4.12). In
this case (the so called light – cone gauge), we have the following con-
straints, after introducing the complex-light-cone euclidean coordinate on
the domain Ω
(∂ξ1 − i∂ξ2 )
∂+ = = ∂z
2 (4.19)
(∂ξ1 + i∂ξ2 )
∂− = = ∂z̄ .
2
We have that
∂+ Xμ ∂+ Xμ = ∂− Xμ ∂− Xμ (⇔ ∂ξ1 Xμ ∂ξ2 Xμ = h12 = h21 = 0) (4.20-a)
6
h(Xμ (ξ  )) = ∂+ Xμ ∂− Xμ (⇔ h11 = h22 = ∂+ Xμ ∂− Xμ ). (4.20-b)
In this gauge the path integral eq. (4.4) for C = {φ} (the string partition
functional) takes the form (for the simple case of trivial topology surface
g = 0).
> 1 dz ∧ dz̄
Z = dh μ[Xμ (z)] exp −
2πα Ω 2i
 
?
× (∂z X μ )(∂z̄ X μ )(z, z̄)
× δ (F ) [(∂z Xμ ∂z Xμ )] × δ (F ) [(∂z̄ Xμ ∂z̄ Xμ )]. (4.21)
Note that in the practical use of eq. (4.1) to evaluate string observable
average, one already uses the observable on the light-cone gauge eq. (4.20),
which by its turn suppress the explicitly use of the above written delta
functionals insuring that Feynman-Wiener measure dh μ[Xμ (ξ)] is already
in this gauge.
It has been proved by my self ([1]) that the non-linear measure
dh μ[Xμ (ξ)] on the light-cone gauge can be related by the A.M. Polyakov-
Stromingh-Fugikawa conformal anomaly factor to the Feynman-Wiener
simply weighted measure
⎛ ⎞

dh μ[Xμ (ξ)] = ⎝ dXμ (z, z  )⎠

(μ,z,z̄)
 (e)

(0)δ
× ln(h(Xμ (z, z)))dz dz̄
4 Ω
   
(26 − D) dzdz̄ ∂+ (∂+ Xμ ∂− Xμ ) · ∂− (∂+ Xμ ∂− Xμ )
× exp − (z, z) .
48π Ω 2 (∂+ Xμ ∂− Xμ )2
(4.22)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 90

90 Lecture Notes in Topics in Path Integrals

One can see thus that only at D = 26, the pure bosonic Nambu-Goto
string is described by Marsless scalar fields on the string domain Ω, if One
uses formally the Bollini-Giambiagi dimensional regularization scheme to
assign the unity value for the tad-poles Feynman diagramms1 ([1], [4]),
through a not completely understood Feynman diaggrammatics for two-
dimensional (mathematically ill defined) Massless scalar fields for Ω = R2
(the so called Coleman theorem ([1])).
However, One can follows the Virassoro-Sakita proposal to evaluate
string scattering amplitudes using scalar vertex without bothering ourselve
with gauge fixed technical details ([3]).
We thus use the path integral eq. (4.21) for Ω = R2 in order to evaluate
the closed string (scalar) N -point scattering amplitude at D = 26 (with
δ (2) (0) = 0)
 
1> 1
A(ρμ1 , . . . , ρμN ) = DF [Xμ (ξ)] exp −  (∂Xμ · ∂Xμ )(ξ)d2 ξ
Z πα R2
⎡ ⎤
N ?
×⎣ d2 ξj exp(iρjμ · Xμ )(ξj )⎦ . (4.23)
R2N j=1

Since eq. (4.23) is formally a Gaussian Functional Integer and we can

√
re-write the scalar vertexs as string vector position source (i = −1)
⎛ ⎞
N
⎝ exp(iρjμ Xμ )(ξj )⎠
j=1
⎧ ⎡ ⎤⎫
⎨  N ⎬
= exp i ⎣ (ρjμ δ (2) (ξ − ξj ))Xμ (ξ)d2 ξ ⎦ , (4.24)
⎩ R2 ⎭
j=1
and by using the dimensional regularization technique to vanish the tad-
pole term
⎧ ⎫
⎨  N ⎬
exp − (ρjμ )2 (−Δ)−1 (ξj , ξj ) = 1. (4.25)
⎩ ⎭
j=1

One gets the Veneziana N -point amplitude as a result in R26 ([3]. [6])
A(ρμ1 , . . . , ρμN ) = d2 ξ1 . . . d2 ξN
R2N
N μ μ
(ρ ,ρ )
i j
× |ξi − ξj | πα (4.26)
i<j
1 If eikξ
Δ(ξ) = dD k · |K|β
with β ∈ R, then on the dimensional regularization scheme
Δ(0) = dD k|K|−β = 0, even if β = 0.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 91

The Formalism of String Functional Integrals for the Evaluation 91

4.3 The Nambu-Goto Extrinsic Path String

Another important classical local diffeomorphism invariant surface func-

tional, closely related to the Q.C.D. (SU (∞)) string ([1])l, is the so-called
extrinsic functional which is defined by the square of the surface mean
curvature (see eq. (4.9)), with γ ∈ R
√
Fexta (S) = γ 2 [ h (Δh Xμ )2 ](ξ)d2 ξ. (4.27)
Ω

Let us add suh functional to the area functional eq. (4.1) and consider
the associated string path integral propagator eq. (4.4) (in the trivial topo-
logical secta of surface Sc ).

G(C) = dh μ[Xμ (ξ)]

Xμ (ξ)|ξ∈∂Ω =C μ (σ)
 
1
× exp − 1/2 2
(h(Xμ (ξ))) d ξ
2πα Ω
 
√
× exp −γ 2 2 2
[ h(Δh Xμ ) ](ξ)d ξ . (4.28)
Ω

Let us note that even by using the light-cone string coordinate system,
One still has a non-polinomial interacting quantum two-dimensional SO(D)
scalar field theory.
Namelly

G(C) = dh μ[Xμ (z, z̄)]

 
1 1  
× exp −  dzdz̄ (∂+ X μ )(∂− X μ ) (z, z̄)
πα Ω 2
$ % 
&'
2
(∂+ X μ )(∂− 2
X  μ)
× exp −γ 2
dzdz̄
Ω (∂+ X  μ )(∂− X  μ )
 
× δ (F ) [(∂+ X μ
· ∂+ X μ )]
 
× δ (F ) [(∂− X μ
· ∂− X μ )]. (4.29)

However one can evaluate formally scattering amplitudes as done pre-

viously (see eq. (4.23)) by considering the string quantum trajectory as a
small perturbation around the flat metric. The outcome is the following
one-loop string scalar scattering amplitude compare with the Nambu-Goto
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 92

92 Lecture Notes in Topics in Path Integrals

case of the previous section eq. (4.23)


1
A(ρμ1 , . . . , ρμN ) = DF [Xμ (ξ)]
Z
 
1 1
× exp −  (∂Xμ ∂Xμ )(ξ)d2 ξ
πα R2 2
 
× exp −γ 2 (∂ 2 Xμ )(∂ 2 Xμ )(ξ)d2 ξ
R2
⎡ ⎛ ⎞⎤⎫
N ⎬
×⎣ d2 ξj ⎝ exp(iρjμ Xμ )(ξj )⎠⎦ . (4.30)
R2N ⎭
j=1

Since eq. (4.30) is a Gaussian Functional Integral, One obtains an ex-

actly result, where we have used the result
 −1
1 2 2 2
(−∂ ) + γ(∂ ) (ξ, ξ1)
(πα )
%   &
1/2
 1  1 1 
:= πα − ln |ξ − ξ | − K0 |z − z | .
2π 2π πα γ 2
(4.31)

Namelly:

A(ρμ1 , . . . , ρμN )

= d2 ξ1 . . . d2 ξN
R2N
⎛ ⎞
N μ μ
(ρ ,ρ )
×⎝ ⎠
i j
(ξi − ξj ) πα

i<j
⎛ ⎡  ⎤⎞
 
N 1/2
α 1
× ⎝exp ⎣− K0 |ξi − ξj | ⎦⎠ . (4.32)
2 i<j
πα γ 2

Unfortunatelly, it appears that the corrections coming from the string

extrinsic functions do not modify the usual Veneziano-Virassoro bosonic
closed string spectrum. But a clear proof of this no go result is still missing.
At this point we analyze the vacuum string energy for the domain ΩR,T )
being a rectangle of sizes 0 ≤ ξ1 ≤ R; 0 ≤ ξ2 ≤ T through a one-loop ap-
proximation inthe Regge slope constant α (see eq. (4.15-a) of the previous
section).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 93

The Formalism of String Functional Integrals for the Evaluation 93

In this case we have the exact result

 
1
G[C(R,T ) ] = exp − RT
2πα
 − (D−2)
1 2
× det γ 2 (−∂ 2 ) + (−∂ 2 ) . (4.33)
Ω(R,T ) γ2

The evaluation of the fourth-order elliptic operator (with Dirichlet cn-

ditions) can be accomplished through the result

 − (D−2)
2 2 1 2
det γ (−∂ ) + (−∂ 2 )
Ω(R,T ) γ2
%
− (D−2)
2
= det [(−∂ 2 )]
Ω(R,T )
   &
− (D−2)
2 1
× det 2
+ (−∂ ) . (4.34)
Ω(R,T ) γ2

But functional determinants have been evaluated in the literature ([7]).

(See eq. ([7])). The second order massive operator has the following result


− (D−2)

2
2 1
det (−∂ ) + 2
γ
> (D − 2)πT / 1 R
= exp −
R 6 γ 2 2π
4πe−γ̂
2
R
+ ln
2πγ 2 R

2 − 2 0?
3
4 1 ∞
1 R R
+ dx(1 − x) (n2 + x
2 2πγ 2 0 n=1
2πγ 2
⎧ ⎡⎡ ⎛⎛ G ⎞⎞⎤⎤⎫−(D−2)
⎨ +∞ 2 ⎬
× ⎣⎣1 − exp ⎝⎝− 2πT n2 + R ⎠⎠⎦⎦ .
⎩ R 2πγ 2 ⎭
n=−∞

(4.35)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 94

94 Lecture Notes in Topics in Path Integrals

/ 0
R
Let us remark that t = 2πγ 2 (appendix B)

⎡ ⎤
1 ⎢ ⎥
⎢ 1 ⎥
dx(1 − x) ⎢ ⎥
0 ⎣ / 02 3/2 ⎦
R
n2 + x 2πγ 2

2 1 2n2
=− +
t2 (n + t2 )1/2
2 t4 (n2 + t2 )1/2
2 2 2
+ − (n + t2 )1/2 . (4.36)
t2 n t4

which unfortunatelly leads to an – in principle – divergence on the string

path integral G(CR,T ) for the extrinsic string. However it is possible to
get a somewhat phenomenological finite result for the vacuum energy (the
interquark potential) for large R (R → ∞)

 
1
EVac (R) = lim − ln G(C(R,T ) ) . (4.37)
T →∞ T

The point is that for large R (R → ∞), the Epstein sum below written
has a finite asymptotic behavior (see appendix A for a detailed evaluation),
s∈R

∞
 1
Sepst (s, a2 ) :=
n=1
(n2 + a2 )s
⎛ ⎡ ⎤⎞
1 ⎜ ∞
2 ⎢
/ d2
0⎥⎟
= ⎜ duU s−1 · e−U(λt ) ⎢ ⎧ ⎫ eU(− dz 2 ) ⎥⎟
Γ(s) ⎝ ⎣T r⎨ C 2 ([0, 1]) ⎬ ⎦⎠
0
⎩f (0) = f (1) = 0⎭

1 ; 
(a2 )−(S− 2 +1) .
a→∞ 3
∼ (4.38)
4πΓ(s)

As a consequence of the finite behavior eq. (4.38), One has the asymp-
totic (distributional L. Schwartz sense)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 95

The Formalism of String Functional Integrals for the Evaluation 95

∞
 1 R→∞ 1
/ 02 ∼ / 02 . (4.39)
R R
n=1 n2 + x 2πγ 2
2(π)3/2 x 2πγ 2

After substituting this (formal) result on eq. (4.35), by taking into

account the (regularized) integration:

⎧ ⎡ ⎛ ⎞⎤⎫
⎪
⎨1 ∞ ⎪
⎬
⎜
4 1
R ⎢ 1 ⎟⎥
lim ⎣ dx(1 − x) ⎝ / 02 ⎠⎦
R→∞ ⎪
⎩2 2πγ 2 0 R ⎪
⎭
n=1 (n2 + x )3/2
2πγ 2

⎡ ⎤
4 1
1 R ⎢ 1 ⎥
= ⎣ dx(1 − x) / 02 ⎦
2 2πγ 2 0 R
2π 3/2 x 2πγ 2

/ 02 
R
2πγ 2
1
dx(1 − x)
∼
4π 3/2 εQCD x
/ 02
R
2πγ 2
= [−y(εQCD ) − 1]
4π 3/2

 
(y(εQCD ) + 1)
=− R2 . (4.40)
16π 5/2 γ 2

Here εQCD is the underlying cut-off on the “Feynman” parameters x

(εQCD ≤ x ≤ 1). It is worth remark that the regularization parameters will
be absorbed in the bare extrinsic coupling constant γ 2 (somewhat related
to the Q.C.D(SU (∞)) coupling constant γ 2 = (g∞ )2 (0|F 2 |0SU(∞) ) (see
chapter 1). Namelly:

1 1
2 (ε) = 2 ((ln εQCD ) + 1).
γbare γbare

The contribution of this term to the phenomenological “interquark

potential” on the extrinsic string theory is exactly given below (to be added
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 96

96 Lecture Notes in Topics in Path Integrals

to the pure Nambu-Goto term eq. (4.18).

(D − 2) π (D − 2) 1
V extrinsic(R) = − · + · 2
6 R 2 γren

(D − 2) 1
− · 4 R · ln(4πe−γ̂ )
4π γren

(D − 2) 1
+ 4
(R ln R)
4π γren

(D − 2) 1
+ 2
·R . (4.41)
16π 5/2 γren
We think that eq. (4.41) is an important result on applications of string
representations for QCD, since it leads to a growing force that goes to
infinite for R → ∞, confining the static color quarks charge, certainly
a real “quark confinement”, opposite to the pure Nambu-Goto case with
a “weak confinement” by a constant force. quarks are really confined if
a string for QCD holds true at the limit of large R. So a QCD point
particle description should only be phenomenological.
We comment also that such hind of behavior for the interquark potential
is compatible with a Mandelstan Gluonic propagator of the form ([8])
1
Dm (x − y) = dD p · eiρ(x−y) [(y(|p|2 ))/|p|4 ] . (4.42)
(2π)D

4.4 Studies on the perturbative evaluation of closed Scat-

tering Amplitude in a Higher order Polyakov’s Bosonic
String Model

It is well known that Polyakov covariant string theory with exactly soluble
Liouville two dimensional model (the famous non critical string) has in its
protocol the main hope of suppressing the tachionic excitation of the usual
Nambu-Goto string ([10], [11], [12])2 .
By the other side it is less known that Polyakov-Liouville effective
string theory is somewhat phenomenological in the sense that it has al-
ready built in the assumption of “weak” two dimensional induced quantum
gravity, since one replaces its covariant path integral measure DF [eϕ/2 ] ≡
2 For a detailed presentation of the Polyakov’s string path integral see §3 of [9].
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 97

The Formalism of String Functional Integrals for the Evaluation 97

B
d(eϕ/2 (ξ)) by the usual Feynman path measure DF [eϕ/2(ξ) ] ∼ DF [ϕ(ξ)]
ξ∈D
when considering the final effective Liouville-Polyakov field theory on string
world-sheet.
So in this approximate scheme of non critical strings (but quite useful
([2])), we in this section (somewhat pedagogical) introduce a somewhat toy
model of higher-order two dimensional Polyakov covariant string with im-
proved ultra violet behavior and show the exactly solubility of the resulting
covariant Polyakov path integrations.
Let us start our analysis by considering the theory’s N -point off-shell
closed Scattering amplitude defined by the following Polyakov’s bosonic
string general path integral

μ2  √
A(P1μ , . . . , PNμ ) = dcov μ[gab (ξ)]e− 2 R2
( g)(ξ)d2 ξ

⎡ ⎤
+ ,- .
=newproposed term

⎢ γ 4 4  ⎥
⎢ − 2 2    ⎥
⎢ 2 R2 d ξd ξ g(ξ) g(ξ )R(ξ)Kg (ξ, ξ )R(ξ ) ⎥
× ⎢e ⎥
⎢ ⎥
⎣ ⎦

  
1 √
× dcov μ[X μ (ξ)] exp − ( g g ab ∂a X μ ∂b X μ )(ξ)d2 ξ
2 R2

$ '
N ;4 
× g(ξi ) · exp(Piμ Xμ (ξi )) d2 ξi . (4.43)
R2N i=1

Here the covariant functional measures are the functional volume ele-
ments associated to the functional Riemann metrics ([1], [2]. [3])
;√    

||δgab ||2 = g((δgab )(g aa g bb + cg ab g a b )(δga b )) d2 ξ
R2

√
||δX μ ||2 = [ g δX μ δX μ ](ξ)d2 ξ. (4.44)
R2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 98

98 Lecture Notes in Topics in Path Integrals

The Green function Kg (ξ, ξ  ) of the Laplace-Beltrami operator on the

presence of the metric gab (ξ) = eϕ(ξ) δab , in the conformal gauge, with a
covariant cut-off already built in is given explicitly by the Riesz-Hadamand
formula ([4], [5]) on R2

⎧
⎪
⎪
1
− y|ξ − ξ  | ξ = ξ
⎪
⎪
⎨ 2π
Kϕ (ξ, ξ  ) = (4.45)
⎪
⎪
⎪
⎪ 1 1 1 1
⎩ − ϕ(ξ) − y(ε) ξ = ξ 
2π ε 2π 2π

Note that eq. (4.3) is a direct result that in the conformal gauge
gab = eϕ(ξ) δab the Laplace-Beltrami operator reduces to the usual Laplace
operator for ξ = ξ  . And for those points ξ = ξ  , one should use the fa-
mous parameter formula of J. Hadamard for the Laplace Beltrami Operator
on R2 togheter by taking into account the fact that the geodesic distance
S(ξ, ξ  ) in R2 , endowed with a metric in the conformal gauge is exactly
given by S(ξ, ξ  ) = eϕ(ξ) |ξ − ξ  | for ξ → ξ  .
In other words, for ξ → ξ  , we have the regularized asymptotic behavior
an RN ([5])

$ '
 Γ( 12 N ) (2−N )g(S(ξ,ξ  ))
K(ξ, ξ ) ∼ lim N e
N →2 2π ( 2 ) (N − 2)

N →2
1 1 1 / 0
ε→0
∼ − g eϕ(ξ) (|ξ − ξ  |2 + ε2 )1/2
2π N −2 2π

|ξ−ξ  |→0 1 ϕ(ξ) 1

∼ − − y(ε). (4.46)
2πε 2π 2π

In order to evaluate perturbativelly eq. (4.1) around the flat metric

gab (ξ) ∼ δab , we consider the following approximation on the resulting
effective functional measure on the metric field by replacing it by its weak
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 99

The Formalism of String Functional Integrals for the Evaluation 99

metric feynman measure

ϕ
dμ[gab ] = DF [e 2 (ξ) ]
$ '
26 1
× exp − 2 2
(∂ϕ) (ξ)d ξ
48 R2+ 2
$ '
μ2
× exp − e ϕ(ξ) 2
d ξ
2 R2+
 2 
γ
× exp − (∂ 2 ϕ)2 (ξ)d2 ξ
2 R2
 
(replace by) 26 1
∼ DF [ϕ(ξ)] exp − (∂ϕ)2 d2 ξ
48 R2 2
 2 
γ
× exp − (∂ 2 ϕ)2 (ξ)d2 ξ
2 R2
 
μ2
× exp − eϕ(ξ) d2 ξ . (4.47)
2 R2
Note that we have further replaced the fourth-order (non-local) 2D grav-
ity term by the simple local result exposed below
R(ξ)
> γ 
/+ ,- .0
exp − d2 ξd2 ξ  eϕ(ξ) eϕ(ξ ) e−ϕ(ξ) ∂ 2 ϕ(ξ)
2
δ (2) (ξ − ξ  ) −ϕ(ξ ) 2 ?
× ( ∂ ϕ(ξ))
eϕ(ξ)
 
γ 2 2  2 δ (2) (ξ − ξ  ) 2 
= exp − d ξd ξ (∂ ϕ(ξ)) (∂ ϕ)(ξ )
2 eϕ(ξ)
 
γ
∼ exp − d2 ξ(∂ 2 ϕ)2 (ξ) . (4.48)
2
Here we have substituted the covariant Laplace Beltrami operator Green
Function by its mathematical distributional limit δcov (ξ −ξ  ), but for a back
(2)

ground “weak metric” gab (ξ) = δab .

In other words:
δ (2) (ξ − ξ  ) δ (2) (ξ − ξ  )
∼ ∼ δ (2) (ξ − ξ  ).

eϕ(ξ ) 1 + ϕ(ξ  ) + . . .
It is worth to call the reader attention that
> 2we have preserved
? in its

integrity the Liouville cosmological term exp − μ2 R2 eϕ(ξ) d2 ξ , since the
main interest in this section is to present perturbative calculations on the
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 100

100 Lecture Notes in Topics in Path Integrals

N -point off-shell closed scattering amplitude that show us that such im-
portant term does not affects its pole singularities of those usual closed
string scattering amplitudes Kobo-Milsen-Virasoro result for μ2 = 0; even
in presence of metric fourth-order term as proposed by ours.
After collecting all the above results and remarks, we are lead to evalu-
ate perturbativelly on the Liouville-Polyakov cosmological constant μ2 , the
following standard Liouville-Polyakov path integral
 
(26 − d) 1
Ã(P1 , . . . , PN ) = D [ϕ(ξ)] exp −
F
(∂ϕ)2 (ξ)d2 ξ
48π R2 2
 
μ2
× exp − e ϕ(ξ) 2
d ξ
2 R2
 2 
γ
× exp − 2 2
(∂ ϕ) (ξ)d ξ 2
2 R2
$% N &% N &'
μ
·Xμ (ξi ))
× eϕ(ξi ) ei(Pi . (4.49)
i=1 i=1

By taking into account

/@ the momentum 0 conservation on the scattering of
N
the string excitations i=1 (Pμ
i 2
) ≡ 0 , One obtains the following Liouville
path integral without the ε-covariant cut-off
/ ;
Ã(P1 , . . . , PN ) = d2 ξ1 . . . d2 ξN exp − (Pμi · Pμj )
R2N i,j=1
(i=j)

1 0
× − ln |ξi − ξj |2
4π
  
(26 − D) 1
× D [ϕ(ξ)] exp −
F 2
(∂ϕ) (ξ)d ξ2
48π R2 2
 2   
γ μ2
× exp − (∂ ϕ) (ξ)d ξ exp −
2 2 2
e ϕ(ξ) 2
d ξ
2 R2 2 R2
 % N N & '
  ϕ(ξi )
× exp ϕ(ξ1 ) − (Pμi )2 . (4.50)
i=1 i=1
4π

It is worth to point out that the zeroth order 2D quantum metric

gab (ξ) = δab (ϕ(ξ) = 0), the Liouville path integral eq. (4.8) produces a
scattering amplitude structurally similar to the famous Veneziane-Koba-
Nielsen-Virasoro closed scattering amplitude (in the Euclidean space-time
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 101

The Formalism of String Functional Integrals for the Evaluation 101

RN )

Ã(0) (P1 , . . . , PN )
⎛ ⎡ ⎤⎞
⎜ ⎢ i 1 ⎥⎟
= d2 ξ1 . . . d2 ξN ⎜ ⎢
⎝exp − ⎣ (Pμ · Pμj ) − ln |ξi − ξj |2 ⎥
⎦⎠
⎟
R2N i,j=1
4π
i=j
⎛ ⎞
 
⎜ i ·P j )
(Pμ ⎟
d2 ξ1 . . . d2 ξN ⎜ ⎟.
μ
= ⎝ |ξi − ξj | 2π
⎠ (4.51)
R2N (i,j=1)
i=j

The key point of our study is the exactly path-integral evaluation of the
fourth-order improved Liouville path integral in a perturbative expansion
in the Polyakov’s cosmological constant μ2 . Namelly
 
(26 − D) 1
FLiouville (Pμi ) ≡ DF [ϕ(ξ)] exp − (∂ϕ)2 (ξ)d2 ξ
48π R2 2
 2 
γ
× exp − (∂ 2 ϕ)2 (ξ)d2 ξ
2 R2
$∞ '
 μ2
n
× − e ϕ(z1 )
...e ϕ(zN ) 2 2
d z1 . . . d zN
n=0
2 R2N
 %N &
 (Pμi )
× exp ϕ(ξi )(1 − ) . (4.52)
i=1
4π

Since the above written path integral is a fourth-order Gaussian func-

tional integral, with correlation function given explicitly by
 −1
26 − D
(−∂ 2 ) + γ 2 (∂ 2 )2 (ξ, ξ  ) ≡ L−1 (ξ.ξ  )
48π
%  G &
48π 1  1 26 − d
≡ − ln |ξ − ξ | − K0 |ξ − ξ  | , (4.53)
26 − D 2π 2π 48πγ 2

one obtains the result below, with the path integral average Notation,
 

1 − 12 ϕ(ξ)Lξ ϕ(ξ)dR ξ
O(ϕ)ϕ = F
D [ϕ]e O(ϕ) . (4.54)
2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 102

102 Lecture Notes in Topics in Path Integrals

Namely: (for each perturbative order M in μ2 )

EN M  i )2
(Pμ
F
(M,N ) ϕ(ξi ) 1− 4π
FLiouville (Pμi ) = <e ϕ(z1 )
...e ϕ(zn )
×e
j=1 i=1 R2N
ϕ
⎧⎛ %  &⎞⎫
⎨ N
1 (Pμt )2 (P s )2 ⎬
= ⎝ exp + 1− 1− L−1 (ξt , ξs ) ⎠
⎩ 2 4π 4π ⎭
(t,s=1)
⎛ ⎞
M  
1
×⎝ exp + L−1 (zh , zq ) ⎠
2
(h,q)
N M %  &
1 (Pμq )2
× exp + 1− L−1 (zq , zt ) . (4.55)
t=1 q=1
2 4π

We remark now that the singularities type pole (paticle string excita-
tion that would modify the zeroth-order Veneziano-Kobe-Nielsen-Virasoro
closed bosonic string dual model will come from the “pinch” ultra-violet
singularities ξt → ξs ! However from the ultra-violet behavior below de-
picted
G
1 26 − D
lim K0 |ξt − ξs |
ξt →ξs 2π 48πγ 2
% G &
1 1 26 − D ψ(1)
= lim  + ln |ξt − ξs | + ln − , (4.56)
ξt →ξs 2π 2π 96πγ 2 2π

one can see that there is the cancealling of the logaritmic terms on eq. (4.12)
for the “pinch” points protocol to analyze poles on string scattering am-
plitudes.
As a result, at each perturbative order M , these terms are expected to
not contribute to the bosonic closed string spectrum eq. (4.10).
That is the main conclusion of our section: the Polyakov proposal of
a quantum bosonic string as a theory of D (ill defined!) two-dimensional
massless scalar fields interacting with two-dimensional induced Liouville-
Polyakov quantum gravity, (expected to be computationally effective for
“weak” two-dimensional quantum gravity), still does not alter the usual
(tachionic) old closed bosonic string of the Veneziano-Virasoro dual model,
even with a improved ultra-violet behavior as proposed by ours.
It appears thus that the introduction of induced quantum gravity in
the Polyakov proposal, could not alter the usual tackionic spectrum of the
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 103

The Formalism of String Functional Integrals for the Evaluation 103

bosonic string, even if one considers a somewhat “higher order” Polyakov-

Liouville induced 2D quantum gravity. As a result, one is naturally lead
to consider further fermionic (intrinsic, extrinsic or supersymetric) degrees
of freedom on the string world sheet ([1], [6]), in oder to have candidate to
be physically sensible (see chapter 3).

4.5 Appendix A: The distributional limit of the Epstein

function

Let us try to evaluate (in some yet undiscovered asymptotic distributional

theory) the limit of large a of the so called Epstein function for S ∈ R

∞
 1
Sepstein (s, a2 ) =
μ=1
(μ2 + a2 )s

∞
1
U s−1 e−a U 2 −Un2
2
= e dU
Γ(s) 0
⎛ ⎞
 
1 ⎜ ⎜
∞
−U −d 2 ⎟
⎟.
U s−1 · e−a
2
U2
= · T r⎧ ⎫ e dξ 
Γ(s) ⎝ 0
⎨ C 2
([0, 1])⎬ ⎠
⎩ Dirichlet ⎭

(A-1)

For a → ∞,k certainly U → 0 on the U -integrand, and the Seeley

d2
asymptotic expansion for the second-order operator − dz 2 holds true.

Theta is:

 ∞ 
a2 →∞ 1 1
dU · U s−1 e−Ua
2
Seps (s, a2 ) ∼
Γ(s) 0 4πU 1/2
 ∞ 
a2 →∞ 1
dU · U s− 2 e−Ua
3 2
∼
4πΓ(s) 0
 
a2 →∞ 1 Γ(s − 32 + 1)
∼
4πΓ(s) (a2 )s− 32 +1
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 104

104 Lecture Notes in Topics in Path Integrals

4.6 Appendix B: Integral Evaluation

Let us elementarly evaluate the following integral

1 1 1
1 dx x
dx(1 − x) = − dx
0 (a2 + xb)3/2 0 (a2 + xb)3/2 0 (a2 + xb)3/2
 2
 2
2 a +b a +b
(x= v−a ) 1 1
dv · v − 2 dv · v − 2
3 1
b
= −
b a2 b2 a2

a2 +b
a2
dv · v − 2
3
− 2
b a2
 
2 2a2
= (a2 + b2 )− 2
1
− + 2
b b
 
2 2a 2a 2
+ + 2 − 2 − 2 (a2 + b)1/2
ab b b b
2b + 2a2 1 2 2
= − + − (a2 b)1/2 . (B-1)
b2 (a2 + b)1/2 ab b2 +

As a consequence (μ2 = n2 )

1
1 2 1
dx(1 − x) =−
0 (μ2 + xa2 )3/2 a2 (μ2 + a2 )1/2
1 2μ2
+ 4
· 2
a (μ + a2 )1/2
2 2
+ − (μ2 + a2 )1/2 . (B-2)
a2 μ a4

4.7 Appendix C: On the perturbative evaluation of

the bosonic string closed scattering amplitude on
Polyakov’s framework

The fundamental observable on Polyakov’s bosonic string theory ([1], [2]) is

the closed scattering amplitude which is given explicitly by the covariants
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 105

The Formalism of String Functional Integrals for the Evaluation 105

2D induced quantum gravity path integral in moments space

N 
1
Ã(Pμ1 , . . . , PμN ) = d2 ξj × DF [ρ̄(ξ)]
j=1
Z
$  '
2
(26 − D) 1 ∂a ρ̄ 1 2
× exp − (ξ)d ξ − μR
2 2 2
ρ̄ (ξ)d ξ
12π R2 2 ρ̄ 2 R2
⎧ ⎫
⎨ N ⎬
× exp 2 ln(ρ̄(ξj ))
⎩ ⎭
j=1
⎧ ⎫⎞⎤
⎨  N ⎬
× exp − (Pμi · Pμj )RN K (ε) (ξi , ξj ) ⎠⎦ . (C-1)
⎩ ⎭
(i,j)=1

Here the covariant regularized Polyakov’s Green function is given ex-

plicitly by eq. (4.3) (with the identification (ρ̄(ξ))2 = eϕ(ξ) ).
After substituting eq. (4.3) into eq. (C-1), one obtain the outcome ex-
pressed now as a sigma model like perturbativelly renormalizable path in-
tegral defined by a trully Feynman measure DF [ρ̄(ξ)] ([1], [9]).
1
We have thus to evaluate in term of the D -expansion, the natural pertur-
bative parameter expansion on two-dimensional quantum gravity applied
for quantum strings with domain parameter being R2 (closed string) (see
footnote)
1/2
12π
ρ̄(ξ) = 1 + h(ξ). (C-2)
26 − D
Here we have considered our classical background metric, the flat metric
gab (ξ) = 1δab on R2 , as it should be in Einstein like gravitation theories.
At one-loop order, one arrives at the following path integral (with the
normalization factor Z = 1).
1
The full D expanded path integral is given by
$ % ∞ & '
1 
Z = D [h] exp −
F
(∂h) 2
(−1)n−1
nε n−1 n−1
h 2
(ξ)d ξ
2 R2 n=1
 
× exp −μ2 h2 (ξ)d2 ξ (C-3)
R2

with the ortoghonal constraint of the fluctuating piece h(ξ) in relation to

the flat R2 background

h(ξ) · 1 · d2 ξ = 0. (C-4)
R2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 106

106 Lecture Notes in Topics in Path Integrals

From the usual renormalization power counting / interactions

0 of the form
((∂h)2 hm )(ξ) have the renormalization index r = (D−2)
2 b+ (D−1)
2 f +δ−D
for an interaction of general for g(∂)δ φb ψ F . In our case g = ε, D = 2, f = 0,
δ = 2, b = m, which leads to the theory’s renormalizability r = 0 in R2 .
 
1 1
A(Pμ1 , . . . , PμN ) = DF [h] exp − [(∂h)2 + μ2 h2 ](ξ)d2 ξ
2 2 R2
⎧ ⎛ ⎞⎫

⎨ ( (Pμj )2 )
N
N i ·P j ) ⎬
(Pμ
d2 ξj ⎝ |ξi − ξj | 2π R ⎠
μ N
× (ε) j=1 ×
⎩ R2 j=1 ⎭
1<j
⎡   ⎤
N 1/2
12π (Pμj )2
×⎣ exp 2 1− h(ξj ) ⎦ .
j=1
26 − D 4π
(C-5)
Here we remark the use of the one-loop approximation on the object
/ 01/2
12π
ε = 26−D
⎡ ⎤ ⎡ ⎤
N j N j
(Pμ )2 (Pμ )2
⎣ (ρ̄(ξj ))2(1− 4π )⎦
=⎣ e 2(1− 4π ) ln(1+εh(ξj )) ⎦

j=1 j=1
⎧ %  &⎫
⎨ N
(Pμj )2 ⎬
one-loop order
∼ exp 2ε 1 − h(ξj ) . (C-6)
⎩ 4π ⎭
j=1

1
At this point one can see that at a perturbativelly renormalized “ D dx-
pansion” (D → −∞) ([1], [9]), the contribution of the Liouville-Polyakov’s
@N
dynamics is a multiplication factor as given below ( (Pμj )2 = 0 for a
j=1
physical elastic string excitation scattering)
⎡ ⎤
N j
P i ·Pμ
Ã(Pμ1 , . . . , PμN ) = d2 ξj ⎣ |ξi − ξj | ( μ2π )RN ⎦
R2 i<j
⎧
⎪
⎪ +,-.
2ε2
⎪
⎨ 
N
24π
exp (1 − (Pμi )2 )(1 − (Pμj )2 )
⎪
⎪ (26 − D)
⎪
⎩i<j=1

1
× K0 (μR |ξi − ξj |) . (C-7)
2π
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 107

The Formalism of String Functional Integrals for the Evaluation 107

By considering ε → 0 (D → −∞), one re-obtains the well-known A.M.

Virasoro-B. Sakita closed scattering from the old dual models.
Another useful remark is that the behavior of eq. (C-5) for ξi → ξj is
the same of the Virasoro-Sakita result, implying, thus, that by just taking
into account the Liouville-Polyakov degree of freedom on Bosonic Srings
does not alter the poles of the scattering amplitudes, so leasing to the same
drawbacks of the old theory.
We conclude that nonperturbative effects (or others intrinsic degrees of
freedom ([1])) must be taken into account to remove the tachion from the
Polyakov Bosonic string spectrum ([1]).

4.8 References

[1] Luiz C.L. Botelho, Methods of Bosonic and Fermionic Path Inte-
grals, Nova Science Publisher, Inc., 2009.
[2] A.M. Polyakov, Gauge Fields and Strings, Harwoud Academic Pub-
lisher, Chur., (1987).
[3] Michio Kako, Introduction to Superstrings, Springer Verlag, 1988.
[4] Luiz C.L. Botelho, Lecture Notes in Applied Differential Equations
of Mathematical Physics, World Scientific, 2008.
[5] M.J. Duff, Partial Differential Equations, Toronto Press, (1956).
[6] M.B. Green, J.H. Schwarz & E. Witten, Superstring theory, vol. 2,
Cambridge Monographs on Mathematical Physics, (1996).
[7] C. Itzykson & J.M. Drouffe, Statistical Field theory, vol. 1, Cam-
bridge Monographs on Mathematical Physics, (1991).
[8] Luiz C.L. Botelho, Mod. Physics Letters A, vol. 20, No. 12,
(2005).
[9] Luiz C.L. Botelho, ISRN High Energy Physics, vol. 2012, Arti-
cle ID 674985, doc. 10.5402/2012/674985. (Research Article) –
chapter 1.
[10] J. Teschner, Class. Quant. Grav., 18, (2001), R153–R222.
[11] Adel Bilal, F. Ferrari, S. Klevitson, arxi No. 1310.1951v2, [hep-th],
(1800–2013).
[12] Luiz C.L. Botelho, Gen Relative Gravity (research article), DOI
10.1007/S, 10714-012-1372-1, (May/11/2012).
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 109

Chapter 5

The D → −∞ saddle-point spectrum

analysis of the open bosonic Polyakov
string in RD × SO(N ) – The
QCD(SU (∞)) string
In this chapter we investigate the role of the chiral anomaly in de-
termining the spectrum at the saddle-point approximation D → −∞ of
the recently considered Polyakov formulation of bosonic strings moving in
RD × G with K = 2, where G is the group manifold SO(N ). The main re-
sult is, opposite to the critical dimension, that the spectrum is not sensitive
to the model chiral anomaly in the D → −∞ limit.

5.1 Introduction

There is currently strong interest in the study of strings moving in a com-

pact manifold, mainly connected with the problem of dimensional reduction
in superstring theories.
Redlich and Schnitzer have considered a formalism for quantization of
closed bosonic strings moving in the space RD × G with G being the group
manifold associated to the group U (N ) [or SO(N )]. One important feature
of these models is its characterization by a positive integer K which is
proportional to the ratio of the radius of the inner compactified dimensions
and the Regge slope parameter. The main result obtained by these above-
quoted authors was the change in the string’s critical dimension as a result
of the model chiral anomaly.
The purpose of this chapter is to take a further step in the study of these
compactified string models by analyzing the role of the chiral anomaly
in determining the string spectrum at the saddle-point limit D → −∞
(Refs. 10 and 11). As our main result we show that the spectrum for the
caqse K = 2 does not differ from the case K = 1, thus leading to the
conclusion that the chiral anomaly does not influence the string spectrum

109
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 110

110 Lecture Notes in Topics in Path Integrals

(at least at the natural limit D → −∞ on Polyakov’s framework - chapter

4).

5.2 The non-tachyonic spectrum and scalar amplitudes at

D → −∞

Let us start our analysis by considering the open-bosonic-string action for

the group RD × SO(N ) with K = 2 (cf. Ref. 8) generalized to the open-
string case: Namely,
1
S[φ(A) (ξ), eaμ (ξ), ψ i,β (ξ), Aμ (ξ)] = d2 ξe(ξ)(g μν ∂μ φ(A) ∂ν φ(A) )(ξ)
2 D
1 2
d ξe(ξ)[ψ̄ i,β eμa γa (∂μ + iAμ )ψ i,β (ξ)] + μ20 d2 ξe(ξ) + λ0 ds.
2 D D ∂D
(5.1)
The string “surface” is characterized by three fields: first, the usual
bosonic vector position field φ(A) (ξ) (A = 1, . . . , D): second, by a set of
O(N ) × O(2) two-dimensional Majorana spinors with components denoted
by {ψ1i,β (ξ), ψ2i,β (ξ)} (i = 1, . . . , N ; β = 1, 2) needed to describe the com-
pactified bosonic coordinates; and finally an Aμ − O(2) gauge field without
kinetic term. The two-dimensional parameter domain D with boundary
∂D should be taken in a consistent way with the “bosonic string surface”
topology.
The presence of the zweibein eμa (ξ) (a = 1, 2; μ = 1, 2) ensures that
the action written above is invariant under general Lorentz and coordinate
transformations. The Euclidean (curved) Hermitian γμ (ξ) matrices we are
using satisfy the relations
εμν
{γμ , γν } + (ξ) = 2gμν (ξ), γ μ (ξ)γs = i γν (ξ),
e(ξ) (5.2)
γμ (ξ) = eaμ (ξ)γa , gμν (ξ) = (eaμ ebν )(ξ)δab ,
where γa are the usual flat-space Dirac matrics.
Dual Green’s functions for a scattering of string scalar particles on RD
associated with the SO(N ) Euclidean sring theory described by the action
Eq. (5.1) can be obtained by taking variational derivatives of the generat-
ing functional [{λR } denotes the Hermitian generators of the SO(N ) Lie
algebra]
9  :
j
Γ]JR (x)] = exp − d ξe(ξ)JR [φ (ξ)] × [ψ1 (ξ)(λR )ij ψ̄2 (ξ)] , (5.3)
2 (A) i
D
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 111

The D → −∞ saddle-point spectrum analysis of the open bosonic Polyakov 111

where we have suppressed the O(2) indices and consider the fermionic fields
as complex fields. The average   is defined as in Ref. 8 or chapter 2,
but with the appropriate boundary conditions of Refs. 10–13 and JR (x)
(R = 1, . . . , N ) is an external source with O(N ) indices.
Before turning to the computation of the N -point off-shell amplitudes
it is instructive to remark that Eq. (5.3) is the string generalization of the
analogous formulas in an SO(N ) quantum particle dynamics:

Γ[JR (x)] = d[e(ζ)]d[ϕμ (ζ)]d[θi (ζ)]e−I[JR (x)] ,

ϕμ (0)=ϕμ (T )
T T
1
I[JR (x)] = m dξ e(ζ)ϕ̇μ (ζ)2 + +i θi (ζ)θ̇i (ζ)dζ (5.4)
0 e(ζ) 0
T
+ dζe(ζ)JR [ϕμ (ζ)][θi (ζ)(λR )ij θj (ζ)].
0

In a gauge e(ζ) =const, we find that lim δ 2 Γ[JR (x)]/δJR1 (x1 )δJR2 (x2 )
JR (x)→0
is the quantum-mechanical propagator of a free SO(N ) bosonic particle.
Let us now pass on to the problem of evaluating the scalar N -point
off-shell scattering amplitude through the correlation functions of the asso-
ciated effective quantum Liouville theory for the case of D being the upper
half-plane R+2
= {(ξ1 , ξ2 ) | ξ2 ≥ 0}. In momentum space, it is given by
ÂR1 , . . . , RN (p1 , . . . , pN )
E F
N ;  
(H) (A)
= d2 ξj e(ξj ) exp (A) i l
i(pj ; φ (ξj )) ψ1 (ξj )(λRj )il ψ̄2 (ξj ) ,
R2+ j=1

(5.5)
where ( ; ) means the Euclidean scalar product over the Lorentz indices and
N
(H)
d2 ξj = d2 ξj |ξb − ξa |2 |ξc − ξb |2
j=1 j=1;j=a=b=c

is the Mbius-invariant Haar measure which takes into account the (physical)
residual symmetry of the projective group not fixed by the conformal gauge.
The physical spectrum is determined by considering the poles of the ex-
pression ÂR1 · · · RN (p1 , . . . , pN ) and the associated residues identified with
the on-shell scattering amplitudes for the related string excitation.
Proceeding as in Refs. 10–13 by introducing the family Lj =
−2(j+1)δ(Z,Z ∗ ) ∗
e ∂Z ∗ e2jδ(Z,Z ) ∂Z of self-adjoint operators acting on an appro-
priate space of two-component real functions on R+ 2
= {Z = ξ1 +iξ2 ; ξ2 ≥ 0}
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 112

112 Lecture Notes in Topics in Path Integrals

we can perform the Gaussian functional integration over the scalar fields in
∗
the conformal gauge eaμ (Z, Z ∗ ) = eδ(Z,Z ) δμa . This yields the result

⎡ ⎛ ⎞⎤

N
−D
(L0 ) exp ⎣− ⎝ (pi ; pj )K (ε) (Zi , Zj , 2δ(Zi , Zi∗ ))⎠⎦ , (5.6)
(A) (A)
det 4

(i,j)=1

where K (ε) (Zi , Zj , 2δ(Zi , Zi∗ )) denotes the Neumann problem Green’s func-
tion of the covariant Laplacian conformally regularized on the half-plane
R2 . Its expression reads.

$ 
(ε) ∗ ∗ − 2π
1
ln(|Z − Z  ||Z − Z ∗ |), Z = Z ,
K (Z, Z , 2δ(Z, Z )) = ∗
δ(Z,Z )
2π − 1
4π ln ε − 1
2π ln |Z − Z ∗ |, Z = Z .
(5.7)
(eaμ (Z, Z ∗ ) = eaμ (ξ) = exp[δ(Z, Z ∗ )]δμa ).
Let us now evaluate the functional integrations over the fermionic com-
pactified string degrees of freedom and also of the U (1) gauge field Aμ (ξ).
Therefore, we face the problem of evaluating the functional integrals

WR1 · · · RN [eμa (ξ)](Z1 , . . . , ZN )


1
= D[ψ i (ξ)]D[ψ̄ i (ξ)]D[Aμ (ξ)] exp − d2 ξ[ψ i D(Aμ )ψ̄ i ](ξ)
2π R2+
⎛ ⎞
N
×⎝ psii1 (ξj )(λRj )il ψ̄2l (ξj )⎠ ,
j=1
(5.8)
where D(Aμ ) = ieaμ γa (∂μ +iAμ ) denotes the Dirac operator in the presence
of the background O(2) gauge field.
We evaluate Eq. (5.8) by using the fact that due to the chiral anomaly
the fermionic functional measures is not invariant under the chiral trans-
formations. Hence, by making the decoupling variable change in Eq. (5.8)
(see chapter 3),

εμν
ψ i (ξ) = eiγ5 η(ξ) χi (ξ), ψ̄ i (ξ) = χ̄i (ξ)eiγ5 η(ξ) , Aμ (ξ) = − (∂ν η)(ξ),
e(ξ)
(5.9)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 113

The D → −∞ saddle-point spectrum analysis of the open bosonic Polyakov 113

we obtain the result

WR1 · · · RN [eaμ (ξ)](Z1 , . . . , ZN )

= (det L0 )1/2 D[χi (ξ)]D[χ̄i (ξ)]D[η(ξ)]


1
exp − e(ξ)[χi (iγa ∂a )χ̄i ](ξ)d2 ξ
2 R2+

1
× exp − d2 ξe(ξ)(g μν ∂μ η∂ν η)(ξ)
2π R2+
⎛ ⎞
N
iγ5 η(Zj ,Zj∗ ) iγ5 η(Zj ,Zj∗ )
×⎝ [e χi (Zj , Zj∗ )]1 (λRj )il [χ̄l (Zj , Zj∗ )e ]2 ⎠ .
j=1
(5.10)

Since in Eq. (5.10) the (η(ξ), χi (ξ), χ̄i (ξ)) fields are decoupled we can
evaluate the corresponding functional integrations, yielding (in the confor-
mal gauge)

WR1 · · · RN [δ(Z.Z ∗ )](Z1 , . . . , ZN )

⎛ ⎞  N
 
1/4 ⎝
= (det L−1/2 ) Tr (λRσ(1) · · · λRσ(N ) )⎠ exp − δ(Zi , Zi∗ )
(σ) i=1
⎡ ⎛ ⎞⎤

×⎣ ⎝ (iγa ∂a )−1
N (Zi , Zj )
⎠⎦
(σ) (i,j) (a1 ,a2 )
⎛ ⎞

N
× exp ⎝− K (ε) (Zi , Zj , 2δ(Zi , Zi∗ ))⎠ , (5.11)
(i=j)
@
where (σ) means that we have to sum over all ways of pairing the fermion
fields in Eq. (5.10) and (iγa ∂a )−1 N (Zi , Zj ) denotes the flat Dirac Green’s
function with the Newmann condition along the real axis.
It is instructive to point out the O(N ) Chan-Paton factor
@
T r (σ) (λRσ(1) · · · λRσ(N ) ) in the expression quoted above. Additionally
N should be an even number, thus, implying that the RD × SO(N ) K = 2
bosonic string model possesses a quantum number which is subject to con-
servation.
Collecting the results of Eqs. (5.6) and (5.11) and evaluating the
Faddeev-Popov determinant, we get the final expression conformally
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 114

114 Lecture Notes in Topics in Path Integrals

regularized for the scalar N -point off-shell scattering amplitudes, Eq. (5.5):
⎡ ⎤

AR1 · · · RN (p1 , . . . , pN ) = ⎣T r
A (λRσ(1) · · · λRσ(N ) )⎦
(σ)

× D[δ(Z, Z ∗ )] exp{−Seff [δ(Z, Z ∗ )]}

⎧ ⎛ ⎞
⎨ N 
N
d2 ξj exp ⎝ δ(Zi , Zi∗ )⎠
(H)
×
⎩ R2+ j=1 i=j
⎡ ⎛ ⎞⎤

N
× exp ⎣− ⎝ (pi ; pj )K (ε) (Zi , Zi , 2δ(Zi , Zi∗ ))⎠⎦
(A) (A)

i=j
⎡ ⎛ ⎞⎤

N
× exp ⎣− ⎝ [(pi ; pj ) + 1]K (ε) (Zi , Zj , 2δ(Zi , Zi∗ ))⎠⎦
(A) (A)

i=j
⎡ ⎛ ⎞⎤⎫⎞
 N ⎬
× ⎣ ⎝ (iγa ∂a )−1
(N ) (Zσ(i) ; Zσ(j) )
⎠⎦ ⎠ , (5.12)
⎭
(σ) (i,j) (a1 ,a2 )

2
where the effective Liouville action in the upper half-plane R+ is given by
26 − D − (N − 1) 1
Seff [δ(Z, Z ∗ )] = d2 ξ[ (∂a δ)2 + μ2 e2δ ](ξ)
12π R2+ 2
+∞ +∞
λ dξ1 eδ(ξ1 ,0) − D/8π dξ1 (∂N δ)(ξ1 , 0) (Z = ξ1 + iξ2 ).
−∞ −∞
(5.13)
2
Since the solution of the Liouville field theory in R+ was not found
yet, which would provide the complete solution of Eq. (5.12), we imple-
ment a saddle-point approximation to evaluate Eq. (5.12) as introduced √
in Ref. 10. By requiring the usual “boundary” restriction λ = −(5/ 2)μ
[cf. Eq. (3.11) of Ref. 10] the (natural) D → −∞ solution of the theory
2
described by Eq. (5.13) is the Poincar metric in R+ : namely,
 
1 2
δ(ξ1 , ξ2 ) = ln 2 (Z = ξ1 + iξ2 ). (5.14)
2 μ |Z − Z ∗ |2
We remark that this solution renders the action singular. However the
action Seff [δ(Z, Z ∗ )] does not depend on the points Zi , so it cancels with
the normalization factor of the quantum average.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 115

The D → −∞ saddle-point spectrum analysis of the open bosonic Polyakov 115

By substituting Eq. (5.14) in Eq. (5.12), we finally get the D → −∞ (or

equivalently N → −∞) scalar dual Green’s function of the RD × SO(N )
K = 2 string model:
⎛ ⎞

AAR1 · · · RN (p1 , . . . , PN ) = [ḡ(ε, pi )]N ⎝T r (λRσ(1) · · · λRσ(N ) )⎠
(σ)
⎛ ⎞
N N
p2
(N ) ⎝ −1 ⎠
× d2 ξj |Zi − Zi∗ | + πi
R2+ j=1 j=1
⎛ ⎞

×⎝ ⎠
(A) (A)
(|Zi − Zj | |Zi − Zj∗ |) [(pi ;pj )+1]/π

i<j
⎛ ⎞
⎜  ⎟
×⎜
⎝ [(iγa ∂a )−1 ⎟
N (Zσ(i) , Zσ(j) )]a1 a2 ⎠ ,
(σ) (i,j) (a1 ,a2 )
(i=j)
(5.15)
where the conformally regularized dual coupling constant model is given
explicitly by
N
1/2−( p2i /2πN ) 
2 i=j
N
p2i /4πN
ḡ(ε, pi ) = ε i=j . (5.16)
μ2
At this point it is instructive to compare with the similar result obtained
for the fermionic string [cf. Eq. (18) of Ref. 11].
In order to determine the associated spectrum we have to find the poles
in the external momentum variables p2i ) = (pi ; pi ). Such poles occur when
Zi and Zi∗ come close together. As a result, (Euclidean) poles exist when
p2i /π = 1 = −1, −2, . . . or p2i /π = 0, −1, −2, . . . . (5.17)
This fact implies that the spectrum of the RD × SO(N ) K = 2 bosonic
string at the D → −∞ saddle point is not sensitive to the axial anomaly and
coincides with the spectrum of the pure D → −∞ fermionic string, a string
without tachyans. As another comment, we remark that the absence of the
tachyon excitation of these models is entirely due to the covariant treatment
@N
of the fermionic degrees which produces the factor exp[− i=1 δ(Zi , Zi∗ )]
in Eq. (11), thus, suppressing the tachyon excitation from the pure bosonic
spectrum [cf. Ref. 10, Eq. (9.11)]. Opposite to the usual “add-hoc” spec-
trum projection onto an even “G-parity” sector implemented at the string
critical dimension Dc as an operator quantization analysis.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 116

116 Lecture Notes in Topics in Path Integrals

5.3 References

[1] Peter G.O. Freund, Phys. Lett. 151B, 387 (1985).

[2] Goddard, Nucl. Phys. B116, 157 (1976).
[3] C. Lovelace, Phys. lett. 135B, 75 (1984).
[4] J. Maharana and G. Veneziano, Phys. Lett. 169B, 177 (1986).
[5] M.B. Green and J. Schwarz, Phys. Lett. 149B, 117 (1984); 151B,
21 (1985).
[6] D. Gross, J. Harvey, E. Martinec, and R. Rohm, Nucl. Phys.
B256, 253 (1985).
[7] P. Candelas, G. Horowitz, A. Strominger, and E. Witten, Nucl.
Phys. B258, 46 (1985).
[8] A.N. Redlich and Howard J. Schnitzer, Phys. Lett. 167B, 315
(1986).
[9] D Nemeschansky and S. Yankielowicz, Phys. Rev. Lett. 54, 620
(1985); 54, 1736(E) (1985).
[10] B. Durhuus, H.B. Nielsen, P. Olesen, and J.L. Petersen, Nucl.
Phys. B196, 498 (1982).
[11] Luiz C.L. Botelho, Phys. Lett. 152B, 358 (1985).
[12] E.S. Fradkin and A.A. Tseytlin, Phys. Lett. 158B, 316 (1985).
[13] B. Durhuus, P. Olesen, and J.L. Petersen, Nucl. Phys. B198, 159
(1982).
[14] S.P. de Alwis, Phys. Lett. 168B, 59 (1986).
[15] R.E. Gamboa Saravi, F.A. Schaposnik, and H. Vucetich, Phys.
Rev. D30, 363 (1984).
[16] F. Gliozzi, J. Scherk, and D.J. Olive, Nucl. Phys. B122, 253
(1977).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 117

Chapter 6

The Electric Charge Confining in

Abelian Rank two Tensor Field Model

We present a Wilson Loop evaluation of the binding energy of electric static

charges in an Abelian Rank two Tensor Field Model in R4 through path
integrals.

6.1 Introduction

It has been argued by A.M. Polyakov [1] that the presence of monopoles
excitations in the vacuum of Compact Quantum Electrodynamics is the
main fact responsible for the confinement or screening of the electric charge
of the electron excitations.
In this chapter we intend to show a similar, however diferent dynam-
ics of confinement of electrical charges in the presence of tensor fields by
path integral manipulations with abelian Wilson loops path integrals in the
framework of dimensional regularization and by phenomenologically repre-
senting the dynamics of the theory’s vacuum through an effective dynamics
of a low-energy rank-two antisymmetric tensor field ([2]) (see appendix for
some comments).

6.2 The interquark potential evaluation

Let us thus start our analyzes by recalling a propose to represent quantum

field dynamics for a rank-two antisymmetric tensor by means of a Gaussian

117
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 118

118 Lecture Notes in Topics in Path Integrals

path integral ([2])

> 1 ;
Z= DF [Bμν ] exp − d4
x 2
Bμν
4e2
=
dB dB  ?
+ dB arsen( 2 ) + m 1 − ( 2 )2 (x)
2
m m
low-energy
> 1 ?
∼ D F
[B μν ] exp − d 4
x(B μν [−(∂ 2
) + m 2
]B μν )(x)
m2 →∞ 4e2
(6.1)
where m2 is a dimensional transmutation (very heavier) mass parameter
which is (phenomenologically) signaling the energy scale where theory’s
non-perturbative effects are expected to be relevant.
In order to consider the presence of an external electromagnetic field dy-
namics we consider an interaction with the electromagnetic field by means
of a kind of topological term, namely
/ 1 0
Z = DF [Bμν ]DF [Aμ ]δ(∂μ Aμ ) exp − 2 d4 xFμν
2
(A)(x)
4e
/ 1 0
× exp − 2 d4 x(Bμν [−(∂ 2 ) + m2 ]Bμν )(x)
4e
/ 0
× exp θ d4 x(Bμν ∗ Fμν (A))(x) . (6.2)

Let us show the electric charge confinement in the effective theory as

described by the path integral eq. (6.2) under the constraint of the fine
tunning value θ = |m|.
In order to analyze such phenomenon, we consider the binding energy
between two probing electric static charges with opposite charge’s signal
with a space-time trajectory description C(R,T ) , the loop boundary of the
> ?
rectangle SR,T ) = − T2 ≤ x0 ≤ + T2 ; − R2 ≤ x1 ≤ + R2 .
We have thus the standard result for this binding energy in terms of the
gauge invariant Wilson loop
⎧ ⎫
⎪
⎪ +
W [C(R,T ) ]
,- .⎪
⎪
⎪
⎨ 1 ⎪
⎬
Ebin (R) = lim − lg exp(i α μ
Aμ (X (s)dX (s)) (6.3)
T →∞ ⎪
⎪ T ⎪
⎪
⎪
⎩
C(R,T ) ⎪
⎭

where the euclidean path integral average, normalized to unity, is defined

by eq. (6.2). After realizing the Bμν (x) Gaussian path integral, we arrive
at the effective U (1)-invariant path integral result for the Wilson Loop
observable inside eq. (6.3).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 119

The Electric Charge Confining in Abelian Rank twoTensor Field Model 119

W [C(R,T ) ] = DF [Aμ ]δ (F ) (∂μ Aμ )

> 1
× exp − d4 xd4 yAμ (x)[(−∂ 2 )x δ (4) (x − y)
2e2
?
− θ2 (−∂ 2 )x (−∂x2 + m2 )−1 (x, y)]Aμ (y)

× exp i Aμ (X α (s)dX μ (s) . (6.4)

C(R,T )

The searched binding energy takes thus the simple form in a dimensional
regularized integral-distributional form
 2 
e dν k
Ebin (R) = lim − 2 μ
fμ (k, C(R,T ) )D(k )f (−k, C(R,T ) ) (6.5)
T →∞ T (2π)ν
with the rectangle form factors ([4])

dxμ (s)
fμ (k, C(R,T ) ) = e−i(k0 x0 (s)+k1 x1 (s))
C(R,T ) ds
 
∂
= εμν (e−ikα xα ) d2 xα
S(R,T ) ∂xν
$
− k40 sin( k02T ) sin( k12R ) for μ = 0
= (6.6)
+ k41 sin( k02T ) sin( k12R ) for μ = 1
The effective electromagnetic propagator, local in momentum space, is
explicitly given by
 −1
θ2 k 2 k 2 + m2
D(k 2 ) = k 2 − 2 = 4 · (6.7)
k +m 2 k + k 2 (m2 − θ2 )
Now one can easily see that for the fine tunning choice m = +θ, one
obtains after straightforwardly calculations ([4]), the expected confining
Cornell form ([5]) for the binding energy
e2  
Ebind (R) = − + (4πe2 m2 )R . (6.8)
4πR
It is worth that for m = θ, or θ = iθ (θ ∈ R), call the reader attention that
the binding energy is of the Yukawa form and thus leading to a screening
picture for the inter-electronic potential for compact QED in R4 .
At this point, let us comment and compare ours results with the at-
tempts done in ref.[3] through Hamiltonian methods for a similar quantum
field model.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 120

120 Lecture Notes in Topics in Path Integrals

We feel that these works may be not well-defined by the somewhat cum-
bersome use of cut-offs (otherwise infinite) ordinary integrals. For instance,
in these refs.[5], it has been obtained the following expression for the bind-
ing energy (see eqs. (17)-(29) of ref.[3] where R = (y − y  ), M 2 = m2 + e2
in the refs author’s notation):
(1) (2)
Ebind (R) = V(R) + V(R) (6.9)
with

y y
e2 1
dzi δ (3) (x 
(1)
V(R) = − 3
d x −z ) ∇2 dz i δ (3) (x − z)
2 y ∇2x − M 2 x y
2 −MR
H
= (e e ) 4πR (6.10)
and

y y
e 2 m2 1
V (2)
(R) = + d3 x dzi δ (3) (x − z  ) dz i δ (3) (x − z)
2 y ∇x − M 2
2
y
e2  −MR  e2 M ε3
=− e /R + R n 1 − 2 (6.11)
4π 8π M
where ε is a cutt-off to be imposed in order to make sense for their diver-
gent ordinary integrals. However, it appears that this process of handling
distributions has drawbacks, since one can easily use the screening result
eq. (6.10) to evaluate eq. (6.11) through the use of the simple operatorial
decomposition
y
e 2 m2 (∇x )2
V (2)
(R) = 3
d x dzi δ (3) (x − z  )
2 y (∇x )2 (∇2x − M 2 )

y
× dz i δ (3) (x − z)
y
 y  
e 2 m2  1 1 1 1
= d x 3
dz1i δ (3) (x −z ) − 2 + 2
2 y M (∇x )2 M (∇x − M 2 )
2
 
y
e2 M 2 1 e 2 m2 e−MR
× dz i δ (3) (x − z) =− +
y 2M 2 R 2M 2 4πR
(6.12)
which clearly is finite and differs from the their claimed cut-off implicit
confining result. At this point let us suggest that the use of Hamiltonian
formalism to handle mathematically gauge theories, specially Yang-Mills
theory and its variants is somewhat difficulted by the fact that one must
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 121

The Electric Charge Confining in Abelian Rank twoTensor Field Model 121

fix the gauge and this makes recourse for formal BRST realizations of Gauge
Invariance in the observables evaluation in order to make the results phys-
ically acceptable. Certainly such somewhat formal technique making use
of “infinitesimal” grassmanian parameters in a theory with solely physical
bosons fields as imputs must be applied very carefully outside the well-
founded evaluations of matrix (on-sheel) scattering amplitudes which does
not exist in pure Yang-Mills theory due to severe infrared divergencies.
That was the main reason of reformulating all the gauge theories by means
of gauge invariant path integrals and mainly with the objective that in the
confining phase all the evaluations should be done non-perturbativelly (see
appendix for supplementary comments).

6.3 Appendix A: The dynamics of the QCD(SU (∞)) ten-

sor fields from strings

In this appendix we intend to high-light on some ideas and loop space

formulae about non-space time supersymmetric random surfaces represen-
tations for bosomic QCD(SU (Nc )) (including the t’Hooft planer diagrams
limit of Nc → ∞).
Alexandre M. Polyakov in the article ref [6], has argued that one possible
random surface path integral representation for Wilson Loops in bosonic
QCD(SU (Nc )) shall be given by our discovery that mathematical bosonic
strings interacting with A Migdal intrinsic fermionic Strings degree of
freedom in the SU (3) × SU (2) × U (1) fundamental representation [in an
non-abelian bosonized form] and interacting with a rank-two antisymmetric
tensor field through the closed string world-sheet orientation tangent tensor
namely ([6], [7]).
9    
−1
W [C] = exp − 2
d ξζ μν β
(X (ξ)). Tr(g Bμν (X(ξ)g)
Si ∂s= {g}

exp[−(σ − model action for g with Wess-Zumino-Novikov

:
Topological Term)] (A-1)
B

The author of the second set of ref.[7] present arguments that the av-
erage over the Bμν (x), probably representing the non-trivial random flux
structure of the QCD-vacuum should be defined by a pure white-noise set
of random fluxes. In the path integral language ([6], ]8]), the B-average
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 122

122 Lecture Notes in Topics in Path Integrals

should be defined by the functional measure:

9:  
1
= D [Bμν (x)] exp −
F
2
d4 2
x Bμν (x) (A-2)
B 4(g ∞)

2
where g∞ = lim (g 2 Nc ) < ∞ denotes the Nc = +∞ t’Hooft QCD cou-
Ni →∞
pling constant.
It is very important to remark now that eq. (A-1) with the White-
Noise random flux average eq. (A-2), satisfies the QCD loop wave equation
exactly under the geometrical hypothesis that the bosonic surfaces (string
world-sheets) defining the QCD string should satisfies the self-intersecting
orthogonality tangent plane constraints (ξ = (s, σ)), X μ (s, 0+ ) = u (s))

ζμν (X β (s, σ) · ζ μν (X β (s , σ  )) = 0 if s = s

ζ μν (X β (s, σ)) ζμν (X β (s , σ)) = 0 if s = s (A-3)

Note that the second constraint in eq. (A-3) means that one allows
non-trivial topology in the intrinsic (mathematical) string time-direction
evolution 0 ≤ σ < ∞. However, the first geometrical constraint may be
connected (not proved yet) to the fact that the bosonic loops X μ (s, σ̄) (with
σ̄ fixed) should physically corresponding to R4 space-time euclidean quark-
antiquark trajectories on the fermionic quark functional determinant in the
presence of an abelian color singlet vectorial source, for instance:

1 ∞ dΓ
g det(∂(Aμ ) + Jμ ) = − d4 z β DF [X(s)]
2 0 T β β
X (0)=X (T )=z β
 
T T
1
exp − ds · X 2 (s) × exp ie ds Jμ (X(s))
2 0 0
%    &
g2 T δ
Tr P Spin exp μ ν
ds[γ , γ ]
4i 0 δσμν (X(s))
% &
T
1
× Trcolor P exp ig Aμ (X(s))dX (s)μ
(A-4)
Nc 0

δ
with denoting the formal Mandelstam-Migdal area functional
δσμν (xβ )
δ δ
derivative = and satisfying thus the
β
δσμν (X (s)) δ[(Ẋμ Xν − Ẋν Xμ )(s)]
Pauli Exclusion principle which by its turn translates geometrically by
the condition that at non-trivial proper-time trajectories self-intersections
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 123

The Electric Charge Confining in Abelian Rank twoTensor Field Model 123

dX μ (s) dXμ (s )

X μ (s) = X μ (s ) (0 < s, s < T ), we have always that · ≡ 0.
ds ds
This important Pauli Exclusion random surfaces probably signals that
strings representations for scalar quark QCD is ill-defined in relation to
strings representations for fermionic quark QCD as proposed in refs [7],
[8].
At this point one may expect that in the very low energy scale of QCD,
one can integrate out all the strings degrees of freedom and arrives at the
somewhat crude however local effective rank-two antisymetric tensor field
simplest action eq. (A-1), but added with Abelian Wilson Loop as a sort
of remnants of the full non-Abelian colour Wilson Loop phase factor. Note
that the closed loops are already under a fixed proper-time parametrization,
as one can see from the loop space expression for eq. (A-4).
As an important point to be singlet out is that the Loop Space/String
program for QCD semms much better defined if one introduces super-
symetry on the quark-antiquark world lines and string (spinning) world
sheets to handle directly the Lorentz spin content of the quarks, instead
of the obligatory operational loop Feynman Lorentz spin factor inside
eq. (A-1) with the operational expression:
    
g2 δ
ΦSpin [] = PDirac exp − ∞ ds[γ μ , γ ν ] ×
2 C δζμν (X β (s, σ)) σ→0+
(A-5)
equivalently written by the introduction of a new set of intrinsic neutral
Fermion fields in the string world sheet: {ψ μ (ξ)}μ=0,1,2,3 and representing
the “String Lorentz Spin”. Namely:
 
1
ΦSpin
αβ [] = [D F μ
ψ(ξ) ] × exp − 2 μ
d ξ(ψ (iγ)ψμ )(ξ) (ψα (0, 0)ψβ (T, 0)
2
g2
× exp − ∞ d2 ξ(ψα [γ μ , γ ν ]αβ ψβ )(ξ)
2

δ
× Tr g −1 (ξ) g(ξ) (A-6)
δζμν (X β (ξ))

It is hoped that some sort of solitons/Lorentz Spin Vertexs in the QCD

Spinning, flavor charged string theory eq. (A-1)–eq. (A-5) should be can-
didates to describe the Baryons Physics, basic issue not yet handled in the
loop space framework and others holographic proposals for strings repre-
sentations in supersymmetric Yang-Mills theory.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 124

124 Lecture Notes in Topics in Path Integrals

6.4 References

[1] A.M. Polyakov, Gauge Field and Strings. Harwood Academic

Chur., Switzerland, 1987.
[2] A.M. Polyakov, Particle Phys. B486, 23, (1997).
[3] P. Gaete, C. Wotzasik, Phys. Lett. B 601, 108, (2004).
[4] Luiz C.L. Botelho, Phys. Rev. 700, 045010, (2004).
[5] Eichten, Gottpried, T.Kinoshita, K.O. Lane, T.M. Yau, Phys. Rev
170, 3050, (1978).
[6] Alexandre M. Polyakov, String Theory and Quark Confinement –
arKiv: hep-th/971100201, 1 Nov 1997.
[7] Luiz C.L. Botelho - CBPF - NF - 051/85 (Pre-Print) - (1985)
- CBPF - NF - 034/86 (Pre-Print) - (1986)
- CBPF - NF - 045/86 (Pre-Print) - (1986)
- Brazilian Journal of Physics, vol. 21, no¯ 4, (1991)
- Brazilian Journal of Physics, vol. 19, no¯ 3, (1989)
- Caltech preprint 68 - 1444, (1987)
[8] Luiz C.L. Botelho, J. Math. Phys. 30 (9), 2160, (1989).
[9] A.M. Polyakov, Nucl. Phys. B 486, 28, (1997)
[10] L.C.L. Botelho, Phys. Lett. B 152, 358, (1985)
- Phys. Lett. B 169, 428, (1986)

6.5 Appendix B: Path-integral bosonization for a nonrenor-

malizable axial four-dimensional fermion model

We study the bosonization and exact solubility of a nonrenormalizable four-

dimensional axial fermion model in the framework of anomalous chiral path
integrals.

6.6 Introduction

The study of two-dimensional fermion models in the framework of chiral

anomalous path integrals has been shown to be a powerful nonperturbative
technique to analyze the two-dimensional bosonization phenomenon.
It is the purpose of this appendix to implement this nonperturba-
tive technique to solve exactly a nontrivial and nonrenormalizable four-
dimensional axial fermion model which generalizes for four dimensions the
two-dimensional model studied in Ref. 2.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 125

The Electric Charge Confining in Abelian Rank twoTensor Field Model 125

6.7 The model

Let us start our analysis by considering the (Euclidean) Lagrangean of the

proposed Abelian axial (mathematical) model
1
L1 (ψ, ψ̄, φ) = ψ̄γμ (i∂μ − gγ5 ∂μ φ)ψ + g 2 (∂μ φ)2 + V (φ), (B-1)
2
where ψ(x) denotes a massless four-dimensional fermion field, φ(x) a pseu-
doscalar field interacting with the fermion field through a pseudoscalar
derivative interaction, and V (φ) is a φ self-interaction potential given by
−g 4 2 2 g2
V (φ) = φ(∂ μ φ) (−∂ φ) + (−∂ 2 φ)(−∂ 2 φ). (B-2)
12π 4 4π 2
the presence of the above φ potential is necessary to afford the exact solu-
bility of the model as we will show later [eq. (18)].
The Hermitian γ matrices we are using satisfy the (Euclidean) relations

{γμ , γν } = 2δμν , γ5 = γ0 γ1 γ2 γ3 . (B-3)

The Lagrangian L1 (ψ, ψ̄, φ) is invariant under the global Abelian and
chiral Abelian groups

ψ → eiα ψ, ψ → eiγ5 β ψ, (α, β) ∈ R, (B-4)

with the Noether conserved currents at the classical level:

∂μ (ψ̄γ 5 γ μ ψ) = 0, ∂μ (ψ̄γμ ψ) = 0. (B-5)

In the framework of path integrals, the generating functional of the

correlation functions of the mathematical model associated with the La-
grangian L1 (ψ, ψ̄, φ) is given by
1
Z[J, η, η̄] = × D[φ]D[ψ]D[ψ̄]
Z[0, 0, 0]

× exp − d4 x[L1 (ψ, ψ̄, φ) + Jφ + η̄ψ + ψ̄η](x) . (B-6)

In order to generalize for four dimensions the chiral anomalous path

itnegral bosonization technique as in Refs. [2] and [3], we first rewrite the
full Dirac operator in the following suitable form:

D[φ] = iγμ (∂μ + igγ5 ∂μ φ)

= exp(igγ5 φ)(iγμ ∂μ ) exp(igγ5 φ). (B-7)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 126

126 Lecture Notes in Topics in Path Integrals

Now we proceed as in the two-dimensional case by decoupling the

fermion field from the pseudoscalar field Φ(x) in the Lagrangian L1 (ψ, ψ̄, φ)
by making the chiral change of variables:
ψ(x) = exp[igγ5 φ(x)]χ(x),
(B-8)
ψ̄(x) = χ̄(x) exp[igγ5 φ(x)].
On the other hand, the fermion measure D[ψ]D[ψ̄], defined by the eigen-
vectors of the Dirac operator D[φ], is not invariant under the chiral change
and yields a nontrivial Jacobian, as we can see from the relationship
 
D[ψ]D[ψ̄] exp − d x(ψ̄ D[ψ]ψ)(x)
4

= det(D[φ])

= J[φ] D[χ]D[χ̄] exp − |d4 x(χ̄iγμ ∂μ χ)(x) . (B-9)

Here J[φ] = det( D[φ])/ det( D[φ = 0]) is the explicit expression for this
Jacobian.
It is instructive to point out that the model displays the appearance
of the axial anomaly as a consequence of the nontriviality of J[φ], i.e.,
∂μ (ψ̄γμ γ5 ψ)(x) = {(δ/δφ)J[φ]}(x).
So, to arrive at a complete bosonization of the model Eq. (B-6) we face
the problem of evaluating J[φ].
Let us, thus, compute the four-dimensional fermion determinant
det( D[φ]) exactly, In order to evaluate it, we introduce a one-parameter
family of Dirac operators interpolating the free Dirac operator and the
interacting one D(ζ) [φ]: namely,
D(ζ) [φ] = exp(igγ5 ζφ)(iγμ ∂μ ) exp(igγ5 ζφ) (B-10)
with ζ ∈ [0, 1].
At this point we introduce the Hermitian continuation of the operator
D(ζ) [φ] by making the analytic extension in the coupling constant ḡ = ig.
This procedure has to be done in order to define the functional determinant
by the proper-time method since only in this way (Dζ [φ]) can be considered
as a (positive) Hamiltonian.
The justification for this analytic extension in the model coupling con-
stant is due to the fact that typical interaction energy densities such as
ψ̄γ 5 ψ, ψ̄γμ Δμ ψ, which are real in Minkowski space-times, become complex
after continuation in Euclidean space-time. As a consequence, the above
analytic coupling extension must be done in the proper-time regularization
for the Dirac functional determinant.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 127

The Electric Charge Confining in Abelian Rank twoTensor Field Model 127

By using the property

d ζ
D [φ] = ḡγ5 φ Dζ [φ]+ Dζ [φ]γ5 ḡφ, (B-11)
dζ
we can write the following differential equation for the functional determi-
nant (see Ref. [4] and chapter 3):
d
ln det Dζ [φ]
dζ

= 2 lim d4 x Trx|(ḡγ5 φ) exp{−ε(Dζ [φ])2 }|x, (B-12)

ε→0+

where Tr denotes the trace over Dirac indices.

The diagonal part of exp{−ε(Dζ [φ])2 } has the asymptotic expansion
x| exp{−ε(Dζ [φ])2 }|x
1 ε2
∼ 1 + εH1 (φ) + H2 (φ) (B-13)
ε→0+ 16π 2 ε2 2!
with the Seeley coefficients given by (see the Appendix)
H1 (φ) = ζ ḡγ5 ∂ 2 φ − ḡζ 2 (∂μ φ)2 (B-14)
and
H2 (φ) = 2∂ 2 H1 (φ) + 2ζ ḡγ5 (∂μ φ)∂μ H1 (φ) + 2[H1 (φ)]2 . (B-15)
By substituting Eqs. (B-13)–(B-15) into Eq. (B-12), we obtain finally
the result for the above-mentioned Jacobian:
J[φ] = J0 [φ, ε]J1 [φ]. (B-16)
Here J0 [φ, ε] is the ultraviolet cutoff-dependent Jacobian term
g2
J0 [φ, ε] = exp d4 x[φ(−∂ 2 )φ](x) (B-17)
4π 2 ε
and J1 (φ) is the associated Jacobian finite part
g2
J1 [φ] = exp − d4 x(−∂ 2 φ)(−∂ 2 φ)(x)
4π 2
g4
× exp d4 x[φ(∂μ φ)2 (−∂ 2 φ)](x) . (B-18)
12π 2
From Eqs. (B-17) and (B-1), e can see that the (bare) coupling constant
2
g gets an additive (ultraviolet) renormalization. Besides the Jacobian term
cancels with the chosen potential V (φ) in Eq. (B-2).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 128

128 Lecture Notes in Topics in Path Integrals

As a consequence of all these results, we have the following expression

for Z[J, η, η̄] with the fermions decoupled:
1
Z[J, η, η̄] = D[φ]D[χ]D[χ̄]
Z[0, 0, 0]
/ 1 1 2
× exp − d4 x[ x̄(iγμ ∂μ )χ + gR (∂μ φ)2
2 2
0
+ η̄ exp(igR γ5 φ)χ + χ̄ exp(igR γ5 φ)η + Jφ](x) . (B-19)
This expression is the main result of our paper and should be com-
pared with the two-dimensional analogous generating functional analyzed
in Ref. [2]. Now we can see that the quantum model given by Eq. (B-6)
although being nonrenormalizable by usual power counting and Feynman-
diagrammatic analysis it still has nontrivial and exactly soluble Green’s
functions. For instance, the two-point fermion correlation function is easily
evaluated and produced the result
ψα (x1 )ψ̄β (x2 ) = Sαβ
F
(x1 − x2 ; m = 0)
× exp[−ΔF (x1 − x2 ; m = 0)]. (B-20)
Here Δ(x1 −x2 ; m = 0) is the Euclidean Green’s function of the massless free
scalar propagator. We notice that correlation functions involving fermions
ψ(x), ψ̄(x) and the pseudoscalar field φ(x) are easily computed too (see
Ref. [2]).
As a conclusion of our paper let us comment to what extent our proposed
mathematical Euclidean axial model describes an operator quantum field
theory in Minkowski space-times.
In the operator framework the fields ψ(x) and φ(x) satify the following
wave equations:
iγμ ∂μ ψ = igγμ γ5 (∂μ φ)ψ,
δΔ (B-21)
φ = ∂μ (ψ̄γμ γμ ψ) + .
δφ
It is very difficult to solve exactly Eq. (B-21) in a pure operator framework
because the model is axial anomalous [∂μ (ψ̄γμ γ5 ψ) = 0]. However the path-
integral study Eq. (B-19) shows that the operator solution of Eq. (B-21)
[in terms of free (normal-ordered) fields] is given by
Ψ(x) =: exp[iγ5 φ(x)] : χ(x) (B-22)
since it is possible to evaluate exactly the anomalous fivergence of the above-
mentioned axial-vector current and, thus, choose a suitable model potential
V (φ) which leads to the above simple solution.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 129

The Electric Charge Confining in Abelian Rank twoTensor Field Model 129

It is instructive to point out that the operator solution Eq. (B-22) co-
incides with the operator solution of the U (1) vectorial model analyzed by
Schroer in Ref. [8] [the only difference between the model’s solutions being
the γ5 factor in the phase of Eq. (B-22)].
Consequently we can follow Schroer’s analysis to conclude that the Eu-
clidean correlation functions Eq. (B-20) define Wightman functions which
are distributions over certain class of analytic test functions. But the model
suffers the problem of the nonexistence of time-ordered Green’s functions
which means that the proposed axial model in Minkowski space-times does
not satisfy the Einstein causality principle.
Finally we remark that the proposed model is to a certain extent less
trivial than the vectorial model since for nondynamical φ(x) field the as-
sociated S matrix is nontrivial and is given in a regularized form by the
result
 +∞ 
4
S = T exp i (ψ̄γμ γ5 ∂μ φψ)(x)d x
−∞
 +∞ 
δ
= exp −i φ(x) Jε [φ] (x)d4 x (B-23)
−∞ δφ(x)

with Jε [φ] expressed by Eq. (B-16).

6.8 Appendix C

We now briefly calculate the asymptotic term of the second-order positive

differential elliptic operator:

(Dζ [φ])2 = (−∂ 2 ) − (ḡγ5 ζ∂μ φ)∂μ − ḡζγ5 ∂ 2 φ + (∂μ φ)2 . (C-1)

For this study, let us consider the more general second-order elliptic four-
dimensional differential operator (non-necessarily) Hermitian in relation to
the usual normal in L2 (RD ) (Ref. [7]):

Lx = −(∂ 2 )x + aμ (x)(∂μ )x + V (x). (C-2)

Its evolution kernel K(x, y, ζ) = x|e−ζD |y satisfies the heat-kernel

equation
∂
K(x, y; zeta) = −Lx K(x, y; ζ),
∂ζ (C-3)
lim+ K(x, y; ζ) = δ (D) (x − y).
ζ→0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 130

130 Lecture Notes in Topics in Path Integrals

The Green’s function K(x, y, ζ) has the asymptotic expansion

 ∞

lim K(x, y; ζ) ∼ K0 (x, y; ζ) ζ m Hm (x, y) , (C-4)
ζ→0+
m=0

where K(x, y, ζ) is the evolution kernel for the n-dimensional Laplacian

(−∂ 2 ). By substituting Eq. (C-4) into Eq. (C-3) and taking the coinciding
limit x → y, we obtain the following recurrence relation for the coefficients
Hm (x, x):
∞
∞
 1 n  ζn
ζ Hn+1 (x, x) = − (−∂x2 )Hn (x, x)
n=0
n! n=0
n!
∞
ζn x
+ an (x) ∂ Hn (x, x)
n=0
n! μ
∞
ζn
+V (x) Hn (x, x) . (C-5)
n=0
n!
For D = 4, Eq. (C-5) yields the coefficients
H0 (x, x) = 14×4 ,
H2 (x, x) = −V (x), (C-6)
H2 (x, x) = 2[−∂x2 V (x) + aμ (x)∂μ V (x) + V (x)]. 2

Now substituting the value aμ (x) = [−ζ ḡ∂μ φ(x)]14×4 and V (x) =
[(−ḡζγ5 ∂x2 φ + ∂μ φ)2 ]14×4 into Eq. (C-6) we obtain the result Eqs. (B-14)
and (B-15) quoted in the main text.

6.9 References

[1] K. Fujikawa, Phys. Rev. D 21, 2848 (1980); R. Roskies and F.

Schaposnik, ibid. 23, 518 (1981).
[2] L. Botelho, Phys. Rev. D 31, 1503 (1985); A. Das and C.R.
Hagen, ibid. 32, 2024 (1985).
[3] L. Botelho, Phys. Rev. D 33, 1195 (1986).
[4] O. Alvarez, Nucl. Phys. B 238, 61 (1984).
[5] V. Romanov and A. Schwartz, Teor. Mat. Fiz. 41, 190 (1979).
[6] K. Osterwarder and R. Schrader, Helv. Phys. Acta 46, 277 (1973).
[7] P.B. Gilkey, in Proceedings of Symposium on Pure mathematics,
Stanford, California, 1973, edited by S.S. Chern and R. Osserman
(American Mathematical Society, Providence, RI, 1975), Vol. 127,
Pt. II, p. 265.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 131

The Electric Charge Confining in Abelian Rank twoTensor Field Model 131

[8] B. Schroer, J. Math. Phys. 5, 1361 (1964); High Energy Physics

and Elementary Particles (IAEC, Vienna, 1965).
[9] K. Bardakci and B. Schroes, J. Math. Phys. 7, 16 (1966).
[10] A. Jaffe, Phys. Rev. 158, 1454 (1966).
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 133

Chapter 7

Infinities on Quantum Field Theory:

A Functional Integral Approach

We analyze on the formalism of probabilities measures-functional integrals

on function space, the problem of infinities on Euclidean field theories. We
also clarify and generalize ours previous published studies on the subject.

7.1 Introduction

The subject of infinities of Euclidean Quantum Field Theories, or in gen-

eral Minkowskian Quantum Field Theories, has been well grounded with
practical Calculations on Quantum Physics since its full inception on 1950
years ([1]).
The purpose of this chapter is to study the nature of ultraviolet infinities
on Euclidean Quantum Field Path integral through an analitically regular-
ized, mathematical rigorously path integrals. This study is presented in
Section 1. In Section 2, we clarify our previous studies on the subject ([2],
[7]) by analyzing in detail all those estimates leading to a correct under-
standing of the well-known problem of coupling constant renormalization
on QFTs in a two-dimensional scalar interacting field model. In Section 3,
mainly of pedagogical purpose, we present a mathematically detailed con-
struction of the Wiener Kac functional measure. In Section 4, we analyze
the mathematical structure of the so called Feynman geometrodynamical
propagation on Euclidean QFT, by using the mathematical rigorous results
of Section 3.
In several appendixes, we add several calculations useful to understand
the bulk of this chapter.

133
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 134

134 Lecture Notes in Topics in Path Integrals

7.2 Infinities on Quantum Field Theory on the Functional

Integral Formalism

Euclidean quantum fields are expected to be mathematically defined objects

by the analytic continuation of Minkowskian self-adjoints quantum fields to
imaginary time, where this precise analytic continuation is well expressed
through the famous Bargmann-Hall-Wightman theorem ([1]). It is thus
expected that Euclidean self-adjoints quantum field operators on RD should
be naturally constructed (from a rigorous mathematical point of view) from
a probability measure dμ(φ) on the L. Schwartz Tempered Districutions
S  (RD ). It is also expected that the following functional, the called theory’s
generating functional Z(j) ∈ C(S(RD ), R) should furnishes the bridge of
such time analytic extension though the identity

vac | exp iφ(j)|Ωvac  = Z(j).

Ωeucl eucl
(7.1)
Here φ(j) = φ, j is the usual canonical pairing between the distribution
valued euclidean field operator φ (take from here on as a neutral scalar field
for simplicity of our exposition).
Since eq. (7.1) is a Bochner-Martin positive definite functional on
C(S(RD ), R) one can apply the Minlos Theorem to represent eq. (7.1) by
means of a probability measure on S  (RD )

Z(j) = dμ(φ) exp iφ(j). (7.2)

S  (RD )

For example, massive free scalar fields on RD given by the explicity

generating functional below (exponential of a continuous bilinear form on
S(RD ))
1
Z(j) = exp − T(−Δ+m2 )−1 j (j) . (7.3)
2
Here our T (continuous application) is explicited by the formula
T : (S(RD ), strong topology) → (S  (RD ), weak topology)
j → T(−Δ+m2 )−1 j (7.3-a)
where for f ∈ S(RD ), we have

T(−Δ+m2 )−1 j (f ) = ((−Δ + m2 )−1 j)(x)f (x)dD x. (7.3-b)

RD

As a consequence of above exposed, one simple application of Minlos-

Bochner Theorem to eqs. (7.3-a)/(7.3-b) give us the massive free scalar
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 135

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 135

euclidean generating functional

1
Z(j) = exp − T(−Δ+m2 )−1 j (j)
2
1
= exp − j, (−Δ + m2 )−1 jL2 (RD )
2

= d(−Δ+m2 )−1 μ(φ)(exp iφ(j)). (7.3-c)

Interacting quntum field theories as perturbation around massive scalar

fields, by theirs turn are also defined by an absolutely continuous mea-
sure dν(φ) in relation to previously euclidean massive free scalar field
d(−Δ+m2 )−1 μ(φ), which in a rigorous way would be writen as of as

d(ε) ν(φ) = (d(−Δ+m2 )−1 μ(φ))(e−gV (φ∗ρε ) ) (7.4)

where (φ ∗ ρε ) denotes the field sampling in eq. (7.3-c) in its regularized

form (which are C ∞ (RN ) functions).
Classifically gV (φ) is a non-linear function of the C ∞ -regularized dis-
tribution (φ ∗ ρε ) and g is the bare theory’s coupling constant.
Note that trying to formalize rigorously eq. (7.4) for ε → 0 is meaning-
less, since one expect that if these probabilistic distributional field configu-
rations were described by locally integrable functions, they would be infinite
almost everywhere on RD . It is no difficult to expect such behaviour for
scalar neutral free euclidean fields since (for D > 1)1

d(−Δ+m2 )−1 μ(φ)φ2 (x) = lim d(−Δ+m2 )−1 μ(φ)(φ(x)φ(y))

x→y
1 1
= dD k = +∞ (7.5)
(2π)D/2 k2 + m2

As a consequence it does not make sense to consider interactiong

euclidean QFT interactions directly from their classical counterpart La-
grangians as usually done in Physics textbooks. For instance, the functional
below it holds true the triviality relation on the mentioned field measure
space.
 
λ
exp − φ4 (x)dD x = 0 a.e. on (S  (RD ), d(−Δ+m2 )−1 μ(φ)).
4!
(7.6)
1 See Appendix B
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 136

136 Lecture Notes in Topics in Path Integrals

For interacting field theories as defined as perturbation such “triviality”

behavior is also expected.
In order to overcome such mathematical problems, it is used “regular-
ized” forms of the QFT, which in many cases, although useful for com-
putations, they shadow on the famous problems of infinities or the usual
Feynman-Dhyson perturbative scheme (intrinsic for defining interaction
field path measures around free field path integrals).
Let us now propose a new form of regularizing euclidean quantum field
theories by considering suitable regularization on the Kinetic free operator
eq. (3-b).
We have the following basic theorem of ours ([2]).

Theorem 1. Let us consider the α-power of the Laplacean operator acting

on H 2α (RD ). Let us consider also a compact domain Ω ⊂ RD and χΩ (τ )
its characteristic function
 $
1 x∈Ω
χΩ (x) =
0 x∈
/Ω
then one has:
1 – the operator L(α,Ω,m) here defined by its inverse which is given by an
integral operator with kernel
L−1 α 2 −1
α,Ω,m (x, y) = χΩ (x)[(−Δ) + m ] (x, y)χΩ (y) (7.7)
is such that it defines a probability measure on the functional space L2 (RD )
for the parameter range α > D/2
 
1 −1
Z(j) = exp − j, L(α,Ω,m) jL2 (RD )
2
 
= dL−1 (α,Ω,m) μ(φ) exp iφ, j . (7.8)
L2 (RD )

Proof: We note that L−1 α,Ω,m is a positive definite trace class operator on
2 D D
L (R ) for α > 2 (ref. ([2])). Since
 
−1 vol(Ω) dD k D
T rL2 (RD ) {L(α,Ω,m) } = D 2α 2
< ∞ if α > .
(2π) RD k +m 2
(7.9)
Now the result on the support of the probability measure given by
eq. (7.8) as given by the L2 (RD ) (or L2 (Ω)) space is a direct consequence
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 137

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 137

of the Minlos-Bochner theorem on integration theory on Hilbert Spaces

setting ([2] – appendix).

The point now in consider finite volume analitically regularized free

scalar QFT’s is that one can handle directly non-trivial interactions (super
renormalizable QFT’s) since the field condigurations are now the usual real
measurable point L2 (Ω) functions instead of L. Schwartz distributions.
For instance, we have the following theorem:

Theorem 2. The exponential cut-off euclidean P2 (ϕ) interaction (defined

explicitly below) is well defined on the (dL−1 μ(φ), L2 (Ω)) probability
(α,Ω,m)
D
functional space for α > 2
⎡ ⎤
⎢ 2k
λj j ⎥
VΩ (ϕ) = exp(−δϕ2 (x)) ⎢
⎣ 
ϕ (x)⎥ D
⎦d x (7.10)
Ω⊂R D
j=0
j
j=even

here δ > 0, λj ≥ 0, for j = 0, 2, 4, . . . , 2k.

Proof: An immediate consequence of the fact that if ϕ(x) = +∞ for

ϕ ∈ L2 (Ω) and for some point x ∈ Ω, then V (ϕ(x)) = 0, which means that
e−V (φ) = 1. As a result one has the upper bound for all
φ ∈ L2 (Ω) = supp dL−1 μ(ϕ) ⇒ exp{−VΩ (φ)} ≤ 1. (7.11)
(α,Ω,m)

Since (dL−1 μ(φ)) is a truly functional probability measure for α >

(α,Ω,m)
D
2, an application of the Lebesgue dominated theorem gives the finitude of
the associated generating functional eq. (7.8) for functional interactions of
the form eq. (7.10).

Note that remains a non trivial problem to evaluate the n-point field
correlation functions on this proposed scheme of ours. However the same
reasoning below can be applied to prove that the exponential regularized
n-point functions are finite. Namelly

(dL−1 μ(φ)) exp{−VΩ (φ)}

α,Ω,m
L2 (Ω)

(e−δφ φ)(x)j(x)dD x
2
× exp i
Ω
= Z(α,Ω,m2 ,δ) (j(x)) (7.12)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 138

138 Lecture Notes in Topics in Path Integrals

is an analytical functional on L2 (Ω).

Theorem 3. The analytic regularized, finite volume generating functional

Zα,Ω,m [j] is defined now on the Sobolev Space H0m (Ω) (now Ω denoting an
open set with compact closure) if α is sufficiently higher (2α − 2m − D > 0)
and for Drichlet conditions on the field configurations on the Ω-boundary
(φ|∂Ω ≡ 0).

On the light of such theorem, one can use the Sobolev immersion
Theorem H0m (Ω) ⊂ C P (Ω), for P < m − D/2, to have continuous or even
differentiable euclidean field sample configurations on the theory’s path in-
tegral for higher order free field path integral.
At this point appears the very difficult problem of cut-off remotions
vol(Ω) → ∞, α → 1, δ 2 → 0 or and m2 → 0 in these analitically “reg-
ularized” field thories. In next section one implement the finite volume
and the analitically regularized remotion on a class of non trivial massless
scalar field theories, just for exemplifying that cutt-off remotions on ours
proposed regularized path integrals are as possible as well.
Let us point out that theorem 3 can be considered as a sort of generalized
Wiener theorem on the continuity of Brownion motions for ours volume-
analitically regularized euclidean fields path integrals.
The proof of Theorem 3 is again a direct result of the Minlos-Bochner
theorem ([1], [2], [6]).
dL−1 μ(φ)||ϕ2 ||H m (Ω) < ∞ (7.13-a)
α,Ω,m
L2 (RD )
if
T rH m (Ω) [L−1
α,Ω,m ] < ∞. (7.13-b)
This can be verified by a direct computation
T rH m (Ω) [L−1
α,Ω,m ]
 
1 k 2m
= dD k dD u 2α | ˆΩ (k − U )|2 < ∞ (7.13-c)
I
vol(Ω) RD RD u + m2
if m + D < α.
The reader can check eq. (7.13-c) by means of the finitude condition of
the integral 2
K 2m
dD k 2α < ∞ if 2α − 2m − 2D > 0. (7.13-d)
K + m2
2 See also Appendix B for a discussion on the sample differentiability of the Euclidean

Field Theory Path Integraly on Hilbert Spaces.

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 139

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 139

7.3 On the cut-off remotion on a two-dimensional Euclidean

QFT model

We start our studies in this section by considering the (bare) euclidean

functional integral on a finite volume smooth compact region Ω ⊂ R2 as
given below (see ref. [2]) for α > 1, associated to a real scalar field on R2
$ '
1
Zα (j(x)) = (dL−1 μ)(ϕ)
Zα (0) L2 (Ω)
α

× exp −gbare V (ϕ(x))d2 x

R2

× exp i j(x)ϕ(x)d2 x . (7.14)

R2

Here the functional measure on the path space of real square integrable
function on Ω, denoted by L2 (Ω) is given through the Minlos Theorem (see
Section 1) for real field sources j(x) on L2 (R2 ) and the parameter α on
the range α > 1. Here m2 is a mass parameter eventually vanishing at the
end of our estimate, since we are only interested on the ultraviolet field
singularities
 
1 −α
exp − 2 2 2
d xd y j(x)(χΩ (x)((−Δ) + m )(x, y)χΩ (y))j(y)
2 R2 ×R2

= (dL−1
α
μ)(ϕ) exp(ij, ϕL2 (R2 ) ). (7.15)
L2 (Ω)

The interaction potential is a continuous function vanishing at infinite

such that it posseses an essential bounded L1 (R) Fourier Transform (for
λ −δx2 4
instance V (x) = 4! e x , etc...). In others words:

||V ||L∞ (Ω) ≤ ||V< ||L∞ (R) < ∞. (7.16)

Let us show that by defining the bare coupling constant by the renor-
malization prescriptions (v = vol(Ω))
grem
gbare (α, v) = (7.17)
(1 − α)1/2 v
The functional integral eq. (7.1) has a finite limite at α → 1, v → ∞, when
understood in the R.P. Feynman sense as an expansion perturbative on the
renormalized constant grem . Namely (see Chapter 5, §5.2, eq. (5.11), [7])

Zα=1 (j(x)) = lim lim lim lim IN (gbare (α, v), [j]) . (7.18)
V →∞ N →∞ α→1 m2 →0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 140

140 Lecture Notes in Topics in Path Integrals

Theorem 1. The functional IN (gbare (α, v), [j]) satisfies the upper bound
at the limit α → 1
/ 0
lim lim |IN,α (gbare(α, v), [j]|)
N →∞ α→1

≤ C exp(C)
2
(7.19-a)
where the constant C is given by
C = (4π)1/2 gren ||V< ||L∞ (R) . (7.19-b)


Proof: By noting that C(Ω) (dL−1 α
μ)(ϕ) exp(ikϕ(x)) = 0, one has the fol-
lowing result eq. (7.8) (see [2]), where the integral kernel of the our “free”
propagator is given explicitly by (for α > 1; see Appendix A)
 
−1 1 Γ(1 − α) 2(1−α)
Lα,m2 =0 (xi , xj ) = χΩ (x) (|xi − xj | 2(α−1)
) 2 χΩ (y).
(2π) Γ(α)

IN,α (gbare (α, v), [j])

N >
 (−1)n
= (gbare (α, v))n d2 x1 . . . d2 xn
n=0
n! Ω

dk1 dkn <

× ··· (V (k1 ) · · · V< (kn ))
R (2π)1/2 R (2π) 1/2
 n
; 
× (dL−1
α
μ)(ϕ) exp ik ϕ(x )
C(Ω) =1
?
× exp i j(x)ϕ(x)d2 x (7.20)
Ω

As a consequence of the positiviness of the “kinetic” Green function

L−1
α (x, y); one has

(IN,α (gbare (α, v), [j]) ≤ 1

> N
(|(gbare (α, v))|)n <
+ (||V ||L∞ (Ω) )n d2 x1 . . . d2 xn
n=1
n! Ω
; −1/2 ?
× det [L−1 α,m 2 (xi , xj )] . (7.21)
N ×N

Note that due to the continuity on the infrared cut-off mass parameter,
it is possible to consider directly its limit on the determinant formed by the
Green’s functions.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 141

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 141

At this point we note the Taylor expansion of the below written object
−1/2
LN (α, v) = d2 x1 · · · d2 xn det [L−1
α (xi , xj )]
Ω N ×N1≤i≤N
1≤j≤n
N
= (1 − α)N/2 Cn + (1 − α) 2 +m1 CN +1 + . . . (7.22)
with
⎛ ⎞−1/2

Cn = v n (4π)N/2 ⎝ det [Aij ]⎠ (7.23-a)

1≤i≤N
1≤j≤N

and the matrix [Ai,j ] is defined by the rule

$
0 if i = j
[A[ij = (7.23-b)
1 if i = j
It yield thus ([2], [3]) for N > 1
/
lim lim |(gbare (α, v)LN (α, v))|
V →∞ α→1
%1 1&
> 1 N 1
1 gren 1
= lim lim 1 LN (α, v)1
V →∞ α→1 1 (1 − α)v 1
/ > |(gren )| N
= lim lim
V →∞ α→1 v |(1 − α)|N/2
N
; ?0
× |(1 − α)|N/2 V N (4π)N/2 (|(−1)(N − 1)(−1)N |)−1/2
1 0
= (|gren |)N ((4π) 1/2 N
) (7.24)
(N − 1)1/2
We have thus the uniform bound on the “interaction order” N in our
Euclidean QFT model
lim lim (|IN,α (gbare (α, v), [j])|)
V →∞ α→1
> M
(4π)N/2 (|gren |)N
≤1+
N ! (N − 1)1/2
N =2
?
× [||V< ||L∞ (Ω) ]N
 ∞
 CN
≤C 2
N!
N =0
2
= C exp(C) (7.25)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 142

142 Lecture Notes in Topics in Path Integrals

with
C = |gren | · ||V< ||L∞ (4π)1/2 . (7.26)
Note that we have used the elementary estimate to arrive at eq. (7.25)
for N > 1
1
≤ 1, for N ≥ 2. (7.27)
(N − 1)1/2
We conclude this, that the functional path integral eq. (7.14) under the
renormalization coupling constant eq. (7.17) and rigorous Feynman pertur-
bative definition eq. (7.18) has a finite limit for α = 1.
It is worth that one could also consider the most general multiplicative
renormalization including the functional form of the interaction
gren
gbare (α, v||V< ||L∞ (Ω) ) = . (7.28)
(1 − α) v||V< ||L∞ (Ω)
1/2

Now allowing interactions satisfying the constraint V< (k) = lim V< (k)
→∞
with ||V< (k)||L∞ (Ω) =  ∈ N+ .
It is worth to recall that we have proven that the full generating func-
tional eq. (7.14) at α → 1 as defined by a Feynman’s perturbative series:
Feynman’s diagrammas renormalized order by order in a power serie expan-
sion on the bare coupling constant is finite and it is a continuous functional
on the source space j(x) ∈ L2 (Ω).
However it appear that the use of the propagator prescription
/; 1 Γ(1 − α)
L̂−1 2
α,m =0 (xi , xj ) = χ Ω (x) 22(1−α)
4π Γ(α)
 4π 0
× |x − y|2(α−1) − χΩ (y), (7.29)
(1 − α)
which converges on the D (R2 ) L. Schwartz distributional sense to the usual
non positive definite two-dimensional Laplacean Green function for α → 1
does not lead to well defined Euclidean QFT generating functional. A
result already expected since Massless 2D Euclidean Q.F.T. Theories built
already as perturbation around free scalar Massless fields on R2 do not make
mathematical sense due to the fact that the two-dimensional Laplacean
Green function does not belongs to the “Fourier Transformable” Tempered
Distributional Space S  (R2 ), a fact already observed a long time ago by
S. Coleman ([4]) and fully used by G. Hoft on his studies on (QCD)2 –
solubility at large number colors ([5]).
Another point worth call attention in this Section is that the same proof
works out for a class of four-dimensional analitically regularized Euclidean
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 143

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 143

Field theories with the “Free kinetic operator” defined through the Minlos’s
theorem on a finite volume region Ω ⊂ R4
> 1
exp − d4 x d4 y
2 R4 R4
> ;  ?
× j(x) χΩ (x) (−Δ2 )−α + m20 χΩ (y) j(y)
/ 0?
= dL−1 μ(ϕ) exp i ϕ(x)j(x)d4 x . (7.30-a)
α,m2
0 R4
Here the Integral Kernel of the square D-fimensional Laplacean is given
by (for α > 1)
2 − 2α)
Γ( D
(−Δ2 )−α = (|x − y|4α−D ). (7.30-b)
Γ(2α)24α π D/2
Finally, we call attention that into another publication we will address
to the “differentiability” of the generating functional eq. (7.1) at α → 1
as defined in the Bulk of this section. However it is straithforward to
obtain such differentiability for sources j(x) coupled to field configurations
interaction of the form exp(−δϕ2 (x))ϕ(x). Note that in this case, the N -
point Taylor’s coeficients of Z[j(x)] are explicitly given by
δ N Z[j(x)] 1
1
1
δj(x1 ) · · · δj(xN ) j(x)≡0
N
= (dL−1
α
μ)(ϕ) [exp(−δϕ2 (x ))ϕ(x )]
C(Ω) =1

< ∞, (7.31)
since the domain of the above functional integral for α > 1 is the space of
measurable square integrable functions on Ω (and for δ > 0)
||e−δϕ ϕ(x)||L∞ (R) = (max |e−δx x|) = C < ∞,
2 2
(x)
(7.32)
x∈R

leading to the finitiness of eq. (7.31) by the use of the Lebesgue dominated
convergence theorem.
The limite of δ → 0 on the momentums eq. (7.31) will appears elsewhere.
Finally we wishe to point out that non trivial homological topology of
the compact planar two-dimensional domain Ω ([8]) in ours path integral
can be easily taken into account by the Ω set indicator function χΩ (Ω)
on eq. (7.30-a) of this section, specially on Fourier Space by means of the
Ω-domain Fourier Integral form factor for Ω with holes inside

Iˆk (Ω) = d2 ξ exp(ikξ) ; (7.33)

Ω
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 144

144 Lecture Notes in Topics in Path Integrals

which appears on the expression of the theory’s propagator on momentum

space for general RD space-time
IˆΩ (k − p)IˆΩ (p − k  )
L̂−1 (k, k  ) = dD p (7.34)
RD p2α + m2
and leading thus to the Feynman diagrammotic generating functional on
the Fourier Space
Z[j̃(k)]/Z[0]
 
1 δ
= exp − dD x V d D
ke +ikx
Ω (2π)D/2 RD δ j̃(k)
> 1
× exp − dkdk  j̃(k)
2 R2D

ˆ − p)I(p
I(k ˆ − k ) ?

× dp j̃(k ) (7.35)
RD p2α + m2
As a last remark, we conjecture that the ultra-violet limit α → 1 on the
usual correlations functions associated to our path integral should expected
to be finite. The argument follows by considering ||V< ||L∞ (Ω) = 1, since
x → 0 and thus, obtaining the general structure of the (for instance) two-
point function at perturbative order N
ϕ(x1 )ϕ(x2 ) ∼ −(L−1
α→1
α,m2 =0 (x1 , x2 ))


N >
+ L−1
α,m2 =0 (x1 , xp )
p=1,q=1
?
−1 −1
× [L−1
α (xi , xj )] L
pq α,m =0
2 (xq , x2 ) (7.36)
and noting the Laplace formula for evaluate the inverse of the propagator
matrix
1
[L−1 −1
α (xi , xj )]pq =
detN ×N [L−1
α (xi , xj )]

× (C(xi , xj )]qp , (7.37)

with the cofactor matrix [C(xi , xj )]qp associated to the [L−1
α (xi , xj )] prop-
agator matrix eq. (7.21). One expects thus that the singular behavior for
α → 1 of the determinant
det [L−1 −N
α→1
α (xi , xj )] ∼ (1 − α) ,
N ×N
cancels out with the factor
N
L−1 −1
α,m2 (x1 , xp )[C(xi , xj )]qp Lα,m2 =0 (xq , x2 ) ∼ (1 − α)
−N
,
p=1,ε=1
on eq. (7.36).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 145

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 145

7.4 On the construction of the Wiener Measure

On next Section 4, we intend to analyze the somewhat different functional-

path integral on functional space, mainly due R.P. Feynmann and M. Kac:
the so called geometrodynamical end points fixed field propagator.
However, such objects to be defined mathematically, one must review
the construction of the famous Wiener path measure ([6]–[7]). This is our
objective in this short section.
Let us first introduce some notations and mathematical objects.
We first write the time fixed Heat equation green function on the one-
point compactified of the real line R, the interval [− π2 , π2 ] as a integral
kernel of a continuous linear functional on the compact support continuous
function f on R. So, let ε > 0 and f ∈ Cc (R)
exp(−|x − y|2 /2ε)
Lε (f ) = f (y)dy
R (2πε)1/2
:=Ĝ0 (x,y,ε)
+ ,- .
π
−|tg(θx )−tg(θy )|2 /2ε
2 e
= f (tgθ)dθ. (7.38)
−π
2
(2πε)1/2 cos2 θ
Note that supp f (tgθ) ⊂ (− π2 , − π2 ). For a given g(θ) ∈ Cc ([− π2 , π2 ]),
eq. (7.38) defines a positive continuous linear functional on C0 (Ṙ)(f (∞) =
g(− π2 ) = g( π2 ) = 0).
Let us define the following projective family a positive linear functionals
B
∞
on Cc ( ([− π2 , π2 ])n ), firstly defined on the σ-algebra of the infinite variable
n=0
B
∞
space Cc ( Ṙ) ≡ C0 ((Ṙ)∞ ). For N a given positive integer fixed, but
n<0
π π
arbitrary and x̄ a fixed point on [− 21,2 ], we consider the projected positive
continuous linear functionals ε = N ;
0
(N )
Lx̄ (f (x1 , . . . , xN )) = dx1 . . . dxN (f (x1 , . . . , xN ))
Ṙ Ṙ
× (Ĝ0 (x, x1 , ε) . . . Ĝ0 (xN −1 , xN , ε))dx1 . . . dxN .
(7.39)
We point out the “projective” properties of the family of positive con-
(N )
tinuous functionals {Lx̄ }:

a) For M ≤ N ,
L(M)
x (f (x1 , . . . xM )) = L(N )
x (f (x1 , . . . , xM )) (7.40)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 146

146 Lecture Notes in Topics in Path Integrals

b) Cc,finite variables (Ṙ∞ ) = {f ∈ Cc (Ṙ∞ ), but with finite variables} is

a dense subset of Cc (Ṙ∞ ) (endowed with the usual supremum norm!).
All theses results above remarked show that exists the
on Cc (Ṙ∞ ).
(N ) (∞)
lim Lx̄ = Lx̄
N →∞

Since Ṙ∞ is a compact topological space one can apply the Riesz Markov
theorem to represent L∞
x̄ through a well defined measure on Ṙ
∞
(the Bare
∞
σ-albegra of Ṙ ).
Note that one could take ε = t/N with t > 0, a real fixed, and this
obtain the famous Wiener measure ending at x̄ at time t

L∞
(x̄,t) (f ) = dWiener
(x̄,t) μ̂[g(σ)]f (g(σ)) (7.41)
Ṙ∞

where g(σ) ∈ Ṙ∞ is identified with the set of all real functions on Ṙ [the
“compactified” Wiener path trajectory], with the domain σ ∈ [0, t].
It is worth to remark that on eq. (7.39), all the “time parameters” are
at the same value t = ε.
It is an open problem to show the existence and unicity of the Wiener
measure d(x̄,t) μ[g(σ)] under general (different) time steps on eq. (7.39).
It is worth also to note that f ∈ Cc (Ṙ∞ , R) by the hypothesis of the
Riesz-Markos theorem ([7]).
At this point if is argued that there is a unique “pull-back” of the above
constructed Wiener measure on the space of compact paths to the full R
paths. Namely, for F ∈ Cc (C(R, R), R) and x ∈ R

L∞
(x,t) (F ) = dWiener
(x,t) μ(X(σ))F (X(σ)). (7.42)

The above construction generalizes straightforwardly for RD (D > 1).

We have thus the following theorem (Feynmann Wiener-Kac): Let (−Δ)
be the essential self-adjoint extension of the usual Laplacean acting on
Cc∞ (RD ).
We have the formula (7.7), for F ∈ Cc (RD ) on the sense the topology
of Cc (RD )
(e− 2 Δ F )(x)
t

= (dWiener
(x,t) μ(X(σ))F (X(t)). (7.43)

For general F ∈ L2 (RD ), eq. (7.43) is obtained by (unique) extension,

since ((Cc (RD ))L2 (RD ) = L2 (RD )).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 147

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 147

It is important to call attention that due to the C ∞ -regularizing prop-

erty of the heat Kernel eq. (7.38), the functional integral representation
eq. (7.43) remains correct in the L2 (RD ) sense for F (w) = δ (D) (w − y) and
leading thus to the formal Brownian Bridge path integral measure repre-
sentation for the Heat Kernel
X(t)=y
L2 (RD )
x|e− 2 Δ F |y
t
= dWiener μ(X(σ))F (X(t)) (7.44-a)
X(0)=y

or in the correct mathematical meaning of the above written eq. (7.44-a)

for f and g ∈ L2 (RD )

X(t)=y
f (x)x|e− 2 Δ F |yḡ(y) =
t
dWiener μ(X(σ))(f (X(t))ḡ(X(0))).
RD X(0)=y
(7.44-b)

7.5 On the Geometrodynamical Path Integral

Sometimes it appears to be useful for calculational purposes on euclidean

quantum field theory to give a generalized meaning for the Brownian Bridge
Wiener path integral eqs. (7.44-a)–(7.44-b), called now the Geometridy-
namical propagator connecting a classically observed field configuration
φ(x, t1 ) = βin (x) to another final one φ(x, t2 ) = βout (x), with t2 > t1 .
Let us formulate the problem for the free case of a real scalar field
φ(x, t) with classical action and with Dirichlet boundary conditions on the
D-dimensional space-time cylinder manifold propagation D = Ω × [t1 , t2 ]
with A denoting an inversible positive definite self-adjoint elliptic operator
on Ω

1 ∂2
S[D] = β − + A β (x, t)dD−1 xdt. (7.45)
D 2 ∂t2

One wants to give a rigorous mathematical meaning for the Euclidean

Feynman-Wheller path integral

G[(βin (x), t1 ); (βout (x), t2 )]

β(x,t2 )=βout (x)
= DF [β(x, t)] exp{−S[D]}. (7.46)
β(y,t1 )=βin (x)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 148

148 Lecture Notes in Topics in Path Integrals

The most usual way to give a mathematical meaning for eq. (7.46) is
to use the spectral theorem for A (Aϕμ = λμ ϕμ ) and regard eq. (7.46)
as the (enumerable) infinite product of Brownian Bridge Wiener mea-
sures eq. (7.44-a) and under the hypothesis that all the field configurations
entering on the support of the resulting field path integral measure is of the
form

∞

β(x, t) = Cμ (t)φμ (s) ∈ C([t1 , t2 ], L2 (Ω)). (7.47)
μ=0

We thus define eq. (7.46) as

G[(βin (x), t1 ), (βout (x), t2 )]

∞
% out
&
Xn (t2 )=βn t2
1
= d Wiener
μ(Xn (σ)) × exp − 2 2
(λn ) (Xn (σ)) .
n=0
in
Xn (t1 )=βn 2 t1
(7.48)

Here

∞

in
β (x) = βnin ϕn (x) (7.49-a)
n=0

∞

β out (x) = βnout ϕn (x). (7.49-b)
n=0

Let us note that the enumerable infinite product of Wiener-Harmonic

Oscilator measure is still a well behaved σ-measure on the product measure
B
∞
space (C([t1 , t2 ], R))n . Note that if one user the compactified of the real
n=0
line as in Section 3, one would gets as the measure space; the compact space
B
∞
C([t1 , t2 ], Ṙ)n .
n=0
In the presence of an external source f (t, x) ∈ C([0, T ], L2 (Ω)), one has
the usual Feynman closed expression in terms of Feynman-Wiener notation
for the Wiener-Harmonic oscillator path measures ([7])
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 149

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 149

G[(β in (x), 0); (β out (x), T ), [j(x, t)]]

∞ out
$ % &'
Xn (T )=βn T
1 d2
= DF [Xn (σ)]exp − Xn − 2 + λ2n Xk (σ)
n=0
in
Xn (0)=βn 2 0 dσ

T
× exp dσjn (σ)Xn (σ)
0

∞
$G
λn
=
n=0
Sinh(λn T )

λn  out 2
× exp − (βn )
2sinh(λn T )
?
+ (βnin )2 cosh(λn T ) − 2βnout βnin

T
2βnout
− dσjn (σ)sinh(λn σ)
λn 0
T
2βnin
− dσjn (σ)sinh(λn (T − σ))
λn 0
T T
2
− dσ dσ  jn (σ)jn (σ  )sinh(λn (T − σ))
(λn )2 0 0
?
× sinh(λn (σ  )) (7.50)

Another more attractive prescription to eq. (7.46), specially useful on

String Theory ([7]) is to suppose that the sample space for geometrody-
namical propagation is composed of field configurations made by random
perturbations of a (fixed) classical field configuration as exposed in §5.3,
eqs. (7.41)–(7.47) of ref. [7]. However, this method does not appears to be
canonically invariant, since all the objects on the theory are dependent of
the choosen background field configuration the classical choose field config-
uration, i.e. for different background field configuration one could obtain
different path integrals.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 150

150 Lecture Notes in Topics in Path Integrals

7.6 Appendix A

Let us recall the following integral form of a Fourier Transform of a Tem-

pered distribution Tf defined by a L2loc (RD ) radial function f (r)
F (Tf (r) ) = TF̂ (k) (A-1)
with 
∞ f (r)rD−1 J D−2 (kr)dr
D/2 2
F̂ (k) = (2π) . (A-2)
(kr) 2 −1
N
0
By using our proposed distributional sense integral formulae for μ and
ν complex numbers and a > 0
∞
Γ( 1 + 1 ν + 12 μ)
xμ Jν (ax)dx = 2μ a−μ−1 21 21 . (A-3)
0 Γ( 2 + 2 ν − 12 μ)
One obtains the result on the S  (RD ) sense for α ∈ C
F T( 1 Γ(1−α) 2(j−α) 2(α−1)
2π Γ(α) 2 r

= TK −2α . (A-4)
The complete distributional sense is given below for f (x) ∈ S(R ), with
D

fˆ(k) = F [f (x)]
T( 1 Γ(1−α) 22(1−α) r2(α−1) (f ) = Tk−2α (fˆ(k)). (A-5)
2π Γ(α)

Just for completeness, let us evaluate on the S  (R2 ) sense the Fourier
Transform below
2
1 1
Gα (x, y, m2 ) = √ d2 keik(x−y) 2
2π R2 (k + m2 )α
⎛ ⎞
2J0 (kr)
1 ⎜ ∞ + ,- .
k ⎟
= ⎝ dk 2 2 )α
(J0 (kr) + J0 (−kr))⎠
2π 0 (k + m
∞
1 2(α−1) pJ0 (p)
= r × dp 2
2π 0 (p + m2 r2 )α
(mr) 1−α
· K1−α (mr)
= (A-6)
2α−1 Γ(α)
where we have used the distributional sense integral relation for μ and ν
complex parameters and a, b ≥ 0:
∞
JV (bx)xν+1 aν−μ bμ Kν−μ (ab)
dx = . (A-7)
0 (x2 + a2 )μ+1 2μ μ(μ + 1)
Just for completeness, one can use the above exposed formulae to obtain
the Integral Kernel of the S  (RD ) distribution (−Δ)−α . Namely
2 − α)
Γ( D
(−Δ)−α (x, y) = . (A-8)
Γ(α)2 π D/2 (x − y)D−2α
2α
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 151

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 151

7.7 Appendix B

Theorem. Let A be a positive definite trace class operator on L2 (Ω), with

@
∞
spectral resolution Aϕn = λn ϕn such that λn n2p = +∞ for p > 0.
n=0
Let also dA μ(ϕ) denotes the cylindrical measure asociated to A through
the Minlos-Bochner theorem applied to the bilinear source function Z(j) =
exp{− 21 j, Aj L2 (Ω) }.
@ @
∞
Let Hp = {f ∈ L2 (Ω) | f = cn ϕn with c2n n2p < ∞} be the
n=0
“Sobolev” sequence measurable sub-sets of L2 (Ω). Then we have that for
any p > 0
μ(Hp ) = 0.
Roughly this result means that C ∞ (Ω)-smooth path integral field sam-
ple configurations on the path probability space (L2 (Ω), dA μ(ϕ)) make a
set of zero measure Proof ([6]).
By using the following representation of the charachteristic function of
the sub-set Hp
$ % ∞
&'
α  2 2p
χHp (ϕ) = lim exp − c n , (B-1)
α→0 2 n=0 n
L2 (Ω) @
we have the identity (ϕ ∈ L2 (Ω); ϕ = cn ϕn )
μ(Hp ) = lim dj1 . . . djN
α=0 RN
 

N
× dA μ(ϕ) exp i jn cn
n=0
 ∞
1  jn2
× exp −
2α n=0 n2p
% N
&
(2παn2p )− 2
1
× (B-2)
μ=1
A firect evaluation of the cylinder path integration on the right-hand
side of eq. (B-4) give us the following outcome
 ∞

dA μ(ϕ) exp i jn cn
μ=0
$ ∞
'
1  jn2
= (2π) N/2
exp − . (B-3)
2 μ=0 λn
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 152

152 Lecture Notes in Topics in Path Integrals

As a consequence we have the final result on the measure of the “Sobolev

Spaces” Hp
N
>
(2παμ2p )− 2
1
N/2
μ(Hp ) = lim lim (2π
α→0 N →∞
μ=1
 N
1 2 ? −1
× λn +
n=1
αn2p
$  '
1 
N
≤ lim lim exp − α λn n2p
α→0 N →∞ 2 μ=1
$
0 if p > 0
= (Minlos-Bochner Theorem) (B-4)
1 if p = 0
where we have used the straightforward identity
 1/2 $ '
1
N 1/2 N
1 1
≤ @N ≤ exp − αN .
1 + αn 1+ αn 2

I
∞
As it is usual to expect that C ∞ (Ω) ⊂ H p (Ω) (where H p (Ω) denotes
p=0
the usual function Sobolev Spaces on Ω), the theorem of this appendix as
expressing the fact that differentiable sample on non enough sufficiently
regularized euclidean path integrals makes a set of zero measure. And clas-
sical smooth field configurations being useful only in the realm μ of formal
saddle-point (WKB) path integral evaluations. So one can not manipulate
path integral integrands with College Calculus rules.
For the less restringent condition of path integral sample continuity, one
has to use our generalization of the Wiener theorem eq. (7.9).
In the general case of non-Gaussian cylindrical measures, one should
imposes the bound restriction below, as a reasonable thechnical condition
  
1
sup |Z(j)| ≤ C exp − j, Aj (B-5)
2
J+
for some positive definite trace class positive operator A ∈ 1 (L2 (Ω)) and
C > 0, in order to obtain the validity of our theorem – Appendix B.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 153

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 153

7.8 Appendix C

In this somewhat pedagogical appendix, we intend to presente a formal

operational functional calculus to write the cylindrical Fourier Transforms
as an inversible operation. We present such formal results in order to
highlight the necessity of a clean distribution theory in Hilbert Spaces, get
to be developed ([7]).
So let us consider a trace class, inversible and strictly positive operator
−1
A acting on a separable Hilbert Space H.
Let f (x) ∈ L1 (dA μ(x)(x), H) ∩ L∞ (dA μ(x), H). Since the function
exp ix, k, for a k ∈ H fixed is bounded, the L1 -Hilbert Space Fourier
Transform is well defined
F̂ (k) = f (x) exp ix, kdA μ(x). (C-1)
H
In the Physicist’s operational notation for the cylindrical measure with
DF [x] denoting the Feynman Formal (when H realized as some L2 (Ω))
+ 12
1
dA μ(x) = det(A) exp(− (x, Ax)H )DF [x] (C-2)
2
It still to be an open problem in Analysis in Infinite Dimensions or
and Hida calculus to obtain an rigorous mathematical inversion formula
for eq. (C-1). However, it is fully possible to write an inversion formula
for eq. (C-1) in a more larger vectorial space: the called algebraic dual
space of H i.e: H alg . Let us sketchy for completeness such result of
ours. As a first step one introduces an one-parameter ε (ε ∈ [0, 1]) fam-
ily of auxiliary inversible operators ε−2 C(ε) on Lbounded(H, H) such that
C(0) = 1. Let us now consider the new (perturbed) family of operator
B(ε) = (A + ε−2 C(ε))−1 . Note that B(ε) exists for ε small enough and
B −1 (ε) = A + ε2 (C(ε)) do not belong to the trace class althought being
positive definite. But even in such situation one can define a cylindri-
cal measure in the more ample space H alg through the positive-definite
charachteristic functional associated to the operator B −1 (ε)
 
1
ZB [x] = dB(ε) ν(X)eiX(x) = exp − x, (B(ε))−1 xH . (C-3)
H alg 2
Let us choose our interpolating family of bounded, strictly positive op-
erators C(ε) such that for ε > 0, det− 2 (C(ε)) < ∞. We now define the
1

following (continuous) linear functional on H for each ε > 0, under the

non-proved hypothesis that H is a sub-set of H alg of non zero measure
− 12
 
1 −ik(x)
I (x, [F̂ ]) = det( 2 C(ε))
(ε)
F̂ (k)e dB(ε) ν(k) . (C-4)
ε H alg ∩H
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 154

154 Lecture Notes in Topics in Path Integrals

Now if one substitutes eq. (C-4) into eq. (C-1) and by applying the
Fubbini theorem to the Product Measure Space (H × H alg ; dA μ ⊗ dB(ε) ν),
one obtains the result
− 12
1
I (ε)
(x) = det( C(ε))
ε2
>   − 12
−i(x,k)
× f (z)e iz,k
dA μ(z) e dB(ε) ν(k) = det(ε−2 C(ε))
H alg H
 
× f (z)e iz−x,k
(dA μ(z)dB(ε) ν(k))
H alg H
− 12
 
= det(ε−2 (I(ε))) × f (z)ZB(ε) (z − x)dA μ(z)
H
− 12 (z−x,A(z−x))H
= f (z)e
H
% &
− 12
1
× det(ε−2 C(ε)) exp − 2 z − x; C(ε)(z − x) dA μ(z).
2ε H
(C-5)
At this point we take from the Hida Calculus, the formal definition of
the Dirac delta functional on Hilbert Spaces
− 12
 
1
lim+ det(ε−2 C(ε)) exp − 2 (z − x), C(ε)|z − x|H
ε→0 2ε
S  (H).
(F )
= δH (z − x) on (C-6)

As a consequence, one has the operational result

I(x) := lim I (ε) (x)

ε→0+
1
= f (z) exp − z − x, A(z − x)H
H 2
(F )
× δH (z − x)dA μ(z)
+ 12
:= f (x)e− 2 x,Ax det(A).
1
(C-7)

As a consequence we have the operational formulae for Fourier Trans-

forms in separable Hilbert Spaces.
If one has the usual cylindrical Hilbert Space Transform

F̂ (k) := f (x) exp(ix, kH )dA μ(x) (C-8)

H
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 155

Infinities on Quantum Field Theory: A FunctionalIntegral Approach 155

then formally, one has the “inversion” formula on the algebraic dual of H
− 12
f (x) = e+ 2 x,Ax det(A)
1

× lim I (ε) (x, [F̂ ]). (C-9)

ε>0
Anyway the necessity of using mathematically rigorous infinite-
dimensional Fourier Transforms on Tempered Schwartz Distributions has
not appeared fully yet on mathematical physics metrods. However, on light
of the results presented in this paper, the time for such endoavours may be
approaching ([7]).

Acknowledgments: Thanks to professor D. Pickrell of Mathematics

Department of University of Arizona for discussions on P (φ)2 Field Theo-
ries on Riemman Surfaces (ref[8]).

7.9 References

[1] J. Glimm and A. Jaffe, Quantum Physics, Springer, New Yorkl,

NY, USA, 2nd edition, 1987.
- B. Simon, The P (φ)2 Euclidean (Quantum) Field Theory, Prince-
ton University Press, Princeton, NJ, USA, 1974.
[2] Luiz C.L. Botelho, Some Comments onj Rigorous Finite-Volume
Euclidean Quantum Field Path Integrals in the Analytical Regu-
larization Scheme – Hindawi Publishing Corporation, Advances in
Mathematical Physics, vol 2011, Article ID 257916,
DOI: 10.1155/2011/257916.
[3] Luiz C.L. Botelho, “A simple renormalization scheme in random
surface theory”, Modern Physics Letters B, vol 13, No. 6–7, pp.
203–207, 1999.
[4] Green, M.R., Schwarz, J.L., Witten, E., Superstring Theory, Cam-
bridge Monographs on Mathematical Physics, vol 182, CUP, Cam-
bridge (1996).
[5] B. Klaiber, in Lectures in Theoretical Physics: Quantum Theory
and Statistical Theory, edited by A. O. Barut. Gordon and Breach,
New York, 1960, vol XA, pp. 141–176.
[6] Luiz C.L. Botelho, A note an Feynman Kac path integral represen-
tations for scalar wave motions, Random Operators and Stochastic
Equations (print), v. 21, pp. 271–292, (2013).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 156

156 Lecture Notes in Topics in Path Integrals

- Luiz C.L. Botelho, Semi-linear diffusion in RD and in Hilbert

Spaces, a Feynman-Wiener path integral study, Random Oper.
Stoch. Equ-19 (2011), Issue 4, pages 361–386, DOI 10.1515/Rose
2011.020.
- Luiz C.L. Botelho, A method of integration for wave equation and
some applications to wave physics, Random Oper. Stoch. Equ-18
(2010), No. 4, 301–325.
- Luiz C.L. Botelho, Non-Linear Diffusion and Wave Damped
Propagation: Weak Solutions and Statistical Turbulence Behavior,
Journal of Advanced Mathematics and Applications, vol 3, 1–11,
(2014).
[7] Luiz C.L. Botelho, Lecture Notes in Applied Differential Equations
of Mathematical Physics World Scientific, (2008), Singapore ISBN:
10981-281-457-4.
[8] Pickrell, Doug, P (φ)2 Quantum Field Theories and Segal’s Axioms.
Commun. Math. Phys. 280, 403–425, (2008).
[9] Luiz C.L. Botelho, On the rigorous ergodic theorem for a class of
non-linear Klein Gordon wave propagations, Random Oper. Stoch.
Equ. (March 2015), vol 23, Issue 1 DOI:10.1515/rose-2014-0029.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 157

Chapter 8

Some comments on rigorous

finite-volume euclidean quantum field
path integrals in the analytical
regularization scheme

Through the systematic use of the Minlos theorem on the support of cylin-
drical measures on R∞ , we produce several mathematically rigorous finite-
volume euclidean path integrals in interacting euclidean quantum fields
with Gaussian free measures defined by generalized powers of finite-volume
Laplacean operator.

8.1 Introduction

Since the result of R.P. Feynman on representing the initial value solu-
tion of Schrodinger Equation by means of an analytically time continued
integration on an infinite - dimensional space of functions, the subject of
Euclidean Functional Integrals representations for Quantum Systems has
became the mathematical - operational framework to analyze Quantum
Phenomena and stochastic systems as showed in the previous decades of
research on Theoretical Physics ([1]–[3]).
One of the most important open problem in the mathematical the-
ory of Euclidean Functional Integrals is that related to implementation
of sound mathematical approximations to these Infinite-Dimensional Inte-
grals by means of Finite-Dimensional approximations outside of the always
used [computer oriented] Space-Time Lattice approximations (see [2], [3] -
chap. 9). As a first step to tackle upon the above cited problem it will be
needed to characterize mathematically the Functional Domain where these
Functional Integrals are defined.
The purpose of this note is to present the formulation of Euclidean
Quantum Field theories as Functional Fourier Transforms by means of the
Bochner-Martin-Kolmogorov theorem for Topological Vector Spaces ([4],

157
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 158

158 Lecture Notes in Topics in Path Integrals

[5] - theorem 4.35) and suitable to define and analyze rigorously Functional
Integrals by means of the well-known Minlos theorem ([5] - theorem 4.312
and [6] - part 2) which is presented in full in Appendix A.
We thus present studies on the difficult problem of defining rigorously
infinite-dimensional quantum field path integrals in general finite volume
space times Ω ⊂ Rν (ν = 2, 4, . . . ) by means of the analytical regularization
scheme ([12]).

8.2 Some rigorous finite-volume quantum field path inte-

gral in the Analytical regularization scheme

Let us thus start our analysis by considering the Gaussian measure on

L2 (R2 ) defined by the finite volume, infrared regularized and α-power
Laplacean acting on L2 (RN ) as an operatorial quadratic form (j(x) ∈
L2 (RN )) (see Appendix B)
 
1C −α 2 −1
D
(0)
Zα,ε [j] = exp − j, (χΩ [(−Δ) + εIR ] χΩ )j
IR
2 L2 (R2 )
C D
≡ d(0)
α,εIR μ [ϕ] exp i j, ϕ 2 2
(8.1-a)
L2 (R2 ) L (R )

Here χΩ denotes the multiplication operator defined by the characteris-

tic function χΩ (α) of the compact region Ω ⊂ R2 and εIR > 0 the associated
infrared cut-off.
It is worth calling the reader attention that due to the infrared regular-
ization introduced on Eq. (8.1-a), the domain of the Gaussian measure ([4],
[6]) is given by the space of square integrable functions on R2 by the Minlos
theorem of Appendix A, since for α > 1, the operator defines a trace class
operator on L2 (R2 ), namely
 
d2 p
T r1 (L2 (R2 )) {χΩ [(−Δ)α + ε2IR ]χΩ } = vol(Ω) × <∞
|p|2α + ε2IR
(8.1-b)
This is the only point of our analysis where it is needed to consider the
infra-red cut off. As a consequence of the above remarks, one can analize
the ultra-violet renormalization program in the following interacting model
proposed by us and defined by an interaction gbare V (ϕ(x)), with V (x)
being the Fourier Transformed of an integrable and essentially bounded
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 159

Some comments on rigorous finite-volume euclidean quantum field path integrals 159

measurable real function1 . Note that V (x) is thus a continuous real function
vanishing at the infinite point.
Let us show that by defining an ultra-violet renormalized coupling con-
stant (with a finite volume Ω cut off built in).
gren / 0
2−α π − 4
1
gbare (α) = (8.2)
(1 − α) 1/2

one can show that the interaction function

 
exp −gbare (α) 2
d x V (ϕ(x)) (8.3)
Ω

(0)
is an integrable function on L1 (L2 (R2 ), dα,εIR μ [ϕ]) and leads to a well-
defined ultra-violet functional integral in the limit of α → 1.
The proof is based on the following estimates.
Since almost everywhere we have the pointwise limit
 
exp −gbare (α) 2
d x V (ϕ(x))
$N
 (−1)n (gbare (α))n
lim dk1 · · · dkn Ṽ (k1 ) · · · Ṽ (kn )
N →∞ n! [−Λ,Λ]
n=0

× dx1 · · · dxn eik1 ϕ(x1 ) · · · eikn ϕ(xn ) (8.4)
Ω

we have that the upper-bound estimate below holds true

1 1 1 ∞
(−1)n (gbare (α))n
1 α 1 1
1ZεIR [gbare ]1 ≤ 1 dk1 · · · dkn Ṽ (k1 ) · · · Ṽ (kn )
n=0
n! R

N
1
i k ϕ(x ) 1
dx1 · · · dxn d(0)
α,εIR μ[ϕ](e
=1 )1 (8.5-a)
Ω

with
 
ZεαIR [gbare ] = d(0)
α,εIR μ[ϕ] exp −gbare (α) d2 x V (ϕ(x)) (8.5-b)
Ω
1 It could be as well consider also a polinomial interaction of the form V
n,p (x) =
minimum of {(ϕ(x))p , n} with p and n psotive integers. Note that Ṽ (k) ∈ L1 (R)∩L∞ (R)
by hypothesis
   
 
 eikϕ(x) · Ṽ (k)dk d2 x ≤ vol(Ω)||Ṽ ||L1 < ∞

Ω
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 160

160 Lecture Notes in Topics in Path Integrals

(0)
we have, thus, the more suitable form after realizing the d2 ki and dα,εIR μ[ϕ]
integrals respectivelly2
1 1  ∞
(gbare (α))n / 0n
1 α 1
1ZεIR =0 bare 1
[g ] ≤ ||Ṽ ||L∞ (R)
n=0
n!
1 ;  1
1 1
dx1 · · · dxn det− 2 G(N
1
1 α
)
(xi , xj ) 1≤i≤N 1 (8.6)
R2 1≤j≤N
(N )
Here [Gα (xi , xj )] 1≤i≤N denotes the N × N symmetric matrix with
1≤j≤N
the (i, j) entry given by the positive Green-function of the α-Laplacean
(without the infra-red cut off here!).
Γ(1 − α) 1 2(1−α)
Gα (xi , xj ) = |xi − xj |2(α−1) 2 (8.7)
Γ(α) 2π
At this point, we call the reader attention that we have the formulae on
the asymptotic behavior for α → 1 and α < 1 (see ref. [12] - Appendix A).
$ '
lim det− 2 [G(N
1 N
)
α (xi , xj )] ∼ eN πiα × (π 4 · 2N α )
α→1
α>1
(+1)
× (1 − α)N | (8.8)
(−1)N (N − 1)
After substituting eq. (8.8) into eq. (8.6) and taking into account the
hypothesis of the compact support of the nonlinearity Ṽ (k), one obtains
the finite bound for any value grem > 0, without the finite volume cut off
and producing a proof for the convergence of the perturbative expansion in
terms of the renormalized coupling constant for the model
1 1  ∞
(Ṽ L∞ (R) )n n
1 1 gren
lim 1ZεαIR =0 [gbare (α)]1 ≤ 1
α→1
n=0
n! (1 − α) 2
× (1 − α)n/2 × (vol(Ω))n
≤ exp{gren ||V ||L∞ (R) vol(Ω)} < ∞ (8.9)
Another important rigorously defined functional integral is to consider
the following α-power Klein Gordon operator on Euclidean space-time
χΩ (2π)ν/2 χΩ (2π)ν/2
L−1
Ω = [(−Δ)α + m2 ]−1 × (8.10)
vol(Ω) vol(Ω)
2 Note that:
≥0 ≥0
     
N N
2 2 −1 i,j ki kj Gα (xi , xj ) ≤ 2 2 −1 i,j ki kj Gα (xi , xj )
d x1 . . . d xn e 2 d x1 . . . d xn e 2
Ω R2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 161

Some comments on rigorous finite-volume euclidean quantum field path integrals 161

with m2 a positive “mass” parameters.

Let us note that L−1 is an operator of trace class on L2 (Rν ) if and only
if the result below holds true
1 >π νπ ?
T rL2 (Rν ) (L−1 ) = = C̄(ν) m( α −2) ×
ν
dν k 2α 2
cosec <∞
k +m 2α 2α
(8.11)
namely if
ν
α> (8.12)
2
In this case, let us consider the double functional integral with functional
domain L2 (Rν )
(0)
Z[j, k] = dG β[v(x)]

(0)
× dLΩ μ[ϕ]
 
× exp i ν
d x (j(x) ϕ(x) + k(x) v(x)) (8.13)

where the Gaussian functional integralJ on the fields V (x) has a Gaussian
generating functional defined by a 1 -integral operator with a positive-
definite kernel g(|x − y|), namely
 
(0)
Z (0) [k] = dG β[v(x)] exp i dν x k(x)v(x)
 
1
= exp − dν x dν y (k(x) g(|x − y|) k(y)) (8.14)
2 Ω Ω
By a simple direct application of the Fubbini-Tonelli theorem on the
exchange of the integration order on eq. (8.13), lead us to the effective λϕ4
- like well-defined functional integral representation
(0)
Zeff [j] = dL μ[ϕ][ϕ(x)]
 
1
exp − d xd y |ϕ(x)| g(|x − y|) |ϕ(y)|
ν ν 2 2
2 Ω
 
× exp i dν x j(x)ϕ(x) (8.15)
Ω
Note that if one introduces from the begining a bare mass parame-
ters m2bare depending on the parameters α, but such that it always sat-
isfies eq. (8.11) one should obtains again eq. (8.15) as a well-defined
measure on L2 (Rν ). Of course that the usual pure Laplacean limit of
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 162

162 Lecture Notes in Topics in Path Integrals

α → 1 on eq. (8.10), will needed a renormalization of this mass parameters

( lim m2bare (α) = +∞!) as much as it has been done in previous studies.
α→1
Let us continue our examples by showing again the usefulness of the
precise determination of the functional - distributional structure of the do-
main of the functional integrals in order to construct rigorously these path
integrals without complicated lattice limit procedures ([2]).
Let us consider a general Rν Gaussian measure defined by the Generat-
ing functional on S(Rν ) defined by the α-power of the Laplacean operator
−Δ acting on S(Rν ) with a small infrared regularization mass parameter
μ2 as considered in eq. (8.1-a)
 
1C 2 −1
D
Z(0) [j] = exp − j, ((−Δ) + μ0 ) j
α
2 L2 (Rν )

= d(0)
α μ[ϕ] exp(i ϕ(j)) (8.16)
E alg (S(Rν ))

An explicit expression in momentum space for the Green function of the

α-power of (−Δ)α + μ20 given by
dν k ik(x−y) 1
((−Δ)+α + μ20 )−1 (x − y) = e (8.17)
(2π)ν k 2α + μ20
Here C̄(ν) is a ν-dependent (finite for ν-values!) normalization factor.
Let us suppose that there is a range of α-power values that can be
choosen in such way that one satisfies the constraint below

d(0)
α μ[ϕ](ϕL2j (Rν ) )
2j
<∞ (8.18)
E alg (S(Rν ))

with j = 1, 2, · · · , N and for a given fixed integer N , the highest power

of our polinomial field interaction. Or equivalently, after realizing the ϕ-
Gaussian functional integration, with a space-time cutt off volume Ω on the
interaction to be analyzed on eq. (8.16)
1 1 11 1
j1
1 1 1 dν k 1
1 2 −j 1
d x[(−Δ) + μ0 ] (x, x)1 ≤ 1vol(Ω) ×
ν α
1
1 1 k 2α + μ20 1
Ω
1; /π 1
1
1 ν
−2) νπ 0j 11
= 1 Cν (μ0 )(α
× cosec 1<∞
2α 2α
(8.19)
For α > ν2 , one can see by the Minlos theorem that the measure support
of the Gaussian measure eq. (8.16) will be given by the intersection Banach
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 163

Some comments on rigorous finite-volume euclidean quantum field path integrals 163

space of measurable Lebesgue functions on Rν instead of the previous one

E alg (S(Rν )) ([4]–[6]).
K
N
L2N (Rν ) = (L2j (Rν )) (8.20)
j=1

In this case, one obtains that the finite - volume p(ϕ)2 interactions
⎧ ⎫
⎨  N ⎬
exp − λ2j (ϕ2 (x))j dx ≤ 1 (8.21)
⎩ Ω ⎭
j=1

is mathematically well-defined as the usual pointwise product of measur-

able functions and for positive coupling constant values λ2j ≥ 0. As a
(0)
consequence, we have a measurable functional on L1 (L2N (Rν ); dα μ[ϕ])
(since it is bounded by the function 1). Thus, it makes sense to consider
mathematically the well-defined path - integral on the full space Rν with
those values of the power α satisfying the contraint eq. (8.17).
⎧ ⎫
⎨  N ⎬
Z[j] = α μ[ϕ] exp −
d(0) λ2j ϕ2j (x)dx ×exp(i j(x)ϕ(x))
L2N (Rν ) ⎩ Ω ⎭ Rν
j=1
(8.22)
Finally, let us consider an interacting field theory in a compact space-
time Ω ⊂ Rν defined by an iteger even power 2n of the Laplacean operator
with Dirichlet Boundary conditions as the free Gaussian kinetic action,
namely
 
1C D
Z (0) [j] = exp − j, (−Δ)−2n j 2
2 L (Ω)

(0)
= d(2n) μ[ϕ] exp(ij, ϕL2 (Ω) ) (8.23)
W2n (Ω)

here ϕ ∈ W2n (Ω) - the Sobolev space of order n which is the functional
domain of the cylindrical Fourier Transform measure of the Generating
functional Z (0) [j], a continuous bilinear positive form on W2−n (Ω) (the
topological dual of W2n (Ω)) ([4]–[6]).
By a straightforward application of the well-known Sobolev immersion
theorem, we have that for the case of
ν
n−k > (8.24)
2
including k a real number the functional Sobolev space W2n (Ω) is contained
in the continuously fractional differentiable space of functions C k (Ω). As a
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 164

164 Lecture Notes in Topics in Path Integrals

consequence, the domain of the Bosonic functional integral can be further

reduced to C k (Ω) in the situation of eq. (8.24)
(0)
Z (0) [j] = d(2n) μ[ϕ] exp(ij, ϕL2 (Ω) ) (8.25)
C k (Ω)

That is our new result generalizing the Wiener theorem on Brownian

paths in the case of n = 1 , k = 12 and ν = 1
Since the bosonic functional domain on eq. (8.25) is formed by real
functions and not distributions, we can see straightforwardly that any in-
teraction of the form
 
exp −g ν
F (ϕ(x))d x (8.26)
Ω

with the non-linearity F (x) denoting a lower bounded real function (γ > 0)
F (x) ≥ −γ (8.27)
is well-defined and is integrable function on the functional space
(0)
(C k (Ω), d(2n) μ[ϕ]) by a direct application of the Lebesgue theorem
1  1
1 1
1 exp −g F (ϕ(x)) d x 1 ≤ exp{+gγ}
ν
(8.28)
Ω

At this point we make a subtle mathematical remark that the infinite

volume limit of eq. (8.25)–eq. (8.26) is very difficult, since one looses the
Garding - Poincar inequalite at this limit for those elliptic operators and,
thus, the very important Sobolev theorem. The probable correct procedure
to consider the thermodynamic limit in our Bosonic path integrals is to
consider solely a volume cut off on the interaction term Gaussian action as
in eq. (8.22) and there search for vol(Ω) → ∞ ([7]–[10]).
As a last remark related to eq. (8.23) one can see that a kind of “fishnet”
exponential generating functional
 
1C D
Z (0) [j] = exp − j, exp{−αΔ}j (8.29)
2 L2 (Ω)

has a Fourier transformed functional integral representation defined on

the space of the infinitely differentiable functions C ∞ (Ω), which physically
means that all field configurations making the domain of such path integral
has a strong behavior like purely nice smooth classical field configurations.
As a last important point of this note, we present an important result
on the geometrical characterization of massive free field on an Euclidean
Space-Time ([10]).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 165

Some comments on rigorous finite-volume euclidean quantum field path integrals 165

Firstly we announce a slightly improved version of the usual Minlos

Theorem ([4]).

Theorem 3. Let E be a nuclear space of tests functions and9dμ:a given

σ-measure on its topologic dual with the strong topology. Let , be an
90 :
inner product in E, inducing a Hilbertian structure on H0 = (E, , ),
0
after its topological completation.
We suppose the following:
a) There is a continuous positive definite functional in H0 , Z(j), with
an associated cylindrical measure dμ.
b) There is a Hilbert-Schmidt operator T : H0 → H0 ; invertible, such
that E ⊂ Range (T ), T −1 (E) is dense in H0 and T −1 : H0 → H0 is con-
tinuous.
We have thus, that the support of the measure satisfies the relationship
support dμ ⊆ (T −1 )∗ (H0 ) ⊂ E ∗ (8.30)
At this point we give a non-trivial application of ours of the above cited
Theorem 3.
Let us consider an differential inversible operator L : S  (RN ) → S(R),
together with an positive inversible self-adjoint elliptic operator P : D(P ) ⊂
L2 (RN ) → L2 (RN ). Let Hα be the following Hilbert space
$ 9 : 9 : '
Hα = S(RN ), P α ϕ, P α ϕ = , , for α a real number .
L2 (RN ) α
(8.31)
We can see that for α > 0, the operators below
P −α : L2 (RN ) → H+α
(8.32)
ϕ → (P −α ϕ)
P α : H+α → L2 (RN )
(8.33)
ϕ → (P α ϕ)
are isometries among the following sub-spaces
9 :
D(P −α ), , ) and H+α
L2
since
9 : 9 : 9 :
−α −α α −α α −α
P ϕ, P ϕ = P P ϕ, P P ϕ = ϕ, ϕ
H+α L2 (RN ) L2 (RN )
(8.34)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 166

166 Lecture Notes in Topics in Path Integrals

and
9 : 9 :
α α
P f, P f = f, f (8.35)
L2 (RN ) H+α

If one considers T a given Hilbert-Schmidt operator on Hα , the com-

posite operator T0 = P α T P −α is an operator with domain being D(P −α )
and its image being the Range (P α ). T0 is clearly an invertible operator
and S(RN ) ⊂ Range (T ) means that the equation (T P −α )(ϕ) = f has
always a non-zero solution in D(P −α ) for any given f ∈ S(RN ). Note that
the condition that T −1 (f ) be a dense subset on Range (P −α ) means that
9 :
−1 −α
T f, P ϕ =0 (8.36)
L2 (RN )

has as unique solution the trivial solution f ≡ 0.

Let us suppose too that T −1 : S(RN ) → Hα be a continuous applica-
tion and the bilinear term (L−1 (j))(j) be a continuous application in the
L2 L2
Hilbert spaces H+α ⊃ S(RN ), namely: if jn −→ j, then L−1 : P −α jn −→
L−1 P −α j, for {jn }n∈Z and jn ∈ S(RN ).
By a direct application of the Minlos Theorem, we have the result
 
1 −1
Z(j) = exp − [L (j)(j)] = dμ(T ) exp(iT (j)) (8.37)
2 (T −1 )∗ Hα

Here the topological space support is given by

 
 −1 ∗ / 0
(T −1 )∗ Hα = P −α T0 P α (P α (S(RN )))

  (8.38)
= (P α )∗ (T0−1 )∗ (P −α) )∗ P α (S(RN ))

= P α T0−1 (L2 (RN ))

In the important case of L = (−Δ + m2 ) : S  (RN ) → S(RN ) and

(2π)N/2 J
T0 T0∗ = χΩvol(Ω) (−Δ + m2 )−2β × χΩ ∈ 1 (L2 (RN )) since T r(T0 T0∗ ) =
N
1 m2 2 Γ( N2 )Γ(2β − N2 ) N
< ∞ for β > with the choice
2(m2 )β 1 Γ(β) 4
P = (−Δ + m2 ), we can see that the support of the measure in the path-
integral representation of the Euclidean measure field in RN may be taken
as the measurable sub-set below
supp {d(−Δ+m2 ) u(ϕ)} = (−Δ + m2 )+α IΩ (x)(−Δ + m2 )+β (L2 (RN )) (39)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 167

Some comments on rigorous finite-volume euclidean quantum field path integrals 167

since L−1 P −α = (−Δ + m2 )−1−α is always a bounded operator in L2 (RN )

for α > −1.
As a consequence each field configuration can be considered as a kind
of “fractional distributional” derivative of a square integrable function as
written below of the formal infinite volume Ω → RN .
;  N +ε−1 
ϕ(x) = − Δ + m2 4 f (x) (8.40)

with a function f (x) ∈ L2 (RN ) and any given ε > 0, even if originally
all fields configurations entering into the path-integral were elements of
the Schwartz Tempered Distribution Spaces S  (RN ) certainly very “rough”
mathematical objects to characterize from a rigorous geometrical point of
view.
We have, thus, make a further reduction of the functional domain of the
free massive Euclidean scalar field of S  (RN ) to the measurable sub-set as
given by eq. (8.30) denoted by W (RN )
 
1 2 −1

exp − (−Δ + m ) j (j) = d(−Δ+m2 ) μ(ϕ) ei ϕ(j)
2 S  (RN )
9 :
N +ε−1
i f,(−Δ+m2 ) 4 f
= d(−Δ+m2 ) μ̃(f ) e L2 (RN ) (8.41)
W (RN )⊂S  (RN )

8.3 References

[1] B. Simon, “Functional Integration and Quantum Physics” - Aca-

demic Press, (1979).
[2] B. Simon, “The P (φ)2 Euclidean (Quantum) Field Theory”-
Princeton Series in Physics, (1974).
[3] J. Glimm and A. Jaffe, “Quantum Physics” - A Functional Integral
Point of View Springer Verlag, (1988).
[4] Y. Yamasaki, “Measure on Infinite Dimensional Spaces” - World
Scientific Vol. 5, (1985).
[5] Xia Dao Xing, “Measure and Integration Theory on Infinite Di-
mensional Space” Academic Press, (1972).
[6] L. Schwartz, “Random Measure on Arbitrary Topological Space
and Cylindrical Measures” - Tata Institute - Oxford University
Press, (1973).
[7] K. Symanzik, J. Math., Phys., 7, 510 (1966).
[8] B. Simon, Journal of Math. Physics, vol. 12, 140 (1971).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 168

168 Lecture Notes in Topics in Path Integrals

[9] Luiz C.L. Botelho, Phys., Rev. 33D , 1195 (1986).

[10] E Nelson, Regular probability measures on function space; Annals
of Mathematics (2), 69, 630 - 643, (1959).
[11] Walter Rudin, Real and Complex Analysis, second edition, Tata
McGraw-Hill, Publishing Co Limited, New Delhi, (1979).
[12] Luiz C.L. Botelho - “A Simple Renormalization Scheme in Random
Surface Theory” - Modern Physics Letters, 13B, nos 6 and 7, 203–
207, (1999).

8.4 Appendix A: Some Comments on the Support of Func-

tional Measures in Hilbert Space

Let us comment further on the application of the Minlos Theorem in Hilbert

Spaces. In this case one has a very simple proof which holds true in general
Banach Spaces (E, || ||).
Let us thus, give a cylindrical measures d∞ μ(x) in the algebraic dual
alg
E of a given Banach Space E ([4]–[6]).
Let us suppose either that the function ||x|| belongs to L1 (E alg , d∞ μ(x)).
Then the support of this cylindrical measures will be the Banach Space E.
The proof is the following:
Let A be a sub-set of the vectorial space E alg (with the topology of
pontual convergence), such that A ⊂ E c (so ||x| = +∞) (E can always
be imbed as a cylindrical measurable sub-set of E alg - just use a Hammel
vectorial basis to see that). Let be the sets An = {x ∈ E alg | ||x|| ≥ n}.
I∞
Then we have the set inclusion A ⊂ n=0 An , so its measure satisfies the
estimates below:

μ(A) ≤ lim inf μ(An )

n

= lim inf μ{x ∈ E alg | ||x|| ≥ n}

n
 
1
≤ lim inf ||x||d∞ μ(x)
n n E alg
||x||L1 (E alg ,d∞
μ )
= lim inf = 0. (A-1)
n n
Leading us to the Minlos theorem that the support of the cylindrical
measure in E alg is reduced to the own Banach Space E.
Note that by the Minkowisky inequality for general integrals, we have
that ||x||2 ∈ L1 (E alg , d∞ μ(x)). Now it is elementary evaluation to see that
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 169

Some comments on rigorous finite-volume euclidean quantum field path integrals 169

J
if A−1 ∈ 1 (M), when E = M, a given Hilbert Space, we have that

d∞
A μ(x) · ||x|| = T rM (A
2 −1
) < ∞. (A-2)
Malg
This result produces another criterium for supp d∞
A μ = M (the Minlos
Theorem), when E = M is a Hilbert Space.
It is easy too to see that if

||x||d∞ μ(x) < ∞ (A-3)

M
then the Fourier-Transformed functional
Z(j) = ei(j,x)M d∞ μ(x) (A-4)
M
is continuous in the norm topology of M.
Otherwise, if Z(j) is not continuous in the origin 0 ∈ M (without loss
of generality), then there is a sequence {jn } ∈ M and δ > 0, such that
||jn || → 0 with

δ ≤ |Z(jn ) − 1| ≤ |ei(jn ,x)M − 1|d∞ μ(x)

M

≤ |(jn , x)|d∞ μ(x)

M
/ 0
≤ ||jn || ||x||d∞ μ(x) → 0, (A-5)
M
a contradiction with δ > 0.
Finally, let us consider an elliptic operator B (with inverse) from the
Sobelev space M−2m (Ω) to M2m (Ω). Then by the criterium given by
Eq. (A-2) if
T rL2 (Ω) [(I + Δ)+ 2 B −1 (I + Δ)+ 2 ] < ∞,
m m
(A-6)
we will have that the path integral below written is well-defined for x ∈
M+2m (Ω) and j ∈ M−2m (Ω). Namely
1
exp(− (j, B −1 j)L2 (Ω) ) = dB μ(x) exp(i(j, x)L2 (Ω) ). (A-7)
2 M+2m (Ω)

By the Sobolev theorem which means that the embeeded below is con-
tinuous (with Ω ⊆ Rν denoting a smooth domain), one can further reduce
the measure support to the Hlder α continuous function in Ω if 2m− ν2 > α.
Namely, we have a easy proof of the famous Wiener Theorem on sample
continuity of certain path integrals in Sobolev Spaces
M2m (Ω) ⊂ C α (Ω) (A-8-a)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 170

170 Lecture Notes in Topics in Path Integrals

The above Wiener Theorem is fundamental in order to construct non-

trivial examples of mathematically rigorous euclideans path integrals in
spaces Rν of higher dimensionality, since it is a trivial consequence of the
Lebesgue theorem that positive continuous functions V (x) generate func-
tionals integrable in {M2m (Ω), dB μ(ϕ)} of the form below
 
exp − V (ϕ(x))dx ∈ L1 (M2m (Ω), dB μ(ϕ)). (A-8-b)
Ω
As a last important remark on Cylindrical Measures in Separable Hilbert
Spaces, let us point at to our reader that the support of such above measures
is always a σ-compact set in the norm topology of M. In order to see
such result let us consider a given dense set of M, namely {xk }k∈I + . Let
{δk }k∈I + be a given sequence of positive real numbers with δk → 0. Let
@∞
{εn } another sequence of positive real numbers such that n=1 εn < ε.
L∞
Now it is straightforward to see that M ⊂ h=1 B(xk , δk ) ⊂ M and thus
Ln
lim sup μ{ k=1 B(xk , δk )}/= μ(M) = 1. 0As a consequence, for each n,
Lkn
there is a kn , such that μ k=1 B(xk , δk ) ≥ 1 − ε.
I∞ ;Lkn 
Now the sets Kμ = n=1 k=1 B(x k , δ k ) are closed and totally
bounded, so they are compact sets in M with μ(M) ≥ 1 − ε. Let is
now choose ε = n1 and the associated compact sets {Kn,μ }. Let us further
Ln
consider the compact sets K̂n,μ = =1 K,μ . We have that K̂n,μ ⊆ K̂n+1,μ ,
L∞
for any n and lim sup μ(K̂n,μ ) = 1. So, supp dμ = n=1 K̂n,μ , a σ-compact
set of M.
We consider now a enumerable family of cylindrical measures {dμn } in
M satisfying the chain inclusion relationship for any n ∈ I +
supp dμn ⊆ supp dμn+1 .
(n)
Now it is straightforward to see that the compact sets {K̂n }, where
L∞ (m) L∞ (n)
supp dμm = n=1 K̂n , is such that supp {dμm } ⊆ n=1 K̂n , for any
m ∈ I +.
Let us consider the family of functionals induced by the restriction of
(n)
this sequence of measures in any compact K̂n . Namely

μn → L(n)
n (f ) = f (x) · dμp (x). (A-8-c)
(n)
K̂n
L
Here f ∈ Cb (K̂n ). Note that all the above functionals in ∞
(n) (n)
n=1 Cb (K̂n )
are bounded by 1. By the Alaoglu-Bourbaki
/L theorem 0∗they form a compact
∞ (n)
set in the weak star topology of n=1 Cb (K̂n ) , so there is a sub-
sequence (or better the whole sequence) converging to a unique cylindrical
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 171

Some comments on rigorous finite-volume euclidean quantum field path integrals 171

measure μ̄(x). Namely

lim f (x)dμn (x) = f (x)dμ̄(x) (A-8-d)

n→∞ M M
L∞ (n)
for any f ∈ n=1 Cb (K̂n ).

8.5 Appendix B

A straightforward calculation give the following expression for the trace of

the integral operator on Eq. (A-1)
+∞  +∞   
<Ω (k − p)<
χ χΩ (p − k)
Tr1 {χΩ [(−Δ)α +ε2IR ]−1 χΩ } = !dp dk,
−∞ −∞ (p2α + ε2IR )
(B-1)
where
 +∞ 
1
<Ω (p) =
χ ipx 2
χΩ (x)e d x . (B-2)
2π −∞

Since by Parservals theorem

+∞ +∞
dk |<
χΩ (k − p)|2 d2 k = |χΩ (x)|2 d2 x = vol(Ω) (B-3)
−∞ −∞

one has the validity of the result written on Eq. (B-1).

 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 173

Chapter 9

On the Rigorous Ergodic Theorem for

a Class of Non-Linear Klein Gordon
Wave Propagations

We present a complete study of the Ergodic Theorem for the difficult

problem of Non-Linear Klein Gordon Classical Wave Propagations through
Cylindrical Measures, rigorous mathematical Path Integrals and the famous
Ruelle-Amrein-Geogerscu-Enss (R.A.G.E.) theorem on the caracterization
of continuous spectrum of self-adjoint operators

9.1 Introduction

One of the most important phenomenon in numerical studies of the non-

linear wave motion especially in the two - dimensional case - is the existence
of overwhelming majority of wave motions that wandering over all possible
system’s phase space and, given enough time, coming as close as desired
(but not entirely coinciding) to any given initial condition ([1]).
This phenomenon is signaling certainly that the famous recurrence the-
orem of Poincar is true for infinite dimensional continuum mechanical sys-
tems like that one represented by a bounded domain when subject to non-
linear vibrations.
A fundamental question appears in the context of this recurrence phe-
nomenon and concerned to the existence of the time average for the asso-
ciated non-linear wave motion and naturally leading to the concept of a
infinite - dimensional invariant measure for the non-linear wave equation
([1]), the mathematical phenomenon subjacente to the Poincar recurrence
theorem.
In this paper, we intend to give mathematical methods rigorous proofs
for the validity of the famous Ergodic theorem in a special class of polino-
mial non-linear and Lipschitz wave motions. Our approach is fully based
on Hilbert Space methods previously used to study Dynamical System’s

173
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 174

174 Lecture Notes in Topics in Path Integrals

in Classical Mechanics which by its turn simplifies enormously the task of

constructing explicitly the associated Birkoff–Von Neumann invariant mea-
sure on the infinite - dimensional space of wave’s motion initial conditions.
This study is presented on the main section of this paper, namely section
4 and for the Klein Gordon wave motion.
In section 2, we give a very detailed proof of the RAGE’S theorem,
a basic functional analysis rigorous method used in our proposed Hilbert
Space generalization of the usual finite - dimensional Ergodic theorem and
fully presented in section 3.
In section 5, we complement our studies by commenting the important
case of Klein Gordon wave - diffusion under random stirring.

9.2 On the detailed mathematical proof of the R.A.G.E.

theorem1

In this purely mathematical first section 2 of our study, we intend to present

a detailed mathematical proof of the R.A.G.E theorem ([2]) used on the
analytical proof of ours of the Ergodic theorem on section 3 for Hamiltonian
systems of N -particles.
Let us, thus, start our analysis by considering a self - adjoint operator L
on a Hilbert Space (H, (, )) where Hc (L) denotes the associated continuity
sub-space obtained from the spectral theorem applied to L. We have the
following result:

Proposition (RAGE theorem) 1. Let ψ ∈ Hc (L) and ψ̃ ∈ H. We have

the Ergodic limit
1 T 11 12
1
lim 1(ψ̃, exp(i tL)ψ)1 dt = 0 (9.1)
T →∞ T 0

Proof: In order to show the validity of the above ergodic limit, let us
re-write eq. (9.1) in terms of the spectral resolution of L, namely.
 +∞ 
1 T 11 12
1 1 T>
I= 1(ψ̃, exp(i tL)ψ)1 dt = e+i tλ dλ (ψ̃, Ej (λ)ψ)
T 0 T 0 −∞
 +∞ 
−i tμ
× e dμ (ψ̃, Ej (μ)ψ) (9.2)
−∞
1 SeeAppendix C for a rigorous mathematical proof of the Ergodic Theorem for Wide-
Sense Stationary Stochastic Process.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 175

On the Rigorous Ergodic Theorem for a Class 175

Here we have used the usual spectral representation of L

+∞
(g, L h) = λ dλ (g, Ej (λ)h) (9.3)
−∞

with g ∈ H and h ∈ Dom (L).

Let us remark that the function exp(i(λ − μ)t) is majorized by the
function 1 which, by its turn, is an integrable function on the domain
[0, T ] × R × R with the product measure
dt
⊗ dλ (ψ̃, Ej (λ)ψ) ⊗ dμ (ψ̃, Ej (μ)ψ) (9.4)
T
since
T
dt
⊗ dλ (ψ̃, Ej (λ)ψ) ⊗ dμ (ψ̃, Ej (μ)ψ) = (ψ̃, ψ)2 < ∞ (9.5)
0 T
At this point, we can safely apply the Fubbini theorem for interchange
the order of integration in relation to the t-variable which leads to the
partial result below
+∞ +∞
ei(λ−μ)T − 1
I= dλ ψ̃, Ej (λ)ψ ⊗ dμ ψ̃, Ej (μ)ψ (9.6)
−∞ −∞ i(λ − μ)T
Let us consider two cases. Firstly we take ψ̃ = ψ. In this case, we have
the bound
1 ei(λ−μ)T − 1 1 1 2 sen( (λ−μ) T ) 1
1 1 1 1
1 1≤1 2
1≤1 (9.7)
i(λ − μ)T (λ − μ)T
If we restrict ourselvers to the case of λ = μ, a direct application of the
Lebesgue convergence theorem on the limit T → ∞ yields that I = 0.
On the other hand, in the case of λ = μ we intend now to show that
the set μ = λ is a set of zero measure in R2 in respect with the measure
dλ ψ, Ej (λ)ψ ⊗ dμ ψ, Ej (μ)ψ. This can be seen by considering the real
function on R
f (λ) = (ψ, Ej (λ)ψ) (9.8)
We observe that this function is continuous and non-decreasing since
ψ ∈ Hc (L) and range Ej (λ1 )C range Ej (λ2 ) for λ1 ≤ λ2 . Now, f (λ) is
an uniform continuous function on the whole real line R, since for a given
ε > 0, there is a constant λ̃ such that
ε
(ψ, Ej (−∞, −λ̃)ψ) < (9.9)
2
ε
ψ2 − (ψ, Ej (−∞, λ̃)ψ) < (9.10)
2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 176

176 Lecture Notes in Topics in Path Integrals

Additionally, f (λ) is uniform continuous on the closed interval [−λ̃, λ̃]

and f (λ) can not make variations greater than 2ε on (−∞, −λ̃] and [+λ̃, ∞),
besides of being a monotonic function on R. These arguments show the
uniform continuity of f (λ) on whole line R. Hence we have that for a given
ε > 0, exists a δ > 0 such that
1 1
1 1 ε
1(ψ, Ej (λ )ψ) − (ψ, Ej (λ )ψ)1 ≤ (9.11)
ψ2
for
1 1
1  1
1λ − λ 1 ≤ δ (9.12)

In particular for λ = λ + δ e λ = λ − δ

ε
(ψ, Ej (λ + δ)ψ) − (ψ, Ej (λ − δ)ψ) ≤ (9.13)
ψ2
As a consequence, we have the estimate
+∞ λ+δ
dλ (Ej (λ)ψ, ψ) dμ (ψ, Ej (μ)ψ)
−∞ λ−δ
≤ ψ2 × ((ψ, Ej (λ + δ)ψ) − (ψEj (λ − δ)ψ))
ε
≤ ψ2 × ≤ε (9.14)
ψ2

Let us note that for each ε > 0, there is a set Dδ = {(λ, μ) ∈ R2 ; |λ −

μ| < δ} wich contains the line λ = μ and has measure less than ε in relation
to the measure dλ (Ej (λ)ψ, ψ)⊗dμ (Ej (μ)ψ, ψ) as a result of eq. (9.14). This
shows our claim that I = 0 in our special case.
In the general case of ψ̃ = ψ, we remark that solely the orthogonal
component on the continuity sub-space Hc (L) has a non - vanishing inner
product with exp(itL)ψ. By using now the polarization formulae, we reduce
this case to the first analyzed result of I = 0.
At this point we arrive at the complete R.A.G.E theorem ([2])

Theorem 2. Let K be a compact operator on (H, ( , )). We have thus the

validity of the Ergodic limit
T
1
lim K exp(i tL)ψ2H dt = 0 (9.15)
T →∞ T 0

with ψ ∈ Hc (L).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 177

On the Rigorous Ergodic Theorem for a Class 177

We leave the details of the proof of this result for the reader, since any
compact operator is the norm operator limit of finite - dimension operators
and one only needs to show that

1 T 
N
lim  cn (ei tL ψ, en )gn H dt = 0 (9.16)
T →∞ T 0 n=0

for cn constants and {en }, {gn} a finite set of vector of (H, (, )) .

9.3 On the Boltzman Ergodic Theorem in Classical Me-

chanics as a result of the R.A.G.E theorem

One of the most important statement in Physics is the famous zeroth law
of thermodynamics: “any system approaches an equilibrium state”. In the
classical mechanics frameworks, one begins with the formal elements of the
theory. Namely, the phase-space R6N associated to a system of N -classical
particles and the set of Hamilton equations
∂H ∂H
ṗi = − ; q̇i = (9.17)
∂qi ∂pi
where H(q, p) is the energy function.
The above cited thermodynamical equilibrium principle becomes the
mathematical statement that for each compact support continuous func-
tions Cc (R6N ), the famous ergodic limit should holds true ([3]).
$ '
3N 3N 1 T
d q(0)d p(0) lim f (q(t); p(t))dt = η(f ) (9.18)
R6N T →∞ T 0

where η(f ) is a linear functional on Cc (R6N ) given exactly by the Boltz-

man statistical weight and {q(t), p(t)} denotes the (global) solution of the
Hamilton equations (9.12).
For instance:

Theorem 3. Let V (q i ) be a C 2 function on RDN with ||∇qj V (q i )||RDN ≤

C(||q||2RDN + a2 )1/2 for suitable positive non-vanishing constants C and a.

@
3N
Let H(pi , q i ) = Aij pi pj + V (q i ) on R2DN , with A denoting a strictly
i=1
positive definite matrix. Then for any p0 , q0  ∈ R2DN , there is a unique
C 1 function from R to R2DN , denoted by w(t; p0 , q0 ) and satisfying globally
Eq. (9.17) with initial conditions p0 , q0 .
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 178

178 Lecture Notes in Topics in Path Integrals

Moreover w(t; p0 , q0 ) is an C 1 application of R2DN +1 to R2DN (see ref.

[1]).
Note that if V (q i ) is a Lipschitzian function on C 2 (RDN ) the theorem
necessarily holds true.
We aim at this section point out a simple new mathematical argument
of the fundamental eq. (9.18) by means of Hilbert Space Techniques and
the R.A.G.E’S theorem. Let us begin by introducing for each initial con-
dition (q(0), p(0)), a function ωq0 ,p0 (t) ≡ (q(t), p(t)), here q(t), p(t) is the
assumed global unique solution of eq. (9.17) with prescribed initials con-
ditions. Let Ut : L2 (R6N ) → L2 (R6N ) be the unitary operator defined
by
(Ut f )(q, p) = f (ωq0 p0 (t)) (9.19)
We have the following theorem (the Liouville’s theorem) ([1]).

Theorem 4. Ut is a unitary one-parameters group whose infinitesimal

generator is −iL̄, where −iL is the essential self-adjoint operator acting on
C0∞ (R6N ) defined by the Poisson bracket.

3N
∂f ∂H ∂f ∂H
(Lf )(p, q) = {f, H}(q, p) = − (q, p) (9.20)
i=1
∂qi ∂pi ∂pi ∂qi
The basic result we are using to show the validity of the ergodic limit
eq. (9.18) is the famous R.A.G.E’s theorem exposed on section 1.

Theorem 5. Let φ ∈ Hc (−iL̄)(L2 (R6N ), here Hc (−iL̄) is the continuity

sub-space associated to self-adjoint operator −iL̄. For every vector β ∈
L2 (R6N ), we have the result
1 T 11 12
1
lim 1β, Ut ψ1 dt = 0, (9.21)
T →∞ T 0
or equivalently for every ψ ∈ L2 (R6N ) and β ∈ L2 (R6N )
1 T
lim β, Ut ψdt = β, Pker(−iL̄) ψ (9.22)
T →∞ T 0
where Pker(−iL̄) is the projection operator on the (closed) sub-space
ker(−iL̄).

That eq. (9.22) is equivalent to eq. (9.21), is a simple consequence of

the Schwartz inequality below written
 1  1
1 T 1 T 2 T 2
1 1
1 (β, Ut ψ)dt1 ≤ (β, Ut ψ)2 dt 1dt (9.23)
0 0 0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 179

On the Rigorous Ergodic Theorem for a Class 179

or
 1
11 T 1 T 2
1 1 1
1 β, Ut ψdt1 ≤ 2
(β, Ut ψ) dt (9.24)
T 0 T 0

As a consequence of eq. (9.23), we can see that the linear functional

η(f ) of the Ergodic theorem is exactly given by (just consider β(p, q) ≡ 1
on supp of Pker(−iL̄) (ψ), for ψ ∈ Cc (R6N ) arbitrary)

η(f ) = dqdp Pker (−iL̄) (f )(q, p). (9.25)

R6N
By the Riesz’ s theorem applied to η(f ), we can re-write (represent)
eq. (9.25) by means of a (kernel)-function hη(H) (q, p), namely

η(f ) = d3N qd3N p · f (p, q) · hη(H) (p, q) (9.26)

R6N
where the function hη(H) (q, p) satisfies the relationship

3N
∂hη ∂H ∂H ∂hη
{hη(H) , L} = − =0 (9.27)
i=1
∂qi ∂pi ∂qi ∂pi
or equivalently hη(H) (q, p) is a “smooth” function of the Hamiltonian func-
tion H(q, p), by imposing the additive Boltzman behavior for hη(H) (q, p)
namely, hη(H1 +H2 ) = hη(H1 ) · hη(H2 ) , one obtains the famous Boltzman
weight as the (unique) mathematical output associated to the Ergodic
Theorem on Classical Statistical Mechanics in the presence of a thermal
reservoir ([4], [5]).
M
hη(H) (q, p) = exp{−βH(q, p)} d3N qd3N p exp{−βH(q, p)} (9.28)

with β a (positive) constant which is identified with the inverse macroscopic

temperature of the combined system after evaluating the system internal
energy in the equilibrium state. Note that hη(H) L2 = 1 since η(f ) = 1.
A last remark should be made related to eq. (9.28). In order to obtain
this result one should consider the non-zero value in ergodic limit
T
1
lim dt(Ut ψ)(q, p)hη(H) (q, p) = hη(H) , P ker (−iL̄) (ψ) (9.29)
T →∞ T 0
or by a pointwise argument (for every t)
(U−t hη(H) ) ∈ P ker (−iL̄) , (9.30)
that is
hη(H) ∈ P ker (−iL̄) ⇔ Lh = 0. (9.31)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 180

180 Lecture Notes in Topics in Path Integrals

9.4 On the invariant ergodic functional measure for non-

linear Klein-Gondon wave equations with kinetic trace
class operators

Let us start by considering the following initial value non-linear wave equa-
tion with a positive trace class operator as a kinetic operator on RD ([4]),
with U (t, x) ∈ C(R+ , L2 (RN )) (see Theorem 3 for the discreticized case)2
> −1 ? δV (U )
Utt (x, t) = − XΩ ((−Δ)α + m2 )−1 XΩ U (x, t) + (x, t)
δU
(9.32-a)
U (x, 0) = f (x) ∈ S(RD )
(9.32-b)
Ut (x, 0) = g(x) ∈ S(RD )
Here the Klein Gordon kinetic operator is given by an important set
of inverse of class of trace operator on Mathematical Physics of Construc-
tive Field Theory of rigorous path integrals ([4]), and defined by contin-
uum powers of the (strictly positive) Laplacean operator on RD (or any
other strongly uniform elliptic operator) but projected on a given compact
domain
$ Ω ⊂ RD through its charactheristic function IΩ (x) ≡ χΩ (x) ≡
1 x∈Ω
. The integral kernel of such class is given (defined) by
0 x ∈ Ω0
TΩ,α,m2 ) = χΩ (x)((−Δ)α + m2 )−1 (x, y)χΩ (y). (9.33)
The non negative non linear term V (U (x, t)) on our proposed Klein Gordon
Model
eqs. (9.32-a)–(9.32-b) is such that it allows to global solutions of the as-
sociated initial value Klein Gordon problem. Namely
||∇RD V (q i )||RD ≤ C(||q i ||2RD + a2 )1/2 . (9.34)
We now consider the discreticized (N particle) wave motion Hamiltonian
associated to the non-linear Klein-Gordon wave equation (for α > D 2)

N  2 
pi
H(pi , qi ) = + λ2i qi2 + V<N (q11 , . . . , qN
1
) . (9.35)
i=1
2
Here
∞

U (x, t) = qi (t)φi (x) (9.36-a)
i=1
2 We note that the existence, uniqueness and globality on t ∈ [0, ∞] of the U (x, t) ∈

C([0, T ], L2 (RD )) through the application of the fixed point theorem ([2]).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 181

On the Rigorous Ergodic Theorem for a Class 181

  1
T(Ω,α,m2 ) φi (x) = φi (x) (9.36-b)
λi
N

V<N (q1 , . . . , qN ) = V qi (t)φi (x) dD x (9.36-c)
RD i=1

Note the pointwise convergence for each fixed t ∈ R+

/ 0
lim V<N (q1 , . . . , qN ) = V (U (x, t))dD x. (9.37)
N →∞ RD

If F ∈ Cc (R), it holds true also

$ % N & '

lim F qi (t)φi (x) dD x
N →∞ RD i=1

(definition)
F<N (q1 (t), . . . , qN (t))
pointwise convergence for each fixed t
≡ −→

F (U (x, t))dD x (9.38)

RD

After these remarks we have our principal result

Theorem 6. Under the above stateg conditions, we have the Von-Neumann

Birkoff result in the infinite dimension case
$   '
1 T
lim F (U (x, t))dD x dt
T →∞ T 0 RD

 /  0 
D −β V (ϕ(x))dD x
= d x (dT(Ω,α,m2 ) μ(ϕ(x))) e RD F (ϕ(x)) (9.39)
RD

Proof: Let us apply the already proved ergodic theorem for our N -particle
system with Hamiltonian function eq. (9.35). Namely
$ '
1 T <
lim FN (q1 (t), . . . , qN (t))dt
T →∞ T 0
$ % N &'
1 
= F<N (q1 , . . . , qN ) exp −β λ2i qi2 + V<N (q1 , . . . , qN )
Z RN D i=1
(9.40)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 182

182 Lecture Notes in Topics in Path Integrals

where Z is the invariant ergodic measure normalization factor (given ex-

plicitly by considering the integrand on the right-hand side of eq. (9.40)
with F<N (q1 , . . . , qN ) ≡ 1).
At this point we can take safely the limit of N → ∞ on the cylindrical
measures of the right-hand side of eq. (9.40). Since the operator TΩ is a
class trace operator on L2 (RN ) (for α > N2 , and thus by Minlos theorem
defining a σ-measure on L2 (RN ) ([4]).
As a consequence we have the infinite dimensional result below written
$ % &'
1 T <
lim lim FN (q1 (t), . . . , qN (t))dt
N →∞ T →∞ T 0

$ '

−β V (ϕ(x))dD x
= dT(Ω,α,m2 ) μ(ϕ(x)) · e Ω F (φ(x)) . (9.40-a)
RD L2 (RD )

It is straightforward to argument that the left-hand side rigorous math-

ematical object is the physicist ergodic limit for the infinite particle limit
of our Hamiltonian system and taking as a definition for us
$ % &'
1 T <
lim lim FN (q1 (t), . . . , qN (t))dt
N →∞ T →∞ T 0

% &
1 T > ?
lim F<N (q1 (t), . . . , qN (t)) dt
(formally)
= lim
T →∞ T 0 N →∞

% &
T
(physicist’s definition) 1
≡ lim D
F (U (x, t))d x dt . (9.41)
T →∞ T 0 RD

Exactly mathematical conditions for interchange the above (N, T ) limits

still remains an open problem in the subject ([5]).
In the important case of N = 1 and α = 1 and exponentially cut-off
polinomial Klein-Gordon equation on C 2 ([0, ] × R+ ) with vanishing mass
m2 = 0, for 0 ≤ x ≤  (the usual wave equation)
∂ 2 U (x, t) d2 δV (U )
= (U (x, t)) + (x, t)
∂2t dx δU

U (0, ) = U (0, t) = 0

U (x, 0) = f (x) ∈ C 2 ([0, ]); f (3) (x)

∈ L2 ([0, ]); f () = f (0); f (2) (0) = f (2) () = 0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 183

On the Rigorous Ergodic Theorem for a Class 183

Ut (x, 0) = g(x) ∈ C 1 ([0, ]); g (2) (x) ∈ L2 ([0, ]); g(0) = g() = 0. (9.42)
Here the exponentially cut-off interaction (a Lipschitzian function) is
explicitly given by (ε ∈ R1 )

M
V ∗ U (x, t)) = ((λ2j 2 2j
j ) exp(−ε(U (x, t)) ))(U (x, t)) ), (9.43)
j=1
one can substitute the T(Ω,α,m2 ) operator on the two-dimensional Klein-
d2
Gordon wave equation by the usual one-dimensional operator − dx 2 , but

now defined on H 2 (Ω).

It leads thus to the important result that the associated Von-Neumann-
Birkoff invariant measure is given by the usual Wiener-Kac path measure
defined for closed Wiener trajectories with the marked fixed point X(0) =
X(1) = x
 
1
lim F (U (x, t))dx dt
T →∞ T R
 >
dx dWienerμ[X(σ)]
+ ,- .
Wiener closed trajectories
0
{C([0, 1]), X(0) = X(1) = x}
 1  ?
× exp −β V (X(σ))dσ F (X(σ)) (9.44)
0
/ 0−1
d2
Note that one can consider the case m = 0, since the operator − dx 2

with Dirichlet Boundary conditions is a trace class positive operator on

L2 (Ω), for Ω any closed interval on R1 .
Concerning the general case of a strongly non-negative elliptic operator
A of order 2m associated to the free vibration of a domain Ω with compact
closure in a general space RD (with the condition U ) ([6]) we still have the
same results above exposed
> −1 ? δV (U )
Utt = − IΩ [(A)+α + m2 ]−1 IΩ U (x, t) + (x, t)
δU
U (x, 0) = f (x) ∈ C 2 (Ω) ∩ C 1 (Ω)

Vt (x, 0) = g(x) ∈ C 3 (Ω) ∩ C 1 (Ω). (9.45)

Here A should satisfy the Garding coerciviness condition and given
explicitly by

A= (−1)|p| Dxp (Apq (x)Dxq (9.46)
|p|≤m
|q|≤m
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 184

184 Lecture Notes in Topics in Path Integrals

the Garding condition reads as of as:

Real(Au, u)L2 (Ω) ≥ C0 (Ω)||U ||H 2m (Ω) . (9.47)
At this point it appears important to point out that even if the limit of
infinite volume χΩ (x) → 1 is immediate on the Klein-Gordon wave equa-
tion eq. (9.32-a), quiet opposite to the case of the kinetic operator being
(−ΔΩ )α (1 − Pker(ΔΩ ) ); defined by the usual espectral power of the self-
adjoint Laplacean operator projected out of the zero modes (the vector
space of the harmonic functions on Ω); this infinite volume limit on the
associated Birkoff-Von Neumann ergodic invariant measure is highly non
trivial since its support in this case of infinite volume (if any) would be
fully the L. Schwartz distribution space ([4], [7], [8], [9]) and Appendix A
of this paper).
The same remarks apply to the α → 1 limit.
It is thus conjectured by us that the α → 1 and χΩ (x) → 1 limits
shouldbe taken only in the ergodic averages as it is usually expected on
Constructive field theory ([9], [10]) (see the right-hand side of eq. (9.39),
Theorem 6) and solely after considering the infinite renormalization prob-
lem issues ([9], [10], [11]).

9.5 An Ergodic theorem in Banach Spaces and Applications

to Stochastic-Langevin Dynamical Systems

In this complementary section 5, we intend to present our approach to

study long-time ergodic behavior of infinite-dimensional dynamical systems
by analyzing the somewhat formal diffusion equation with polinomial terms
and driven by a white noise stirring ([5]).
In order to implement such studies, let us present the author’s gener-
alization of the R.A.G.E theorem for a contraction self-adjoint semi-group
T (t) on a Banach space X. We have, thus, the following theorem.

Theorem 7. Let f ∈ X. We have the ergodic generalized theorem

T
1
lim dt(e−tA f ) = Pker(A) (f ) (9.48)
T →∞ T 0

where A is the infinitesimal generator of T (t).

Proof: As a first step, one should consider eq. (9.48) re-written in terms
of the “resolvent operator of A” by means of a Laplace Transform (The
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 185

On the Rigorous Ergodic Theorem for a Class 185

Hile-Yosida - Dunford Spectral Calculus)

 
1 T 1 i∞
zt −1
lim dt dze ((z + A) f ) . (9.49)
T →∞ T 0 2πi −i∞
Now it is straightforward to apply the Fubbini theorem to exchange the
order of integrations (dt, dz) in eq. (9.49) and get, thus, the result
T
1
lim dt(e−tA f ) = lim +((z + A)−1 f ). (9.50)
T →∞ T 0 z→0−

The z → 0− limit of the integral of eq. (9.50) (since Real (z) ⊂ ρ(A) ⊂
(−∞, 0) ) can be evaluated by means of saddle point techniques applied to
Laplace’s transforms. We have the following result
lim ((z + A)−1 f )
z→0−
∞
= lim− e−zt (e−tA f ) dt
z→0 0
= lim (e−tA f ) = Pker(A) (f ). (9.51)
t→∞

Let us apply the above theorem to the Lengevin Equation. Let us

consider the Fokker-Planck equation associated to the following Langevin
Equation
dq i ∂V j
(t) = − (q (t)) + η i (t) (9.52)
dt ∂qi
where {η i (t)} denotes a white-noise stochastic time process representing
the thermal coupling of our mechanical system with a thermal reservoir
at temperature T . Its two point function is given by the “Fluctuation-
Dissipation” theorem
η i (t)η j (t ) = kT δ(t − t ). (9.53)
The associated Fokker-Planck equation associated do eq. (9.52) has the
following explicitly form
∂P i i
(q , q̄ ; t) = +(kT )Δqi P (q i , q̄ i ; t) + ∇qi (∇qi V · P (q i , q̄ i ; t)) (9.54)
∂t

lim P (q i , q̄ i ; t) = δ (N ) (q i − q̄ i ). (9.55)
t→0+

By noting that we can associate a contractive semi-group to the initial

value problem eq. (9.55) in the Banach Space L1 (R3N ), namely:

P (q i , q̄ i , t)f (q̄ i )dq̄ i = (e−tA f ). (9.56)

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 186

186 Lecture Notes in Topics in Path Integrals

Here the closed positive accretive operator A is given explicitly by

−A(·) = kT Δqi (·) + ∇qi [(∇qi V )(·)] (9.57)
and acts firstly on X = C∞ (R3N ). It is instructive to point out that
the perturbation accretive operator B = ∇qi · [(∇qi V ) · (·)] on C∞ (R3N )
with V (q i ) ∈ Cc∞ (R3N ) is such that it satisfies the estimate on S(R3N ) :
Bf S(R3N ) ≤ aΔf S(R3N ) + bf S(R3N ) with a > 1 and for some b. As a
consequence A(·) = kT Δqi (·) + (B(·)) generates a contractive semi-group
on C∞ (R3N ) or by an extension argument on the whole L1 (R3N ) since the
L1 -closure of C∞ (R3N ) is the Banach space L1 (R3N ).
At this point we may apply our Theorem 7 to obtain the Langevin-
Brownian Ergodic theorem applied to our Fokker-Planck equation
T
1
lim dt dq(e−tA f )(q) = dqPker(A) (f )(q)
T →∞ T 0 R3N R3N

= dN q̄ i f (q̄ i )P eq (q̄ i ) (9.58)

RN
where the equilibrium probability distribution is given explicitly by the
unique normalizable element of the closed sub-space ker(A).
O = +(kT )Δqi P eq (q̄ i ) + ∇q̄i [(∇q̄i V )P(q̄ i )] (9.59)
or exactly, we have the Boltzman’s weight for our equilibrium Langevin-
Brownian probability distribution3
 
1
P eq (q̄ i ) = exp − (V (q̄ i )) . (9.60)
kT
For the general Langevin equation in the complete phase space {qi , pi }
as in the bulk of this note, one should re-obtains the complete Boltzman
statistical weight as the equilibrium ergodic probability distribution.
$ '
1 ; 1 i 2 
N
1
P (q̄ , p̄ ) = exp −
eq i i i
(p̄ ) + V (q̄ ) (9.61)
Z kT 2
l=1
with the normalization factor
Z= dp̄i dq̄ i P eq (q̄ i , p̄i ) (9.62)
RN RN
Let us apply the above exposed results for the initial-value diffusion
equation on RD (α > D2)

Ut (x, t) = −{[χΩ ((−Δ)+α + m2 )−1 χΩ ]−1 U }(x, t)

3 This result has the conceptual advantage of explaining the Boltzman equilibrium

statistic weight directly from the Ergodic Theorem.

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 187

On the Rigorous Ergodic Theorem for a Class 187

δV (U )
+ (x, t) + η(x, t)
δU

U (x, 0) = f (x) ∈ L2 (RD )

∞

η(x, t) = ηi (t)φi (x). (9.63)
i=1
By using the same discretisized eigenfunction expansion of section 4,
one obtains the associated invariant ergodic measure associated to the non-
linear diffusion equation (9.63)
dinv μ(ϕ(x)) = (dT(Ω,α,m2 ) μ(ϕ)) × e−(βV (ϕ(x))) (9.64)

9.6 Appendix A: The existence and uniqueness results for

some polinomial wave motions in 2D

In this technical Appendix A, we give an argument for the global existence

and uniqueness solution of the Hamiltonian motion equations associated
firstly to eq. (32) - section 3 and by secondly to eq. (53) - section 3 at
the infinite volume limit and for α = 1. Related to the two-dimensional
case, let us equivalently show the weak existence and uniqueness of the
associated continuum non-linear polinomial wave equation in the domain
(−a, a) × R+ .
∂ 2 U (x, t) ∂ 2 U (x, t)
− + g(U (x, t))2k−1 = 0 (A-1)
∂2t ∂2x
U (−a, t) = U (a, t) = 0
U (x, 0) = U0 (x) ∈ H 1 ([−a, a]) (A-2)

Ut (x, 0) = U1 (x) ∈ L2 ([−a, a]) (A-3)

Let us consider the Galerkin approximants functions to eq. (A-2)–
eq. (A-3) as given below

n
π
U n (t) ≡ U (t) sen x (A-4)
a
=1

Since there is a γ0 (a) positive such that

d2
− U n, U n ≥ γ0 (a)(U n , U n )H 1 ([−a,a]) (A-5)
d2 x L2 ([−a,a])
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 188

188 Lecture Notes in Topics in Path Integrals

we have the a priori estimate for any t

0 ≤ ϕ(t) ≤ ϕ(0) (A-6)
with
1 ˙
U n (t)2L2 + γ0 (a)U n 2H 1
ϕ(t) =
2
1
+ U n 2k
L2 (A-7)
2k
As a consequence of the bound eq. (A-7), we get the bounds for any
given T (with Ai constants)
sup essU n 2H 1 (−a,a) ≤ A1 (A-8)
0≤t≤T

sup essU˙ n 2L2 (−a,a) ≤ A2 (A-9)

0≤t≤T

sup essU n 2k

L2k (−a,a) ≤ A3 (A-10)
0≤t≤T

By usual functional - analytical theorems on weak-compactness on

Banach-Hilbert Spaces, one obtains that there is a sub-sequence U n (t) such
that for any finite T
∗
−−−−−−−−−−→
U n (t)(weak − star) U (t) in L∞ ([0, T ], H 1 (−a, a)) (A-11)
∗
−−−−−−−−−−→
U˙ n (t)(weak − star) v(t) in L∞ ([0, T ], L2(−a, a)) (A-12)
∗
−−−−−−−−−−→
U n (t)(weak − star) p(t) in L∞ ([0, T ], L2k (−a, a)) (A-13)
At this point we observe that for any p > 1 (with Ãi constants) and
T < ∞ we have the relationship below
T T
p
U n pH 1 (−a,a) dt ≤ T (A1 ) 2 ⇔ U n pL2 (−a,a) ≤ Ã1 (A-14)
0 0

T T
U˙ n pL2 (−a,a) dt ≤ T (A2 ) 2 ⇔ U˙ pL2k (−a,a) dt ≤ Ã2
p
(A-15)
0 0
since we have the continuous injection below
H 1 (−a, a) → L2 (−a, a) (A-16)

L2k (−a, a) → L2 (−a, a) (A-17)

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 189

On the Rigorous Ergodic Theorem for a Class 189

As a consequence of the Aubin-Lion theorem ([4]), one obtains straight-

forwardly the strong convergence on LP ((0, T ), L2 (−a, a)) togheter with
the almost everywhere point wise equalite among the solutions candidate
t
U n −→ U (t) = p(t) = v(s)ds (A-18)
0
By the Holder inequalite applied to the pair (q, k)
Un − U Lq (−a,a) ≤ Un − U 1−θ
L2 (−a,a) Un − U L2k (−a,a)
θ
(A-19)
with 0 ≤ θ ≤ 1
1 1−θ θ
= + (A-20)
q 2 2k
in particular with q = 2k −1, one obtains the strong convergence of /Un (t) 0in
the general Banach space L∞ ((0, T ), L2k−1 (−a, a)), with θ = 1−k
k 3−2k
2k−1
As a consequence of the above obtained results, one can pass safely the
weak limit on
C ∞ ((0, T ), L2 (−a, a))
$ '
d2 d2 2k−1
lim (U n , v)L 2 (−a,a) + − U n, v + g(U n , v)L2 (−a,a)
n→∞ d2 t d2 x L2 (−a,a)

d2 d2 2k−1
= 2
(U , v)L2 (−a,a) + − 2 U , v + g(U , v)L2 (−a,a) = 0
d t d x L2 (−a,a)
(A-21)
∞
for any v ∈ C ((0, T ), L (−a, a))
2

At this point we sketchy a somewhat rigorous argument to prove the

problem’s uniqueness.
Let us consider the hypothesis that the finite function
((U )2k+1 (x, t) − (v)2k+1 (x, t))
a(x, t) = (A-22)
(U (x, t) − v(x, t))
is essentially bounded on the domain [0, ∞) × (−a, a) where U (x, t) and
v(x, t) denotes, two hypothesized different solutions for the 2D-polinomial
wave equation eq. (A-1)–eq. (A-3). It is straightforward to see that its
difference W (x, t) = (U − v)(x, t) satisfies the “Linear” wave equation
problem
∂2W ∂2W
− + (a W )(x, t) = 0 (A-23)
∂2t ∂x
W (0) = 0 (A-24)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 190

190 Lecture Notes in Topics in Path Integrals

Wt (0) = 0 (A-25)
1 2
At this point we observe the estimate (where H → L !)
1 d d
 W 2L2 (−a,a) + W 2H 1 (−a,a)
2 dt dt
dW
≤ aL∞ ((0,T )×(−a,a)) × W L2 (−a,a) ×  L2 )−a,a)
dt
dW 2
≤M   2 + W 2L2 (−a,a)
dt L (−a,a)
dW 2
≤M   2 + W 2H 1 (−a,a) (A-26)
dt L (−a,a)
which after a application of the Gronwall’s inequalite give us that
d
 W 2L2 (−a,a) + W 2H 1 )−a,a) (t)
dt
dW 2
≤   2 (0) + W 2H 1 (−a,a) (0) =0 (A-27)
dt L (−a,a)
which proves the problem’s uniqueness under the not proved yet hypothe-
sis that in the two-dimensional case (at least for compact support infinite
differentiable initial conditions)
sup a(x, t) ≤ M. (A-28)
x∈(−a,a)
t∈[0,∞)

It is thus expected (but not proven) that the associated invariant mea-
sure would be given by
$ '
a 1
dν d(− d2 −1
)
μ[X(σ)]e−β{g 0
(X(σ))2k dσ}
F (X(σ)) . (A-29)
−a D ((0,1)) dx2

2
d −1
Here the formal distributional operator (− dx 2) acts on D((0, 1)) to

D ((0, 1)) through the rule
−1
d2
− : D((0, 1)) −→ DT ((0, 1))
dx2 2
(− d 2 )−1
dx
(A-30)
f −→ T [f ]
Namely
 −1
1 1
d2
T [f ](g) = T−(( d2 −1
) f)
(g) = f (s) − (s, t)g(t) dsdt.
dx2 0 0 dx2
(A-31)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 191

On the Rigorous Ergodic Theorem for a Class 191

Note that through the Minlos Theorem, one has rigorously ([4]):
 
1
exp − T(− d2 )−1 (j) (j) = d(− d2 )−1 μ(X) exp {i X(j)} .
2 dx2 D ((0,1)) dx2

(A-32)
The above displayed formulae is our mathematical re-wording of our
final comments on section 4 of this paper.

9.7 Appendix B: The Ergodic theorem for Quantized wave

propagations

In this appendix B we intend to present our results on the quantum statis-

tical case, just for completeness of the exposition of our ideas and proposed
mathematical technique.
In quantum theory the equations of motions are operator equations in
terms of a given fixed self-adjoint Hamiltonian Ĥop (p̂i , q̂i )
dq̂i (t) i
= [Ĥop , q̂i ](t) (B-1)
dt 
dp̂i (t) i
= [Ĥop , p̂i ](t). (B-2)
dt 
Added of initial operatorial conditions
(0) (0)
q̂i (0) = q̂i ; p̂i (0) = p̂i . (B-3)
Proceeding exactly as in section 3 of this paper, the self-adjoint in-
finitesimal generator of the time-evolution quantum mechanical operator L̂
satisfy the quantum mechanical Poisson bracket on the system observable
algebra of the trace class operator fˆ(p̂i , q̂i ). Namely
L(fˆ(p̂i , q̂i )) = [fˆ(pi , qi ), Ĥop ]. (B-4)
By imposing again the additive bounded operator Boltzam behavior for
the operatorial kernel for the linear functional η̂f acting on trace class oper-
ator system observable in the norm topology, one arrives at the N -particle
quentum mechanical ergodic theorem for a given quantum mechanical sys-
tem observable described by a trace class operator Ô(p̂i , q̂ i ) ([12], [13])
$ '
T
1
lim T r{Ô(p̂i (t), q̂ i (t))}dt
T →∞ T 0

> ?
= T r(H⊗H⊗...H)n e−β Ĥ(p̂i ,q̂i ) Ô(p̂i , q̂i ) T r(H⊗H⊗...H)n {e−β Ĥ(p̂i ,q̂i ) }.
(B-5)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 192

192 Lecture Notes in Topics in Path Integrals

For a second-quantized Klein-Gordon field associated to eqs. (A-32)–

(B-3), the wave field has the operatorial expansion

N
Û (x, t) = q̂i (t)φi (x). (B-6)
i=1
Here q̂i (t) are operators acting on the Hilbert space of N -free particles
(H ⊗ H ⊗ . . . H)N and φi (x) are the eigenfunctions of the T(Ω,α,m2 ) operator
(see eqs. (A-36)–(B-6)).
Proceeding as in the bulk of this paper, one arrives at the Bosonic Bell
functional integral representation ([11]) for the quantum ergodic theorem
$ '
T
lim T r{Ô(Û (x, t))}dt
T →∞ 0
$ '
T
= lim lim T r{Ô(q̂1 (t), . . . , q̂N (t))}dt
N →∞ T →∞ 0

1>
= d(− d2 +T −1 )−1
μ(φ(x, t))
2 L2 ([0,β]×RD ) dt2 Ω,α,m2
φ(x,t+β)=φ(x,t)
$ % &'
β ?
× exp − V (φ(x, t))d x dt D
× O(φ(x, t)) . (B-7)
0 RD

Here the cilindrical measure d(− d2 +T −1 )−1

μ(φ(x, t)) is defined by its
dt2 Ω,α,m2
generating functional
⎧ E F ⎫
⎨ 1 d2
−1 ⎬
−1
Z(j) = exp − j, − 2 + TΩ,α,m j
⎩ 2 ⎭
2
dt
L2 ([0,R]×RD )

 
= dμ(φ(x, t)) exp iφ(x, t), j(x, t)L2 ([0,R]×RD ) . (B-8)
d2
(− dt2 +T −1 2 )−1
Ω,α,m

Note that we have the bound

$ −1
'
d2 −1
T rL2 ([0,β]×RD ) − 2 + TΩ,α,m 2
dt

∞
 −[2α+ D2 ]
 n2 π 2 1 D D
≤ 2
m + 2 × B ,α− <∞ (B-9)
n=u
β 2 2 2
which is convergent for α > D
2 as supposed.
A more detailed analysis of the quantum case will appears elsewhere.
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 193

On the Rigorous Ergodic Theorem for a Class 193

9.8 Appendix C: A Rigorous Mathematical proof of the

Ergodic theorem for Wide-Sense Stationary Stochastic
Process

Let us start our appendix by considering a wide-sense mean continuous

stationary real-valued process {X(t), −∞ < t < ∞} in a probability space
{Ω, dμ(λ), λ ∈ Ω}. Here Ω is the event space and dμ(λ) is the underlying
probability measure.
It is well-know that one can always represent the above mentioned wide-
sense stationary process by means of a unitary group on the Hilbert Space
{L2 (Ω), dμ(λ)}. Namely [in the quadratic-mean sense in Engineering jar-
gon]
+∞
X(t) = U (t)X(0) = eiwt d(E(w)X(0)) = eiHt (X(0)) (C-1)
−∞

here we have used the famous spectral Stone-theorem to re-write the as-
sociated time-translation unitary group in terms of the spectral process
dE(w)X(0), where H denotes the infinitesimal unitary group operator U (t).
We have supposed too that the σ-algebra generated by the X(t)-process is
the whole measure space Ω, and X(t) is a separable process.
Let us, thus, consider the following linear continuous functional on the
Hilbert (complete) space {L2 (Ω), dμ(λ)} - the space of the square integrable
random variables on Ω
T
1
L(Y (λ)) = lim dtE{Y (λ) X(t, λ) }. (C-2)
T →∞ 2T −T

By a straightforward application of the R.A.G.E. theorem, namely:

$ '
T
1 −iwt
L(Y (λ)) = dμ(λ)Y (λ) lim dte dE(w)X0 (λ)
Ω T →∞ 2T −T

T
1
= lim E{Y e−iHt X}dt
T →∞ 2T −T

= dμ(λ)Y (λ) dE(0)X(0, λ)

Ω
= E{Y (λ) PKer(H) (X(0, λ))}. (C-3)
Here PKer(H) is the (ortoghonal projection) ou the kernel of the unitary-
group infinitesimal generator H (see eq. (C-1)).
By a straighforward application of the Riesz-representation theorem for
linear functionals on Hilbert Spaces, one can see that PKer(H) (X(0)) dμ(λ)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 194

194 Lecture Notes in Topics in Path Integrals

is the searched time-independent ergodic-invariant measure associated to

the ergodic theorem statement, i.e. For any square integrable time inde-
pendent random variable Y (λ) ∈ L2 (Ω, dμ(λ)), we have the ergodic result
(X(0, λ) = X(0).
T
1
lim dtE{X(t)Y } = E{PKer(H) (X(0))Y }. (C-4)
T →∞ 2T −T

In general grounds, for any real bounded borelian function it is expected

the result (not proved here)
T
1
lim dtE{f (X(t))Y } = E{PKer(H) f (X(0)) · Y }. (C-5)
T →∞ 2T −T

For the auto-correlation process function, we still have the result for the
translated time ζ fixed (the lag time) as a direct consequence of eq. (C-1)
or the process’ stationarity property
T
1
lim dtE{X(t)X(t + ζ)} = E{X(0)X(ζ)}. (C-6)
T →∞ 2T −T

It is important remark that we still have the probability average inside

the ergodic time-averages eqs. (C-4)–(C-6). Let us call the reader attention
that in order to have the usual Ergodic like theorem result - without the
probability average E on the left-hand side of the formulae, we proceed [as
it is usually done in probability text-books] by analyzing the probability
convergence of the single sample stochastic-variables below [for instance]
T
1
ηT = f (X(t))dt (C-7)
2T −T

T
1
RT (ζ) = dtX(t)X(t + ζ). (C-8)
2T −T

It is straightforward to show that if E{f (X(t))f (X(t+ζ))} is a bounded

function of the time-lag, or, if the variance below written goes to zero at
T →∞
T T
1
σT2 = lim dt1 dt2 [E{X(t1 )X(t1 + ζ)X(t2 + ζ))
T →∞ 4T 2 −T −T

−E{X(t1 )X(t1 + ζ)}E{X(t2 )X(t2 + ζ)))] = 0, (C-9)

one has that the random variables as given by eqs. (C-7)–(C-8) converge
at T → ∞ to the left-hand side of eqs. (C-4)–(C-6) and producing thus an
ergodic theorem on the equality of ensemble-probability average of the wide
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 195

On the Rigorous Ergodic Theorem for a Class 195

sense stationary process {X(t), −∞ < t < ∞} and any of its single-sample
{X(t), −∞ < t < +∞} time average
T
1
lim dtf (X(t)) = E{PKer(H) (f (X0 ))} = E{f (X(t))} (C-10)
T →∞ 2T −T

T
1
lim dtX(t)X(t + ζ) = E{X(0)X(ζ)}
T →∞ 2T −T
= E{X(t)X(t + ζ)}
= RXX (ζ) (C-11)

9.9 References

[1] W. Humzther, Comm. Math. Phys. 8, 282 -299, (1968).

[2] Luiz C.L. Botelho, Lecture Notes in Applied Differential Equations
of Mathematical Physics, World Scientific, Singapore, 2008.
[3] Y a. G. Sinai, Topics in Ergodic Theory, Princeton University
Press (1994).
[4] Luiz C.L. Botelho, Advances in Mathematical Physics, Volume
2011, Article ID257916.
[5] Luiz C.L. Botelho, RandomOper. Stoch. Eq., 301–325, (2010).
[6] O.A. Ladyzhenskaya, The Boundary Value Problems of Mathemat-
ical Physics, Applied Math Sciences, 49, Springer Verlag, (1985).
[7] Y. Yamasaki, Measures on Infinite-Dimensional Spaces, Series in
Pure Mathematics, World Scientific, vol (5), 1985.
[8] Luiz C.L. Botelho, Random Oper. Stoch Eq. DOI10.1515/Rose
2011-020.
[9] V. Rivasseau, From perturbative to constructive renormalization,
Princeton University, (1991).
[10] James Glimm and A. Jaffe, Quantum Physics, a functional integral
point of view, Springer Verlag (1981).
[11] Luiz C.L. Botelho Mod. Phys. Lett. B35, 391, (1991).
[12] G. Mackey, The Mathematical Foundations of Quantum mechanics,
Benjamin, N.Y., (1963).
[13] A Gleason, J. Math. Mech. 6, 885–894, (1957).
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 197

Chapter 10

A Note on Feynman-Kac Path

Integral Representations for Scalar
Wave Motions

We present Feynman–Kac path integrals representations for scalar wave

motions on variable medium. The main new points on them is about their
rigorous mathematical validity on the space of continuous functions van-
ishing at infinity, besides of possesing intrinsically the physical property of
causal wave field propagation, thus solving mathematically the long stand-
ing problem on the subject of automatically leading to causality wave prop-
agation.

10.1 Introduction

It is well-known that the rigorous Feynman–Kac representaion provides

the solution on L2 (RN ) for the diffusion linear problem ([1]). What is not
known yet is to generalize the Feynman–Kac path integral representation
for hyperbolic problems where the governing partial differential equation is
of the second order in time. Althought there have been several attempts
to arrive at some mathematically feasible result, none of these has lead to
path integrals representations with the causal (retarded) wave propagation
already built in on the path integral scheme.
Our aim in this mathematical rigorous methods oriented paper is to
explore the Feynman–Kac representation, namely the Trotter–Kato Theo-
rem, but now on the context of the Banach Space E = C0 (RN ), to propose
through the use of the Laplace transformation in time; a causal Feynman–
Wiener path integral representation for the Cauchy problem associated to
the linear wave equation in general medium with a spatially varying in-
dex of refraction. These studies are presented on Section 2 and Appendix
A. On Appendixes B, C, D, E we present some results which may be of
some usefullness for understanding from a calculational point of view, our

197
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 198

198 Lecture Notes in Topics in Path Integrals

proposed causal Feynman-Viener path integral for linear wave equation

with variables coeficients.

10.2 On the path integral representation

One of the most important results on the subject of path integrals represen-
tations for wave propagations are those related to the use of rigorous Wiener
measure in infinite dimensional functional spaces ([1]). In this note we in-
tend to present these path integrals results on the context of the Banach
functional space of continuous functions on RN , vanishing at the point ∞:
C0 (RN ) = {f (x) ∈ Cb (RN ) | lim (f (x) = 0}, instead of the usual L2 (RN )
|x|→∞
framework of the usual Hilbert Space setting ([2]).
As a first important mathematical remark, we recall that we still have
the famous Trotter-Kato product formula in this more ample setting of
Banach Spaces as C0 (RN ) for the class of those so called contraction semi
groups on Banach Spaces, see ([2]).

Theorem (Trotter–Kato). Let A and B be the generators of contraction

semigroups on a given Banach Space (E, || ||) ([1]). Suppose that the
closure on D(A) ∩ D(B) associated to the operator sum A + B generates
a contraction semi-group on (E, || ||). Then for all x ∈ E fixed (strong
convergence in the Banach Space norm), we have the result
N  n  N
N t t N
lim N
N exp(−t(A + B)) − exp − A exp − B (x)N =0
N
n→∞ n n E
(10.1)

An immediate application of the mathematical result Equation (10.1)

can be given to problems of diffusion on C0 (RN ). Let us define the “posi-
tive Laplacian” Operator − 12 Δ as the operator acting on the Banach Space
E := (C0 (RN ), || ||sup ) through the closure of the accretive operator − 12 Δ,
originally defined by acting on S(RN ). Now one can show that exp(− 12 tΔ)
is a contractive semi group on E ([1]). It is another result on the sub-
ject that if a given strictly positive bounded continuous function on RN ,
V (x) ∈ Cb (RN ), (V (x) ≥ V0 > 0), one still has that − 12 Δ + V generates a
contractive semigroup on C0 (RN ). As a direct consequence of the Trotter–
Kato Theorem applied to the Laplacean operator, one has the Feynman
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 199

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 199

result for the diffudion propagator (for t ≥ 0)

1
exp(−t(− Δ + V ))(f )(x)
2
− 32 N
(uniform convergence on RN ) 2πt
= lim (10.2)
n→∞ n
 
× exp(−Sn (x; x1 , . . . , xn ; t))f (xn )dN xn . . . dN x1
RnN
with the discreticized Euclidean Feynman action
% &
n
t 1 |xi − xi−1 |
2
Sn (x, x1 , . . . , xn ; t) = + V (xi ) . (10.3)
i=1
n 2 t/n
At this point, one can represent rigorously Equation (10.2) through
a Feynman-Kac-Wiener path integral. (By using the Feynman physicist
notation for the propagator kernel)
9   :
1
x exp −t − Δ + V y
2

(pointwise)
X(t)=x ;  t dX 2 
= DF (X(σ))e−1/2 0 ( dσ ) (σ) dσ
X(0)=y

t
× exp − V (X(σ)) dσ (10.4)
0

Note that the Wiener measure (with two marked fixed-intercept points
(x, y)) is written in the Feynman suggestive form of infinite product of usual
weighted Lebesgue measures of sample trajectories connecting the marked
points (x, y), i.e. ([1])
⎛ symbolic math
⎞
+
⎛ ,- ⎞.
rigorous path measure
+ ,- . ⎜+symbolic
,-
math
.  ⎟
⎜ t
2 ⎟
⎜ 1 dX ⎟
d Wiener
μ[X(σ)] = ⎜ dX(σ) exp ⎝− (σ) dσ ⎠⎟ .
⎜ 2 0 dσ ⎟
⎝0≤σ≤t ⎠

(10.5)
We now intend to apply the above rigorous mathematical results to
second order hyperbolic wave motion with datum in C0 (RN ) ([3]). Let us
thus write the dynamical scalar wave equation governing our dynamics:
1 ∂2 1
U (x, t) = ΔU (x, t) + F (x, t)
C 2 (x) ∂t2 2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 200

200 Lecture Notes in Topics in Path Integrals

U (x, 0) = f (x)

Ut (x, 0) = h(x) (10.6)

Here the spatially slowly variable wave field velocity is supossed to be
given by the continuous bounded function on RN ; C 21(x) = C 2 q(x)
1
, with
0
C2
C0 denoting a reference wave velocity medium and 1/q(x) := C 2 (x)
0
:=
/ 2 0
C̄ (x) 2
1 + ε C2 here ε << 1 with C̄ (x) denoting the “perturbation” wave
0
field velocity (C 2 (x) = C02 + εC̄ 2 (x)).
Since we are now on the context of Banach Spaces C0 (RN ), one should
naturally consider the Laplace transformed problem in relation to the time
(and leading thus to time forward wave propagations automatically). It is
possible to show rigorously that any causal solution of the Cauchy problem
Equation (10.6) U (0, t) is of exponential order in time.
∞
< (x, s) =
U e−st U (x, t) dt. (10.7)
0

s2 < C2 < s h(x)

U (x, s) = 0 ΔU (x, s) + f (x) + + C02 F<(x, s). (10.8)
q(x) 2 q(x) q(x)
s −C 2 2
Let us now note that 2 0 Δ + q(x) generates a contractive semigroup on
N
C0 (R ), with the frequency s being a real number. As a consequence of the
above displayed remarks, one has the following operatorial Feynman–Kac
path integral representation
−1
< y; s) := s2 C2
G(x, − 0Δ (x, y)
q(x) 2
9 1  2  1 :
∞ 1 s C2 1
dζ x 11exp −ζ 1y
pointwise
= − 0Δ 1
0 q(x) 2
$ 
∞ X(ζ)=x ζ 2
pointwise 1 d
= dζ DF [X(σ)] exp − X(σ) dσ
0 X(0)=y 2C02 0 dσ
 '
ζ
dσ
exp −s 2
0 q(X(σ))
(10.9)
The above written Feynman-Kac path integral is the result we are look-
ing for as a first step of our study. Note that the complete expression for
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 201

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 201

the scalar wave field is now given through the inverse Laplace transform at
least in the Schwartz Distributional sense ([2])1 (see also Appendix D).

U (x, t) = L−1 <

D (R+ ) (U (x, s))
  
−1 s h
= LD (R+ ) G̃(x, y; s) 2 N
f + + C0 F̃ (·, s) (y)d y
RN q q
(10.10)
It is worth call attention again that since one has a Cauchy initial value
problem, one in principle does not have any problem with imposing causal-
ity on the obtained wave field configurations. Opposite to previous studies
on the subject ([3], [4] and appendix C). Note also that the “operatorial”
representation Equation (10.10) is a good candidate for numerical-computer
oriented calculations, after time and space suitable discretizations ([4]).
(See appendix B).
For the Mathematical oriented readers, one has the rigorous Wiener
path integral representation (associated to the diffusion “positive”
Laplacean − 21 Δ) for the solution of the wave Equation (10.6) in the fre-
quency domain (see the ref. [1] for Mathematical notation).
The below written rigorous mathematical formulae will not be used in
what follows.
%  &
∞ ζ
< dσ
U (x, s) = dζ d(x,ζ) μ[ω(σ)] exp −s
Wiener 2
0 C([0,ζ],ṘN ) 0 q(C0 ω(σ))

 
f (C0 ω(ζ)) h(C0 ω(ζ))
× s + + C02 F< (C0 ω(ζ), s) (10.11)
q(C0 ω(ζ)) q(C0 ω(ζ))
note that full mathematical conditions on the initial datum leading to
each fixed x ∈ RN to a s-function which admits non-distributional in-
verse Laplace transforms remains on open problem as far as we know.
Certainly no problem if one choose datum such that U < (0, s) is an ana-
lytic function s = σ + iζ in the half-plane Real (s) > ∞, vanishes at
(s) → ∞ at all half-planes Real(s) > α + δ(δ ∈ R+ ) uniformly with respect
to arg(s) = arctg( σs ), besides the following integrability condition ([2]).
+∞
|U (·, α + σ + iζ)|dζ < ∞
−∞
1 The definition of the root square of the “positive” Laplacian inRN is done by
the usual (unitary) Fourier Transformation on L2 (RN ), namely: − 12 Δ(f )(x) :=

f −1 [ 12 |k|f(k)].
ˆ
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 202

202 Lecture Notes in Topics in Path Integrals

Just for completeness of our note, let us write now a non rigorous Feyn-
man path integral representation for our (already built in) casual wave
propagation problem (see Equation (10.9) and refs. [3], [4])
−1
< y, s) = s2 C2
G(x, − 0Δ (x, y)
q(x) 2
9 1  1 :
∞ 1 s2 C02 1
= −i 1
dζ x 1exp iζ − Δ 11 y
0 q(x) 2
=0
+ ,- .
π s2 C02
− δ spectral − Δ
i q(x) 2
 % ζ
∞ X(ζ)=x i X 2
( ddσ ) (σ)dσ
= −i F 2C02 0
dζ D [X(σ)] e
0 X(0)=y
ζ 
−is (
2 dσ
)
× e 0 q(X (σ))
(10.12)

At this point of our note some comments appears to be relevant for an-
alytical manipulations. The first comment is related to the result already
expressed by our Feynman-Wiener and Feynman path integrals representa-
tions are solutions for obtaining full optics propagation from the idealized
geometrical optics, with its idealized rays paths given by the stationary so-
lutions of the geometrical optics Lagrangeans. Namely (see Equation (10.9)
and Equation (10.12) respectivelly and chapter twenty, first reference in [3]).
%  &
ζ 2
1 d 1
LWiener(X(σ), ζ) = X(σ) + dσ (10.13)
0 2C02 dσ q(X(σ))
%  &
ζ 2
1 d 1
LFeynman(X(σ), ζ) = X(σ) − dσ (10.14)
0 2C02 dσ q(X(σ))
The second comment is that our path integrals representations open a
good deal in principle to understand, through semi-classical and W.K.B.
methods, the entire range of interference and diffraction (caustics) electro-
magnetic scalar (optic) phenomena ([3], [5]) (see appendix A for the wave
propagation acoustic case).
As a last comment, we wishe to point out that alternativelly to the path
integral representation Equation (9), one has naturally associated to it, the
following initial-value diffusion equation on RN
⎧ / 0
⎨ ∂ U (x, ζ) = + C02 Δ − s2 U (x, ζ)
∂ζ s 2 q(x) s
(10.15)
⎩Us (x, 0+ ) = δ (N ) (x − y)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 203

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 203

which in some analytical contexts may be as useful as to solve the associated

path integral Equation (10.9).
Let us show the last assertive of these first set of results by writing
an asymptotic behaviour for our scalar wave field U (x, t) for large times
t → ∞ in a compact domain Ω. Firstly we observe that large times t → ∞
the wave field behavior is equivalent to analyze small frequences s → 0+
on Equation (10.15) and thus large “proper-time” limit ζ → ∞ on the
—“diffusion problem” Equation (10.15) with Dirichlet conditions on Ω,
i.e.,
ζ→∞
Us (x, ζ) ∼ eE0 (s)ζ φ0 (x, s)φ∗0 (x, s) (10.16)
where E0 (s) is naturally the loweste eigenvalue of the eigenvalue problem
on Ω
C2 s2
− 0Δ+ φk (x, s) = −Ek (s)φk (x, s). (10.17)
2 q(x)

φk (x, s)|∂Ω = 0.
A formal first-order perturbative calculation lead us to the result
(1) k 2 C02
Ek (s) = + s2 Ĉ (10.18-a)
2
1
Ĉ = dN x (10.18-b)
Ω q(x)
⎡ (1)
⎤
:=ψ̄k (x)
⎢+ ,- .⎥
⎢  ⎥
(1) 2s2 ⎢ N  Q̂(k − k) +ik x ⎥
φk (x.s) =e ikx
+ 2 ⎢ d k 2 e ⎥ (10.18-c)
C0 ⎢ Ω k − k 2 ⎥
⎣ ⎦

1 i(k−k )x
Q̂(k − k  ) = dN x e . (10.18-d)
Ω q(x)
We have thus the straightforward results (with C := 1 + Ĉ)
 % &
t→∞ −1 1
G(x, y, t) ∼ LD (R+ ) k2 C 2
eik(x−y)
Ĉs2 + 2 0
% &
2( Cs0 )2
+L−1 (e−iky ψ̄k (x) + eikx (ψ̄k (y))∗ )
(1) (1)
D (R+ ) k2 C02
Ĉs2 + 2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 204

204 Lecture Notes in Topics in Path Integrals

 = 
1 2 |k|C0
= sin 4 t eik(x−y)
C0 Ĉ 2Ĉ
 √
2 |k|C02
+ − 4 |k| sin t + δ(t)
C0 Ĉ 2
/ 0
× e−iky ψ̄k (x) + eikx (ψ̄k (y))∗ .
(1) (1)
(10.19)

Here K is the wave vector on the bottom of the spectrum of the well-
defined free problem on Ω
C02 |K|2 2 iK·x
− Δ (eiK·x ) = C0 (e ). (10.20)
2 2
Extensive applications of this direct problem in scalar wave propagation
needs the explicity form of the wave field scatter q(x) = C 2 (x)/C02 and will
be left to geophysics oriented papers to appear elsewhere.
Finally, we would like to address ourselves to the important applica-
tion of using Feynman-Kac path integrals representations to the classical
problem of determining the electromagnetic field strength E[(x, y); z, t] orig-
inating from a monochromatic point source and propagating in a medium
characterized by a deterministic refractive index η(x, y, z) in the half-space
R+3
= {(x, y, z) | 0 ≤ z < ∞, (x, y) ∈ R2 }. In the well-known paraxial
approximation ([6]), this paraxial pulse is supposed to have explicitly the
following structural form
E(K,Ω) ((x, y); z, t) = Real{A[(x, y); z] ei(kz−Ωt) } (10.21)
where the (paraxial) amplitude A[(x, y), z] satisfies the following two-
dimensional Schrodinger initial value equation with the depth coordinate z
playing the role of the time variable
∂ 1
i + Δ(x,y) − k(1 − η(x, y, z) A[(x, y); z] = 0. (10.22)
∂z 2k
Here, the initial date condition is supposed to be known, namely:
A[(x, y); z → 0+ ] = B(x, y). (10.23)
Instead of the full set Eqs. (10.22)–(10.23), let us firstly solve them
under the depth-independent condition on the refractive index η(x, y, z) :=
η(x, y). In order to implement such analysis, let us consider the Feynman
propagator like problem associated to Equation (10.22) (k = 12 , r = (x, y))
∂ψ 1
i = − Δr ψ + V (r)ψ + iδ(r − r )δ(z − z  ). (10.24)
∂z 2
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 205

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 205

We added as a mathematical hypothesis the condition that ψ(z, · ) is

an analytical function coming from a Fourier transform of a function with
compact support (a sort of “frequency” bounded limited signal - [2])
Λ
−1 1
ψ(z, r) = Fω→z [ψ̃(ω, r)] = √ e−iωz ψ̃(ω, r) dω. (10.25)
2π −Λ

Here Λ is such that supp ψ̃(ω, · ) ⊂ [−Λ, Λ].

The operatorial solution of Equation (10.24) is thus given by the math-
ematical well-defined expression (note now the functional equality on the
Hilbert Space L2 (R2 ) sense)
$ E 1 −1 11 F'
1 1
2 2
 L (R ) +iωz  1 1 
ψ̃(ω, r, r ) = +i e r1 Δ − (V − ω) 1r
1 2 1
 ∞ 9 1 1 :
2 2 1 1 1
L (R )
= +ie iωz  1
dζ r 1 exp ζ Δ − (V − ω) 11 r
0 2

$ %
∞ r(ζ)=r
L2 (R2 ) 
= +ieiωz dζ eωζ DF [r(σ)]
0 r(0)=r 

ζ 2
1 dr
× exp − (σ) dσ
2 0 dσ
 &'
ζ
× exp − V (r(σ)) dσ (10.26)
0

In the depth dependent notation, we re-write Equation (10.26) in the

more invariant form
ψ[(z, r), (z  , r )]

1 > Λ

; ∞
= i√ dω e−iω(z−z )
dζ e+ωζ
2π −Λ 0

/ r(ζ)=r / 1 ζ
dr 2 0
× DF [r(σ)] exp − ( ) (σ) dσ
r(0)=r  2 0 dσ

/ ζ 00?
× exp − V [r(σ)] dσ . (10.27)
0
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 206

206 Lecture Notes in Topics in Path Integrals

Since the cut-off Λ is a finite number, one can apply the Fubbini theo-
rem for exchange the (ω, ζ) integrations and obtaining thus the more invari-
ant mathematically rigorous path integral representation for at least with
V (r) ∈ L2 (R2 ) + L∞ (R2 ).
Now if one takes formally the limit of cutt-off removing of Λ → ∞, one
could obtain the complete Feynman-Kac path integral representation below
written.
Since
  
1 e−Λζ e−iΛω (1 − e2Λζ e2iΛω ) D (C)
lim − √ := δ((z − z  ) + iζ).
Λ→∞ 2π (ζ + iω)
(10.28-a)
It yields thus
> ∞
ψ[(z, r), (z  , r )] = +i dζ δ((z − z  ) + iζ)
0

; r(ζ)=r / 1 ζ
dr 2 0
× DF [r(σ)] exp − ( ) (σ) dσ
r(0)=r  2 0 dσ

/ ζ 0?
× exp − V [r(σ)] dσ . (10.28-b)
0

If we now introduce a formal parametrization path pure imaginary σ-

time (resulting on the so called Feynman “quantum” paths!) through the
mathematical formal relationship
R(σ̄) := r(−iσ) (10.29)
one gets the usual operational symbolic Feynman path integral, where we
have re-introduced the K-parameter (see the paraxial Equation (10.22))
ψ((z, r); (z  , r ))

r(iz)=r iz 2
1 dr
= D [r(σ)] exp −
F
(σ) dσ
r(iz  )=r  2 iz  dσ

iz
× exp − V (r(σ)) dσ
iz 

R(z)=r
= DF [R(σ)]
R(z  )=r 
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 207

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 207

⎛  ⎞
2
z
1 dR
× exp ⎝+iK (σ̄) dσ ⎠
z 2 dσ̄

z
× exp −iK V (R(σ̄)) dσ̄ . (10.30)
z
For a depth-dependent full case of the potential V (r, x), one should
just follow R.P. Feynman by introducing the z-ordered product inside the
Feynman-Kac path integral Equation (28)
 ∞
ψ[(z, r); (z  , r )] = +i dζ δ((z − z  ) + iζ)
0

r(ζ)=r ζ 2
1 dr
× DF [r(σ)] exp − (σ)dσ
r(0)=r  2 0 dσ
$  '&
ζ
×Tσ exp − V (r(σ)) dσ . (10.31)
0

Or in the usual symbolic Feynman path integral notation of Equation

(10.30)
ψ((z, r); (z  , r ))

> R(z)=r / z
1 dR(σ̄) 2 0
= DF [R(σ̄)] exp + iK ( ) dσ̄
R(z  )=r  z 2 dσ̄
/ / z 0?
×Tσ̄ exp − iK V (R(σ̄), σ̄) dσ̄ . (10.32)
z

Here Tσ̄ (. . . ) means the σ̄-ordered product operation.2

As a pedagogical comment, let us highlight the rigorous mathematical
proof of the cutt-off removing of Λ → ∞.
In the rigorous mathematical notation of ref.[1] in terms of Wiener path
integrals, we define the positive quadratic form in L2 (R2 )
f, QΛ
z−z  gL2 (R2 )

1 > Λ

:= √ dω e−iω(z−z ) × dζ eωζ
2π −Λ
2 At this point the reader should not forget that the purely symbolic Feynman path
integral is just a formal string of symbols for the real mathematically meaningful object
Equation (10.28-b).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 208

208 Lecture Notes in Topics in Path Integrals

;
× dWiener μ[r(σ)]f (r(σ))

/ ζ 0 ?
exp − V (r (σ)) dσ g(r(ζ)) . (10.33)
0
One can show that QΛ z−z  is a closable semi-bounded quadratic form.
As a consequence there is a set of self-adjoint operators H(Λ) such that

f, QΛ
z−z  gL2 (R2 ) = f, exp(−(z − z )H
(Λ)
)gL2 (R2 ) .

By proceeding further, one can shows that H(Λ) converges to a unique

self-adjoint operator H(∞) in the strong resolvent sense. Now one can
see that the Feynman path integral is the famous E. Nelson path integral
representation (see second reference in [1]) for the usual Stone Unitary
group generated by H(∞)
f, Q∞
(z−z  ) gL2 (R2 ) = Equation (30). (10.34)
As a final comment, somewhat oriented to Monte-Carlo (stochastic eval-
uations) of our proposed Feynman-Kac path integrals representations above
written, it is worth remark that by defining a pre-potential W (x), such that
it satisfies the first order eikonal partial differential equation
 N
∂Φ ∂Φ
= V (x) (10.35)
i=1
∂xi ∂xi
where V (x) is a given potential function on C 1 (RN ), then we have that
the solution of the system of ordinary differential stochastic Stratonovich
equations ([4], [7]) [note the all Feynman-Kac path integrals should be asso-
ciated to the stochastic Stratonovich calculus through calculus the famous
Feynman mild-point rule]
RN -Brownian drift
∂Φ + ,- .
dStrat X i (t) = (X  (t))dt + dW i (t)
∂xi
X i (0) = xi (10.36)
defines a diffusion process with (time-invariant) transition density explicitly
given by the following Feynman-Wiener path integral ([4], [5])
p((xi , 0), (y i , t)) = exp(Φ(x) − Φ(y))
$ % &'
X i (t)=y i 2
1 t dX i t
× D [X (σ)] exp −
F i
dσ − i
V (X (σ)) dσ .
X i (0)=xi 2 0 dσ 0
(10.37)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 209

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 209

10.3 Appendix A: The Acoustic Case

In this appendix of complementary nature to the bulk of this note, we

intend to write path integral representations for scalar wave propagation
for more realistic medium of both variable density and variable velocity
and modeling physically acoustic scalar wave propagation ([3]) - second
reference.
In this more general realistic acoustic case, the governing dynamical
wave equation initial value problem takes the following form
1 ∂2 1 F (x, t) · C 2 (x)
P (x, t) = ΔP (x, t) − ∇(nρ(x)) · ∇P (x, t) + .
C 2 (x) ∂t2 2 K(x)
(A-1)
Here ρ(x) is the spatially varying medium density, K(x) the associated
stress-strain bulk modulus (both modeled by strictily positive continuous
functions). The dynamical variable scalar wave field is the non-equilibrium
pressure field, instead of the usual medium vector position ([5]).
In what follows we are going to treat the more general wave equation
initial-value problem, now in the Hilbert Space L2 (RN ) although the point-
wise Banach Space C0 (RN ) can be treated straightforwardly as done in the
first part of this note, but now with initial datum on the more restrict
functional space C0 (RN )
 
1 ∂2U 1
(x, t) = − (−i∇x − a(x)) − V (x) U (x, t) + F (x, t)
2
C 2 (x) ∂t2 2
U (x, 0) = f (x) ∈ L2 (RN )
Ut (x, 0) = g(x) ∈ L2 (RN ). (A-2)
Note that Equation (A-1) can be written in the “Gauge-invariant” form
Equation (A-2) by means of the obvious identification
a(x) = −i(∇(nρ))(x)
1 1
V (x) = ||(∇nρ)||2RN (x) − (Δnρ)(x) (A-3)
2 2
Here all the external imputs (a, V, , g) are functional objects on L2 (RN )
and the external source F (x, t) is such that its Laplace Transform in relation
to the time t. F< (x, s) for each fixed s - belongs to L2 (RN ).
We have now the standard result ([1], Chapter V) for the inverse Laplace
transformed operator kernel
E 1 −1 11 F
1 1 2
1 s 1
x1 (i∇ + a)2 + V + 2 1y
1 2 C (x) 1
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 210

210 Lecture Notes in Topics in Path Integrals

⎧ ⎛  ⎞
L2 (RN )
∞ ⎨ X(ζ)=x
1 ζ
dX
2

= dζ DF [X(σ)] exp ⎝− (σ)⎠

0 ⎩ X(0)=y 2 0 dσ

%
ζ ζ
dX i
× exp −i a(X(σ)) (σ) − (div a)(X(σ)) dσ
0 dσ 2 0

&'
ζ ζ
dσ
− V (X(σ)) dσ − s2 (A-4)
0 0 C 2 (X(σ))
which in terms of the original physical medium parameter density can be
written as of as
% 
∞ X(ζ)=x 2
L2 (RN ) 1 ζ dX
G(x, y, s) = dζ D [X(σ)] exp −
F
(σ) dσ
0 X(0)=y 2 0 dσ

ζ
ρ(y) dσ
× × exp −s2
ρ(x) 0 C 2 (X)(σ))

ζ
× exp +1 (Δnρ)(X(σ)) dσ
0

 &
ζ
1
× exp − ||(∇nρ)|| (X(σ)) dσ
2
(A-5)
2 0

here we have used the Stratonovich formula for implementing a needed by

parts stochastic partial integration
ζ
dX
(∇nρ)(X(σ)) (σ)
0 dσ

= nρ(X(ζ) − nρ(X(0)). (A-6)

It is worth recall that if one had used the It stochastic calculus rule
to define the stochastic integral underlying the Feynman-Kac-Wiener path
integral Equation (A-5); it could lead to a wrong wave equation under the
presence of the vector potential a (see first reference-[3]). As a consequence
one is really using the Stratonovich prescription for the path integral when
involving the presence of an externa vectorial field a(x). And even if this
is not the case, in other words: if one has solely a scalar potential in the
path integral, it does not matter the use of the Stratonovich-Feynman mild
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 211

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 211

point rule or the backward It prescriptions. They are both mathematically

equivalent on the realm of Path Integrals.
Another important point to be remarked is that equality on Equation
(A-5) must be understood in a weak-sense in L2 (RN ). Namely, for any pair
of vectors (f, g) ∈ L2 (RN ), one has the weak equality
f |G(x, y, s)|g
= f | Path integral expression on the righ-hand side of Equation (A-5)|g.
(A-7)

10.4 Appendix B: A Toy model for stable numerics on wave

propagation

Let us start this appendix by considering the following velocity medium

variable wave equation in one-dimension R1 (the generalization to the
higher-dimensional spaces is straightforward ([2])
∂2U d d
(x, t) = C 2 (x)
∂t2 dx dx
U (x, 0) = f (x) ∈ L2 (R) (B-1)

Ut (x, 0) = g(x) ∈ L2 (R).

By re-writing Equation (B-1) in the first order system
∂ U 0 1 U
(x, t) = i (x, t)
∂t π − dx
d d
(C 2 (x) dx )0 π
U (x, 0) = f (x)

Ut (x, 0) = g(x) (B-2)

An (weak-sense) operatorial solution in C([0, t], L2 (R)) is given throught
the Stone theorem ([2]) for each fixed t
C([0, t], L2 (R))    
U - .+ , 0 1 f
(x, t) = exp it (x). (B-3)
π − dx
d d
(C 2 (x) dx )0 g
However in most computer modelling, the initial datum (f, y) is a some-
what “band-limited” process. For instance, since − dxd d
(C 2 (x) dx ) is a self-
2
adjoint operator an L (R), it has a spectral representation
+∞
d d
h, − (C 2 (x) )f L2 (R) = λh, dE(λ)f L2 (R) . (B-4)
dx dx −∞
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 212

212 Lecture Notes in Topics in Path Integrals

As a consequence one should consider all initial datum belonging to a

finite-spectral range i.e., there exists Λ > 0, such that
E([−Λ, Λ])f = f
(B-5)
E([−Λ, Λ])g = g.
The so called finite energy physical initial datum ([2]). For slowly vary-
ing medium velocity, one can take the spectral parameter λ as the wave
vector k.
Since on these stales sub-space of fixed finite energy, the generator group
in Equation (B-3) is effectivelly a bounded operator. So, for short time
propagation one has the rigorous result
U U t→0
(x, t + Δt) − (x, t) ∼
=
π π
U (x + Δx, t) + U (x − Δx, t) − 2U (x, t) 2
i π(x, t)Δt C (x)Δt2C  (x)C(x)
(Δx)2
U (x + Δx, t) − U (x, t)
× Δt . (B-6)
Δx
Here
U (x, 0) = f (x) (B-7)

π(x, 0) = g(x).
After introducing a discreticized space-time R1 ⇔ (nΔ); [0, t] ⇔ (mδ);
U (nΔ, mδ) = Unm ; π(nΔ, mδ) = πnm ; U (nΔ, 0) = f (Λ) (nΔ); π(nΔ, 0) =
g (Λ) (nΔ); one gets the difference scheme governing the discreticized dy-
namics
Unm+1 = iδπnm + Unm

C 2 (nΔ) · δ  m 
πnm+1 =i 2
Un+1 + Un−1 m
− 2Unm
Δ
 (B-8)
2C  (nΔ)C(nΔ) · δ m
+ (Un+1 − Un ) + πn ,
m m
Δ
which is stable in the Von Newmann stability criterium for grid spacements
(Δ, δ) satisfying the “Incertanty relationship”
δ
max |C 2 (x)| <1 (B-9)
x∈R Δ2
1
⇔ Δt < . (B-10)
(Δx)2 ||C 2 (x)||L∞ (R)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 213

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 213

10.5 Appendix C

In this somewhat “advenced calculus” appendix, we show in details how

to salve by means of L. Schwartz distribution theory the classical Cauchy
problem in R3 for the scalar wave equation in a wave field case of constant
velocity
∈S(R3 ×R)
1 ∂ 2 + ,- .
2 2
U (x, t) = ΔU (x, t) + F (x, t)
C ∂t

U (x, 0) = f (x) ∈ S(R3 )

Ut (x, 0) = g(x) ∈ S(R3 ). (C-1)

Since the Fourier Transform is a topological isomorphism on the tem-
pered distributional space S  (R3 ), one easily obtains the wave field as the
L1loc (R3 × R+ ) kernel of a distribution on S  (R3 ). Namely:
S  (R3 ) ĝ(k) sin(Ct|k|)
Fx→k [U (x, t)] := Û (k, t) = fˆ(k) cos(Ct|k|) +
C|k|
 
t
sin(C|k|(t − t ))
−C 2
F̂ (k, t) . (C-2)
0 C|k|
Since in S  (R3 ) sense, we have the Fourier Transform formulae
 
sin(Ct|k|) S  (R3 ) 1
F −1 = {δ(Ct−|x|)−δ(Ct+|x|)} := h1 (|x|, t) (C-3)
C|k| 2C|x|
=
−1 π S  (R3 )
1 d
F [cos(Ct|x|)] · = {δ(|x| − Ct) − δ(|x| + Ct)} := h2 (|x|, t),
2 C|x| dt
(C-4)
one gets the usual formulae, after disregarding by hand the “out-going”
parts of Eqs. (C-3)–(C-4)
S  (R3 )
U (x, t) = (f (x) ∗ hret
2 (|x|, t)) + (g(x) ∗ h1 (|x|, t))
ret

t
−C 2 dt (F (x, t ) ∗ hret 
1 (|x|, t )). (C-5)
0

Note that the equality on the above written Equation (C-5) must be taken
on distributional sense S  (R3 ) (for each test function P (x) ∈ S(R3 ) and for
a fixed time t).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 214

214 Lecture Notes in Topics in Path Integrals

It is worth now point out that if the initial datum f (x) and g(x) are
respectivelly considered as elements on the Sobolev Spaces H2 (R3 ) and
H1 (R), one may give a point function Mathematical Meaning for Equation
(C-2).
cos(Ct|k|)
Û (k, t) = (|k|2 fˆ(k))
|k|2
sin(Ct|k|)
+|k|ĝ(k)
|k|2
 
t
sin(C|k|(t − t ))
−C 2
(|k|F̂ (k, t)) . (C-6)
0 C|k|2
Now one has on L2 (R ), the Fourier Transform inverse formulae
3
 $
D (R) 1 if a ≥ 0
O(a) =
0 if a < 0
  = 1 1
cos(Ct|k|) 1 π d |x| + Ct 11 x − Ct 11
F −1 = − 1 2 1 O(x − Ct)
|k|2 2C|k| 2 dt 2
:= j2 (|x|, t)
(C-7)
  =
sin(Ct|k|) 1 1
F −1 = (1 − O(|x| − Ct))
|k| 2π C|x| (C-8)
:= j1 (|x|, t).
We get thus, the solution on L2 (R3 ) for the wave field
L2 (RN )
U (x, t) = j2 (|x − x |, t)Δf (x ) d3 x

+∞
+ j1 (|x − x |, t)((−Δ)1/2 g)(x ) d3 x
−∞
 t 
    3  
−C 2
j1 (x − x , t )((−Δ) 1/2
H)(x , t ) d x dt . (C-9)
0
Note that if g(x) ∈ H2 (R3 ), then one obtains a full L2 (R3 ) solution,
since now
  =  1 1 
−1 sin(Ct|k|) 2 1 |x| + Ct 11 x − Ct 11
F = − 1 2 1 O(|x| − Ct) . (C-10)
|k|2 π C|x| 2
For the case of initial datum and outputs on space of continuous func-
tions see the Method of Spherical Means in second ref.[2].
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 215

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 215

10.6 Appendix D: The Causal Propagator – The Retarded

Potential

In this short appendix we shall deduce the causal retarded portential in R3

from our proposed path integral representation Equation (9).
We have thus, the chain of equations:
G̃(x, y, s)
$ 
∞ X(ζ)=x ζ
1 dx
= D [X(σ)] exp − 2
F
( )2 dσ
0 X(0)=y 2C0 0 dσ

× exp(−S ζ) dζ
2

⎧ ⎫
⎪
⎪ +
:=G̃s (x,y,s)
,- . ⎪
⎪
⎪ ⎪
∞⎪⎨ 3/2 ⎪
⎬
1 1 −(s ζ)
2
= exp − (x − y)2
e dζ
⎪ 2πC02 ζ
⎪ 2C02 ζ ⎪
⎪
0 ⎪
⎪ ⎪
⎪
⎩ ⎭
3/2 ∞  
1 (x − y)2
ζ −3/2 e−(S ζ) e− ζ
2 1
= dζ
2πC02 0 2C02
$ −1/4
 G '
1 1 (x − y)2 (x − y)2 2
= · × 2 K−1/2 2 S
(2π)3/2 (C0 )3 2C02 S 2 2C02
 √
1 1 √ 1 − 2|x − y|S
= · · 2π · exp .
(2π)3/2 C02 |x − y| C0
(D-1)
Since one has the validity of the above Inverse Laplace Transform on
the sense of D (R+ )
L−1
s→t [e
−as
] = δ(t − a). (D-2)
One obtain the final retarded potential for our wave equation eq(6) on
the main text √
−1 δ(ct − |x − y| 2)
G(x, y, t) = Ls→t (G̃(x, y, s)) = . (D-3)
2πC0 |x − y|
On the usual case of Equation (60) written with the pure Laplacian
(+ 12 Δ → +Δ), one obtains the usual result due originally to Linard-
Wiechiert
3/2 3/2
1 1 (x − y)2
G̃s (x, y, ζ) = exp − . (D-4)
2 2πζC02 4C02
δ(C0 t − |x − y|)
Gret (x, y, t) = (D-5)
4πC0 |x − y|
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 216

216 Lecture Notes in Topics in Path Integrals

10.7 Appendix E: The Causal Propagator – The Damped

Case

Let us take into account in the wave equation initial value problem, the
existence of a damping term. Namely
⎧
⎪
⎨ C 3 (x) Utt = 2 ΔU − νUt + F
1 1
⎪
U (x, 0) = f (x) ∈ C0 (R3 ) (E-1)
⎪
⎪
⎩U (x, 0) = g(x) ∈ C (R3 ).
t 0

Here ν ∈ R+ , denotes the damping constant.

By proceeding as in previous studies presented in this note, we arrive
at the proper-time representation for the problem Green function
3/2  ∞
1
G̃(x, y, s) = ζ −3/2 × exp[−ζ(s2 + νsC02 )]
2πC02 0

1 (|x − y|2 ) 1 1
× exp − = ·
ζ 2C02 (2π)3/2 C03
$ −1/4
%√ &'
|x − y|2 2
× 2 × K−1/2 |x − y|(s + νC0 s)
2 2 1/2
2C02 (x2 + νsC02 ) C0
(E-2)
By noting the L. Schwartz distributional Laplace Transform result
; √  D (R+ ) b d b4 2
L−1 e−a s +sb = e− 2 t −
2
I0 t − a2 O(t − a) (E-3)
da 2
One obtains the final expression for the causal Green function for the
damped case of constant medium wave propagation
 
A νC 2 t
− 20 d ; / 24 2 0
G̃(x, y, t) = e − I0 νC0 t − a 2 O(t − a) (E-4)
|x − y| da
Here the constant A is given explicitly by
= 
2− 4
1
1 π
A= (E-5)
(2π) 3/2 2 C02
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 217

A Note on Feynman-Kac Path Integral Representations for Scalar Wave Motions 217

10.8 References

[1] Barry Simon, Functional Integration and Quantum Physics, Aca-

demic Press, (1979).
– M. Reed & B. Simon, Methods of Modern Mathematical Physics,
Vol II, Academic Press, (1980).
– Ya. A. Butko, M. Grothous and O. G. Smolyanov, Lagrangian
Feynman formulae for second order parabolic equations in bounded
and unbounded domains; Inf. Dim. Anal. Quant. Probability, Rel.
Top V. 13, No 3, 377–392, (2010).
[2] Luiz C.L. Botelho, Random Oper. Stoch. Equ 18, 301–325, (2010).
– Lecture Notes in Applied Differential Equations of Mathematical
Physics, World Scientific, (2008).
[3] L.S. Schulman, Techniques and Applications of Path Integration,
A. Wiley, Interscience, (1981).
– R.B. Schlottmann, Geophys. J. Int., 353–363, (1999).
[4] Luiz C.L. Botelho, Int. J. Theor. Phys. 49, 1396–1404 (2010).
– Methods of Bosonic Path Integrals Representations, Random Sys-
tems in Classical Physics, (2006).
[5] Luiz C.L. Botelho, Physical Review 49E (2), R1003-R1004, (1994).
– Mod. Phys. Lett B13 (11), 363–370, (1999).
– J. Phys. A, Math Yen 34 (12), L131–L137, (2001).
– Mod. Phys. Lett 16B (21), 793–806, (2002).
[6] Luiz C.L. Botelho, Modern Physics Letters 14B, No 3, 73–78,
(2001).
[7] Eugene Wony and Bruce Hajek, Stochastic Processes in Engineer-
ing Systems, Springer-Verlag, (1985).
– Zeev Schuss, Theory and Applications of Stochastic Differential
Equations, Wiley Series in Probability and Mathematical Statistics,
John Wiley & Sons, New York, USA, (1980).
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 11:03:06 AM

December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 219

Chapter 11

A Note on the extrinsic phase Space

path Integral Method for quantization
on Riemannian Manifold Particle
Motions - An application of Nash
Embedding Theorem

Through the use of Nash embedding for Riemann smooth manifolds, we

propose a constrained phase space path integral for quantization of one
particle motion in Riemannian manifold.

11.1 Introduction

The subject of writing path integral representations for describing the quan-
tum propagation of particles (spinorial or scalar) in Riemannian Manifolds
remains as a central theme on the search of consistent framework for the
quantization of the gravitation. In last decades, we have seen the appear-
ance of several important studies on the subject. However all these works
make full recourses for the intrinsic geometrical properties of the Riemann
Manifold where the propagation is supposed to take place (see [1]–[4] for a
short sample list of references).
In this note we propose a somewhat different path integral quantiza-
tion geometrical framework, based on a deep theorem due to Nash that
asserts roughly that every Riemannian metric in a given d-dimensional
C ∞ -manifold {M, gμv (x)} can be always obtained from an immersion
f A : M → RS(d) (f A ∈ C 1 (M ) and rank Dx f A = d) in a suitable
Euclidean space is strictly greater than d (S(d) ≥ 2d − 1) ([5],[6]). As
a consequence one can forseen to write the metric field gμv (x) as

 ∂f A ∂f A def ∂f A ∂fA
S(d)
gμv (x) = ≡ (11.1)
∂xμ ∂xν ∂xμ ∂xν
A=1

219
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 220

220 Lecture Notes in Topics in Path Integrals

here {x} ∈ Dom(f A ), a point of RS(d) (containing by its turn the manifold
chart of M which x belongs for). Note that eq. (11.1) is a non-linear first
order set of Partial Differential Equations for f A (xγ )γ=1,...,d with the source
term gμv (x) given explicitly as input.

11.2 The Phase Space Path Integral Representation

So, let us start our analysis by re-writing the Lagrangian for free-motion
on the manifold in terms of the new extrinsic coordinates.
L(X i ,Ẋ i ) L(QA ,Q̇A )
+ ,- . + ,- .
i j A B
1 ∂X ∂X 1 dQ dQ
L = M gij (X μ (σ)) (σ) = M δAB (σ) (11.2)
2 ∂σ ∂σ 2 dσ dσ
where the new particle “extrinsic” coordinates are given explicitly by
QA (σ) = f A (X 1 (σ), . . . , X d (σ)) (11.3)
with A = 1, . . . , S(d) and we suppose besides that the inverse functional
relations hold true (the usual inverse theorem ([2]) of advanced calculus is
supposed to be at least locally being under application!). Namely:
X i (σ) = Gi (Q(A) (σ)) (11.4)
plus (S(d) − d) smooth constraints on the motion, when viewed on the
non-curved (absolute inertial) referential sistem RS(d)
Φ (QB ) = 0  = 1, . . . , S(d) − d. (11.5)
At the classical level, the intrinsic free motion in the Riemannian Mani-
fold (M, gμν (x)) as given by eq. (11.2) is entirely equivalent to the classical
motion in the extrinsic space RS(d) but now in the full presence of con-
straints given by eq. (11.5).
We have thus that the Free-Manifold Motion equation
⎧  
⎪
⎪ d ∂L ∂L
⎪
⎪ =
⎪ dt ∂(Ẋ )
⎨ i ∂X i
(11.6)
⎪
⎪
⎪
⎪ X i
(σ) = A i
= A
⎪
⎩ Ẋ i (σ) = B i = B

is mathematically equivalent to the motion now under constraints ([4];

pages 45-51) in the extrinsic Euclidean Space RS(d) . We have thus the
new set of Euler-Lagrange equations on the ambient space RS(d) .
  
S(d)−d  
d ∂L ∂L ∂Φ
− = B
λ (Q , σ) B
(Q , σ) (11.7)
dt ∂ Q̇A ∂QA ∂QA
=1
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 221

A Note on the extrinsic phase Space path Integral Method for quantization 221

where the Langrange undetermined multipliers λ(QB , σ) are functions of

general co-ordinates {QB } and of the time σ also. They are determined
from the constraints, at least locally in suitable manifold charts

S(d)−d
∂Φ
· δQA ≡ 0. (11.8)
∂QA
=1

Note that the set of equations eq. (11.7)–eq. (11.8) must be added with
the initial conditions
QA (0) = f A (A)
. (11.9)
Q̇A (0) = ∇f (A), BRS(d)−d
Let us pass now to our proposed phase-space path integral for quantiza-
tion of this constraint classical system in RS(d) (eq. (11.2)–eq. (11.5)). So,
let us brieflly display those basic results on Hamiltonian constraints path
integral framework ([5]).
Firstly we consider the classical action functional for such a system. It
is given by
T 
S(d)

S(d)−
S= dσ PA Q̇A − H(PA , QA ) − λ (QB )Φ (QB ) . (11.10)
0 i=1 =1

The set of variables {PA , Q } form the phase space R2S(d) and Φ
A

are constraints. They clearly satisfy the Poisson-algebra closeness prop-

erty (with all Poison-algebra structure constants naturally vanishing in our
model). Namely through an explicitly check:
=P A ≡∇Φ
S() 
+ ,- .+ ,- .
   ∂ 1 B ∂
H, ΦA P B = P PB Φ (QA )
∂P A 2 ∂QA
A=1

=0
+ ,- .
∂ 1 B ∂ (11.11)
− P PB  A
Φ (Q )
∂QA 2 ∂P A

A
=P(σ)
+ ,-A . 
dQ d   A 
= ·∇Φ (Q ) =
 A
Φ (Q (σ)) = 0
dσ dσ
with also the result
 A B
Φ , Φ P B = 0. (11.12)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 222

222 Lecture Notes in Topics in Path Integrals

Here {, }P B denotes the usual operation of taking Poison-Brackets.

An important result on theses quantization frameworks for constraint
classical dynamics is that one must fix a sort of “gauge”, just in order to
assure us that when one is evaluting physical variables Ω(QA , P A ) of the
system phase-space along the physical trajectories, they do not depend of
the choice of Lagrange multipliers, namely: One must have that

  
Ω, Φ P B = dP Φp (11.13)


together with

dΩ   
= {H, Ω}P B + λ Φ , Ω . (11.14)
dσ


This gauge fixing procedure can be achieved-according ref.[5] - by in-

troducing a set of surfaces χ (QA , P A ) = 0 on the system’s phase space
satisfying the conditions below written

 
a) χ , χr PB
=0 (11.15)

  
b) det {χ , Φr } = 0. (11.16)
PB =1,...,S(d)−d
r=1,...,S(d)−d

If this is the case, which always happens in our case as can be easily
seen, one can perform a canonical transformation which turns the gauge
fixing functions χ (QA , P A ) ≡ π  into new canonical moment. Let thus,
Q be the coordinates conjugate to π  and Q∗ , P ∗ the remaining set of
canonical variables. Note that we can always solve the system Φ = 0,
and find Qa = Qa (Q∗ , P ∗ ). So the constraint conditions Φ = 0 and the
supplementary conditions χ define the new physical phase space Γ∗ . And
within Γ∗ , we have that

π ≡ 0  = 1, . . . , S(d) − d

Qa = Qa (Q∗ , π ∗ ). (11.17)
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 223

A Note on the extrinsic phase Space path Integral Method for quantization 223

The main result ([8]) to be fully used in our study is the following: The
matrix element of the quantum mechanical evalution operator is explicitly
given by the following phase-space path-integral (we have re-introduced
a potential V (X i ) ≡ Ṽ (QA ) into the motion Lagrangian without further
mathematical complication)

9 1 1 :
1 iH 1
QA
out , T
1exp 1 A
T 1 Qin , 0
1 
⎧ ⎫
⎪
⎪
1 B
2 P PB +Ṽ (QA )⎪
⎪
⎪ + ,- .⎪
⎪
⎪
⎨i T 
s(d) ⎬
= exp dσ PA Q̇A − H PA , QA (2π)S(d)−d
⎪
⎪  ⎪
⎪
⎪
⎪
0 A=1 ⎪
⎪
⎩ ⎭

   S(d)−d (11.18)
  r
× det χ , Φ × 
δ χ (Q , P ) A A
=1,...,S(d)−d
r=1,...,S(d)−d =1

S(d)
× δ Φ (Q ) A
DF [Qi (σ)]DF [P i (σ)]
=1


×δ (S(d))
Q (0) −
A
QA
in δ (S(d))
Q (T ) −
A
QA
out .

Note that it appears (at least in our proposal) that there is not an
invariant expression in terms of path integral for the matrix elements asked
j
directly into the original manifold variables Xout
i
, T | Xin , 0.
In this approach of ours, through the full use of the Nash Theorem, one
can do quantum mechanics in Riemann manifolds only when the intrinsic
geometrical setting is viewed as usual holonomic mechanical constraints
in the extrinsic space RS(d) of absolute embedding frame of the metric
manifold.
As useful remark, one can prove that the resulting phase-space path
integral eq. (11.19) does not depend on the concrete choice of supplementary
conditions χ (QA , P A ) ([8]).
As useful point in the original Feynman Lagrangean framework, the
Feymman propagation is easily re-written in the extrinsic space in the
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 224

224 Lecture Notes in Topics in Path Integrals

geometrical co-cordinates transformations invariant form as

 
A A
G (Qout , T ); (Qin , 0)

QA (T )=QA
out
 
= dQ (σ) (W [Q (σ)])
A
QA (0)=QA
in 0≤σ≤T
S(d)−d (11.19)
× δ Φ (QB (σ))
=1

$  '
T
i 1
× exp dσ M (Q̇ (σ) − Ṽ (Q (σ))
A 2 A
 0 2
with the path measure weight (see eq. (11.4))
⎧ ⎫
⎨  1
4
⎬
W [Qi (σ)] = gμν (Gi (QA (σ)) . (11.20)
⎩ ⎭
0≤σ≤T

A simple analysis of the Lagrangean path integral eq. (11.20)–eq. (11.21)

shows that analytical disentanglements of them with sensible results on the
intrinsic manifold motion appears to be possible only in a Post-Newtonian
pertubartive framework of small metrical deviations from the usual Eu-
1 1 (2)
clidean space: gμv (x) = δμv + h(1)
μv (x) + h (x) + . . . . Here C denotes
C C 2 μv
the light velocity ([6]).
Let us now apply these phase space path integral manifold motion result
for the important classical problem of diffusion on the Riemann manifolds,
one important problem in Diffusion Physics on Manifolds.
We thus consider the usual linear diffusion equation in the space Rn ,
but now endowed with a metric {gab (x)}a=1,...,n . Namely ([1]).
b=1,...,n
$ 1
U (t, x) = − (Δg(x) U )(x) − (V U )(x)
2 (11.21)
U (x, 0) = f (x) ∈ L2 (Rn , gdn x).
Here V (x) denotes a real valued function on {Rn , gab (x)}, the so called
diffusion potential. By following the previous analitically complex time
continuation arguments on the connection of the diffusion equation and
the Schrdinger equation on Rn ([9]), one can easily writen a path integral
(Feynman-Kac-Wiener) for the solution eq. (11.22).
+∞ 4
U (t, x) = dn y g(y K(x, y, t)f (y) (11.22)
−∞
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 225

A Note on the extrinsic phase Space path Integral Method for quantization 225

with the evolution Kernel: 1

1
K(x, y, t) = G(Qx , Qy , t)11 (11.23)
Qx,A =f A (x1 ,...,xn )
Qy,B =f B (y 1 ,...,y n )
with G(Qx , Qy , t) the analitically imaginary time continued path integral
equation as given by eq. (11.20), namely
K(x, y, t) = G(f A (x), it); (f A (y), 0). (11.24)
We finally pass to the second quantization problem in a perturbative
framework for neutral scalar fields. Here one must given a meaning for the
covariant manifold free field path integral under the presence of an external
(covariant coupled) $
field source ([9])
4
1
Z[J(x)] = Z(0) × 4
g(xα ) dϕ(xα )
xμ ∈M

  
i 1 4
× exp + (ϕ(− Δg )ϕ)(xα ) g(xα )dD x (11.25)
 M 2
 
i 4
× exp + α α α D
(J(x )ϕ(x )) g(x )d x .
 M
After evaluating the exactly soluble Gaussian Feynman covariant field
path integrals and expressing the resulting function determinant by means
of the proper-time method, one obtains the “Loop Space” path integral
representation for the non-normalized generating Functional associated to
the Feynman field path integral>eq. (11.26). Namely:
 4  4
Z[J(x)] = det−1/2 (−Δg ) exp i M g(xα )dν x M g(y α )dν y
? (11.26)
α −1 α α α
J(x )(−Δg ) (x , y )J(y )
where one has the proper-time motion manifold path integrals representa-
tions, see eqs. (11.23)–(11.24).
≡K(xα ,xα ,t)
− 1  + ,- .
∞
2 dt −t(−Δg )
g det −Δg = T rL2 (M,g(x) e
0 2t
(11.27)
≡K(xα ,y α ,t)
∞ + ,- .
(−Δg )−1 (xα , y α ) = dt x, t|e−t(−Δg ) |y, o .
0
After re-inserting Eq. (11.26) into Eq. (11.27) we are able to re-write the
non-normalized Generating Functional in terms of a dynamics of covariant
path integrals ([9]).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 226

226 Lecture Notes in Topics in Path Integrals

11.3 References

[1] N. Ogawa, K. Fuji, A. Kobushkin, Quantum mechanics in Riemann

Manifold, Prog. Theor. Phys. 83, 894 (1990).
[2] T. Homma, T. Inamoto, T. Miyazaki, Schrodinger Equation for the
Nonrelativistic Particle Constrained on a Hypersurface in a Curved
Space, Phys. Rev. D42, 2049 (1990).
[3] N. Ogawa, K. Fuji, The difference of effective hamiltonian in two
methods in quantum mechanics on submanifold, Prog. Theor.
Phys. 87, 513 (1992).
[4] A Shimizu, T. Inamoto, T. Miyazaki, Path integral quantization
of a nonrelativistic particle constrained on a general hypersurface,
Nuovo Cim. B107, 973 (1992).
[5] Sternberg, Shlomo, “Lectures on Differential Geometry”, Chelsea
Publishing, Company, New York, (1983).
[6] Luiz C.L. Botelho, Int J. Theoretical Physics, 48, 1554-1558,
(2009).
[7] H. Goldstein, “Classical Mechanics”, second edition, Addison-
Wesley Publishing Company (1980).
[8] L. D. Faddeev and A. A. Slavnov, “Gauge Fields an introduction
to quantum theory”, second edition, Addison-Wesley Publishing
Company, (1991).
[9] Luiz C. L. Botelho, “Methods of Bosonic and Fermionic Path Inte-
grals Representations: Continuum Random Geometry in Quantum
Field Theory, Nova Science, NY, (2009).
December 8, 2016 14:47 10129 - Lecture Notes in Topics in Path Integrals 9789813143463 page 227

Index

Abelian, 12, 59, 117 Hamiltonian, 182, 187

Anomaly, 38 Hilbert, 137, 153, 154, 165, 169, 173
Atgah-Singer, 44
Index (Topological), 44, 45, 47
Behavior, 7, 68, 76, 94 Integral, 5, 10, 32, 46, 84, 97
Bosonic, 1, 45, 67, 71, 107
Bosonization, 126 Jacobian, 55, 60, 126
Boundary, 26, 71
Kac, 83, 133, 208, 210
Charges, 117 Klein Gordon, 173, 180, 182, 192
Confinement, 3, 7, 117, 120
Constant fields, 1, 22 Liouville, 37, 96, 99, 102, 111
Correction term, 7, 92 Loop Space, 7, 14, 16, 69, 225

Determinants, 42, 46, 50, 93 Matrix, 5, 141, 144

Diffusion, 65, 208, 224 Minlos, 134, 165, 168, 182
Dirac, 14, 54, 60, 126, 154 Motion, 173, 187, 220
Dirichlet, 32, 48, 88, 93
Nambo, 27, 32, 88
Electric Charge, 117, 118 Non-Linear, 89, 135, 173, 180, 220
Ergodic, 174, 176, 180, 184
Euler, 27, 86, 220 Path Integral, 3, 9, 32, 45, 101
Extrinsic, 29, 31, 73, 91, 219 Polyakov, 28, 58, 83

Feynman, 136, 147, 197, 208 Renormalization, 76, 95, 133, 142
Riemann, 31, 147, 219, 223
Field Theory, 76, 123, 133, 137
Fourier, 142, 150, 153, 169, 213
Wave Propagation, 173
Wiener, 146–148
Generating Functional, 48, 51, 58, 77,
134
Green, 30, 52, 97, 216

227