100% found this document useful (2 votes)
67 views

Optimal Control 1st Edition Richard Vinter (Auth.) 2024 Scribd Download

ebook

Uploaded by

nkungalaheen
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (2 votes)
67 views

Optimal Control 1st Edition Richard Vinter (Auth.) 2024 Scribd Download

ebook

Uploaded by

nkungalaheen
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 84

Full download ebook at ebookgate.

com

Optimal Control 1st Edition Richard Vinter


(Auth.)

https://ebookgate.com/product/optimal-control-1st-
edition-richard-vinter-auth/

Download more ebook from https://ebookgate.com


More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Optimal Control An Introduction 1st Edition Arturo


Locatelli

https://ebookgate.com/product/optimal-control-an-
introduction-1st-edition-arturo-locatelli/

Optimal Control 3rd Edition Frank L. Lewis

https://ebookgate.com/product/optimal-control-3rd-edition-frank-
l-lewis/

Counterexamples in Optimal Control Theory Semen Ya.


Serovaiskii

https://ebookgate.com/product/counterexamples-in-optimal-control-
theory-semen-ya-serovaiskii/

Batch Distillation Simulation Optimal Design and


Control Second Edition Diwekar

https://ebookgate.com/product/batch-distillation-simulation-
optimal-design-and-control-second-edition-diwekar/
Dynamic programming and optimal control 3rd Edition
Dimitri P. Bertsekas

https://ebookgate.com/product/dynamic-programming-and-optimal-
control-3rd-edition-dimitri-p-bertsekas/

Optimal Control of Singularly Perturbed Linear Systems


and Applications Automation and Control Engineering 1st
Edition Zoran Gajic

https://ebookgate.com/product/optimal-control-of-singularly-
perturbed-linear-systems-and-applications-automation-and-control-
engineering-1st-edition-zoran-gajic/

Robust industrial control systems optimal design


approach for polynomial systems 1st Edition Michael J.
Grimble

https://ebookgate.com/product/robust-industrial-control-systems-
optimal-design-approach-for-polynomial-systems-1st-edition-
michael-j-grimble/

Solution to Modern Control Systems 11th Edition Richard


C. Dorf

https://ebookgate.com/product/solution-to-modern-control-
systems-11th-edition-richard-c-dorf/

Bleach Vol 29 Kubo

https://ebookgate.com/product/bleach-vol-29-kubo/
Modern Birkhauser Classics

Optimal Control
Modern Birkhäuser Classics

Many of the original research and survey monographs in pure and


applied mathematics published by Birkhäuser in recent decades have
been groundbreaking and have come to be regarded as foundational
to the subject. Through the MBC Series, a select number of these
modern classics, entirely uncorrected, are being re-released in
paperback (and as eBooks) to ensure that these treasures remain
accessible to new generations of students, scholars, and researchers.
Optimal Control

Richard Vinter

Reprint of the 2000 Edition

Birkhäuser
Boston • Basel • Berlin
Richard Vinter
Department of Electrical Engineering
Imperial College of Science, Technology, and Medicine
London SW7 2BT
United Kingdom

Originally published in the series Systems & Control: Foundations & Applications

ISBN 978-0-8176-4990-6 e-ISBN 978-0-8176-8086-2


DOI 10.1007/978-0-8176-8086-2
Springer New York Dordrecht Heidelberg London

Library of Congress Control Number: 2010930511

Mathematics Subject Classification (2010): 49-XX, 93-XX

© Springer Science+Business Media, LLC 2000, First softcover printing 2010


All rights reserved. This work may not be translated or copied in whole or in part without the
written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street,
New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly
analysis. Use in connection with any form of information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed is
forbidden.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if
they are not identified as such, is not to be taken as an expression of opinion as to whether or not
they are subject to proprietary rights.

Printed on acid-free paper

Birkhäuser is part of Springer Science+Business Media (www.birkhauser.com)


Richard Vinter

Optimal Control

With 12 Figures

Birkhauser
Boston Basel Berlin
Richard Vinter
Department of Electrical Engineering
Imperial College of Science, Technology,
and Medicine
London SW7 2BT
UK

Library of Congress Cataloging-in-Publication Data


Vinter, R.B. (Richard B.)
Optimal control / Richard Vinter.
p. cm. - (Systems and control)
Includes bibliographical references and index.
ISBN 0-8176-4075-4 (alk. paper)
1. Automatic control-Mathematical models. 2. Control theory. I. Title. II. Systems &
control.
TJ213.V5465 2000
629.8-dc2l 99-057788
CIP

Printed on acid-free paper.

© 2000 Birkhauser Boston Birkhauser

All rights reserved. This work may not be translated or copied in whole or in part without
the written permission of the publisher (Birkhauser Boston, c/o Springer-Verlag New York,
Inc., 175 Fifth Avenue, Ndw York, NY 10010, USA), except for brief excerpts in connection
with reviews or scholarly analysis. Use in connection with any form of information storage
and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodol-
ogy now known or hereafter developed is forbidden.
The use of general descriptive names, trade names, trademarks, etc., in this publication, even
if the former are not especially identified, is not to be taken as a sign that such names, as
understood by the Trade Marks and Merchandise Marks Act, may accordingly be used freely
by anyone.

ISBN 0-8176-4075-4
ISBN 3-7643-4075-4 SPIN 10748579

Typeset by the author in LaTeX.


Printed and bound by Edwards Brothers, Inc., Ann Arbor, MI.
Printed in the United States of America.
To Donna
Contents
Preface xiii
Notation xix
1 Overview 1
1.1 Optimal Control . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The Calculus of Variations . . . . . . . . . . . . . . . . . . . 4
1.3 Existence of Minimizers and Tonelli's Direct Method . . . . 18
1.4 Sufficient Conditions and the Hamilton-Jacobi Equation . . 21
1.5 The Maximum Principle . . . . . . . . . . . . . . . . . . . . 25
1.6 Dynamic Programming . . . . . . . . . . . . . . . . . . . . 30
1.7 Nonsmoothness . . . . . . . . . . . . . . . . . . . . . . . . . 35
1.8 Nonsmooth Analysis . . . . . . . . . . . . . . . . . . . . . . 39
1.9 Nonsmooth Optimal Control . . . . . . . . . . . . . . . . . 52
1.10 Epilogue . . . . . . . . . . . . . . . . .... . ..... .. . 56
1.11 Notes for Chapter 1 . . . . . . . . . .... . . ...... . 60

2 Measurable Multifunctions and Differential Inclusions 61


2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.2 Convergence of Sets . . . . . . . . . . . . . . . . . . . . . . 62
2.3 Measurable Multifunctions . . . . . . . .. . . . . . . . . . . 64
2.4 Existence and Estimation of F-Trajectories . . . . . . . . . 75
2.5 Perturbed Differential Inclusions . . . . . . . . . . . . .. . 86
2.6 Existence of Minimizing F-Trajectories .. . . . . . . . . . . 91
2.7 Relaxation . . . . . . . . . . . . . . . . . .. . . . . . . . . . 94
2.8 The Generalized Bolza Problem . . . . . . . .. . . . . . . . 100
2.9 Notes for Chapter 2 . . . . .. .. .. .. .. . ... . ... 108
3 Variational Principles 109
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.2 Exact Penalization . . . . . . . . . . . . . . . . . . . . . . . 110
3.3 Ekeland's Theorem . . . . . . . . . . .... . .. .... . . 111
3.4 Mini-Max Theorems . . . . 115
.. .. .... . ........
. . . . . . . . . . . . . . . . . .

3.5 Notes for Chapter 3 . . . . . 125

4 Nonsmooth Analysis 127


4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4.2 Normal Cones . . . . . . . . . . . . . . . . . . . . . . . . . . 128
CONTENTS

4.3 Subdifferentials . . . . . . . . . . . . . . . . . . . . . . . . . 133


4.4 Difference Quotient Representations . . . . . . . . . . . . . 139
4.5 Nonsmooth Mean Value Inequalities . . . . . . . . . . . . . 144
4.6 Characterization of Limiting Subgradients . . . . . . . . . . 149
4.7 Subgradients of Lipschitz Continuous Functions . . . . . . . 154
4.8 The Distance Function . . . . . . . . . . . . . . . . . . . . . 161
4.9 Criteria for Lipschitz Continuity . . . . . . . . . . . . . . . 166
4.10 Relationships Between Normal and Tangent Cones . . . . 170

Subdifferential Calculus 179


5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . .. . . 179
5.2 A Marginal Function Principle . . . . . . . . . . . . . . . . 181
5.3 Partial Limiting Subgradients . . . . . . . . . . . . . . . . . 185
5.4 A Sum Rule . . . . . . . . . . . . . . . . . . . . . .. . . . . 187
5.5 A Nonsmooth Chain Rule . . . . . . . . . . . . . . . . . . . 190
5.6 Lagrange Multiplier Rules . . . . . . . . . . . . . . . . . . . 193
5.7 Notes for Chapters 4 and 5 . . . . . . . . . . . . . . . . . . 197

The Maximum Principle 201


6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
6.2 The Maximum Principle . . . . . . . . . . . . . . . . . . .. 203
6.3 Derivation of the Maximum Principle from the Extended
Euler Condition . . . . . . . . . . . . . . . . . . . . . . . . . 208
6.4 A Smooth Maximum Principle . . . . . . . . . . . . . . . . 214
6.5 Notes for Chapter 6 . . ..
. . . . . . . . . . . . . . . . . . 228

The Extended Euler-Lagrange and Hamilton Conditions 233


7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
7.2 Properties of the Distance Function . . . . . . . . . . . . . . 237
7.3 Necessary Conditions for a Finite Lagrangian Problem . . . 242
7.4 The Extended Euler-Lagrange Condition: Nonconvex Veloc-
ity Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
7.5 The Extended Euler-Lagrange Condition: Convex Velocity
Sets . . . . . . . . . . . .. . . . . . . . . . . . . . . .. . . 259
7.6 Dualization of the Extended Euler-Lagrange Condition . . 264
7.7 The Extended Hamilton Condition . . . . . . . . . . . . . . 277
7.8 Notes for Chapter 7 . . . . . . . . . . . ... .. .. .. . . 280

Necessary Conditions for Free End-Time Problems 285


8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
8.2 Lipschitz Time Dependence . . . . . . . . . . . . . . . . . . 288
8.3 Essential Values . . . . . . . . . . . . . . . . . . . . . . . . . 295
8.4 Measurable Time Dependence . . . . . . . . . . . . . . . . . 297
8.5 Proof of Theorem 8.4.1 . . . .. . . . . . .. . . . . . .. . . 301
8.6 Proof of Theorem 8.4.2 . . . . . . . . . . . . . . . . . . . . . 310
CONTENTS xi

8.7 A Free End-Time Maximum Principle . . . . . . . . . . . . 313


8.8 Notes for Chapter 8 . . . . . . . . .. . ... . . . .. .. . 318

9 The Maximum Principle for State Constrained Problems 321


9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
9.2 Convergence of Measures . . . . . . . . . . . . . . . . . . . 324
9.3 The Maximum Principle for Problems with State Constraints 329
9.4 Derivation of the Maximum Principle for State Constrained
Problems from the Euler-Lagrange Condition . . . . . . . . 334
9.5 A Smooth Maximum Principle for State Constrained Problems339
9.6 Notes for Chapter 9 . . . . . . .. .. . ... .. ..... .
. 359

10 Differential Inclusions with State Constraints 361


10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
10.2 A Finite Lagrangian Problem . . . . . . . . . . . . . . . . . 362
10.3 The Extended Euler-Lagrange Condition for State Constrained
Problems: Nonconvex Velocity Sets . . . . . . . . . . . . . . 368
10.4 Necessary Conditions for State Constrained Problems: Con-
vex Velocity Sets . . . . . . . . . . . . . . . . . . . . . . . . 375
10.5 Free Time Problems with State Constraints . . . . . . . . . 382
10.6 Nondegenerate Necessary Conditions . . . . . . . . . . . . . 387
10.7 Notes for Chapter 10 . . . . . . . . . . . ... . .... .. . . 396

11 Regularity of Minimizers 397


11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
11.2 Tonelli Regularity . . . . . . . . .
.
.. . ... . . ... .. . 403
11.3 Proof of The Generalized Tonelli Regularity Theorem . . . 408
11.4 Lipschitz Continuous Minimizers . . . . . . . . . . . . . . . 417
11.5 Autonomous Variational Problems with State Constraints . 422
11.6 Bounded Controls . . . . . . . . . . . . ... . . ... .. . 425
11.7 Lipschitz Continuous Controls . . . . . . . . . . . . . . . . . 428
11.8 Notes for Chapter 11 . . . . . . . ... . ... . .... .. . 432

12 Dynamic Programming 435


12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
12.2 Invariance Theorems . . . . . . . . . . . . . . . . . . . . . . 442
12.3 The Value Function and Generalized Solutions of the Hamilton-
Jacobi Equation . . . . . . . . . . . . . . . . . . . . . . . . 452
12.4 Local Verification Theorems . . . . . . . . . . . . . . . . . . 465
12.5 Adjoint Arcs and Gradients of the Value Function . . . . . 474
12.6 State Constrained Problems . . . . . . . . . . . . . . . . . . 483
12.7 Notes for Chapter 12 . . ..
. . . . . . . . 487 . ... . . ... .
References 493
Index 505
Preface
"Where shall I begin, please your Majesty?" he asked. "Begin
at the beginning," the King said, gravely, "and go on till you
come to the end: then stop."
- Lewis Carroll, Alice in Wonderland

What control strategy will transfer a space vehicle from one circular or-
bit to another in minimum time or in such a manner as to minimize fuel
consumption? How should a batch distillation column be operated to max-
imize the yield, subject to specified constraints on purity of the product?
Practical questions such as these underlie the field of Optimal Control.
In the language of mathematics, Optimal Control concerns the properties
of control functions that, when inserted into a differential equation, give
solutions which minimize a "cost" or measure of performance.' In engineer-
ing applications the control function is a control strategy. The differential
equation describes the dynamic response of the mechanism to be controlled,
which, of course, depends on the control strategy employed.

Systematic study of optimal control problems dates from the late 1950s,
when two important advances were made. One was the the Maximum Prin-
ciple, a set of necessary conditions for a control function to be optimal. The
other was Dynamic Programming, a procedure that reduces the search for
an optimal control function to finding the solution to a partial differential
equation (the Hamilton-Jacobi Equation).

In the following decacce, it became apparent that progress was being


impeded by a lack of suitable analytic tools for investigating local properties
of functions which are nonsmooth; i.e., not differentiable in the traditional
sense.

Nonsmooth functions were encountered at first attempts to put Dynamic


Programming on a rigorous footing, specifically attempts to relate value
functions and solutions to the Hamilton-Jacobi Equation. It was found
that, for many optimal control problems of interest, the only "solutions" to

1This is a simplification: the field also concerns optimization problems with dynamic
constraints which might be functional differential equations, difference equations, partial
differential equations, or take other forms.
xiv Preface

the Hamilton-Jacobi Equation have discontinuous derivatives. How should


we interpret these solutions? New ideas were required to answer this ques-
tion since the Hamilton-Jacobi Equation of Optimal Control is a nonlinear
partial differential equation for which traditional interpretations of gener-
alized solutions, based on the distributions they define, are inadequate.

Nonsmooth functions surfaced once again when efforts were made to ex-
tend the applicability of necessary conditions such as the Maximum Prin-
ciple. A notable feature of the Maximum Principle (and one that distin-
guishes it from necessary conditions derivable using classical techniques),
is that it can take account of pathwise constraints on values of the control
functions. For some practical problems, the constraints on values of the
control depend on the vector state variable. In flight mechanics, for exam-
ple, the maximum and minimum thrust of a jet engine (a control variable)
will depend on the altitude ( a component of the state vector). The Max-
imum Principle in its original form is not in general valid for problems
involving state dependent control constraints. One way to derive necessary
conditions for these problems, and others not covered by the Maximum
Principle, is to reformulate them as generalized problems in the Calculus
of Variations, the cost integrands for which include penalty terms to take
account of the constraints. The reformulation comes at a price, however.
To ensure equivalence with the original problems it is necessary to employ
penalty terms with discontinuous derivatives. So the route to necessary
conditions via generalized problems in the Calculus of Variations can be
followed only if we know how to adapt traditional necessary conditions to
allow for nonsmooth cost integrands.

Two important breakthroughs occurred in the 1970s. One was the end
product of a long quest for effective, local descriptions of "non-smooth"
functions, based on generalizations of the concept of the "subdifferential"
of a convex function, to larger function classes. F. H. Clarke's theory of gen-
eralized gradients, by achieving this goal, launched the field of Nonsmooth
Analysis and provided a bridge to necessary conditions of optimality for
nonsmooth variational problems (and in particular optimal control prob-
lems reformulated as generalized problems in the Calculus of Variations).
The other breakthrough, a somewhat later development, was the concept
of Viscosity Solutions, due to M. G. Crandall and P.-L. Lions, that provides
a framework for proving existence and uniqueness of generalized solutions
to Hamilton-Jacobi equations arising in Optimal Control.

Nonsmooth Analysis and Viscosity Methods, introduced to clear a bot-


tleneck in Optimal Control, have had a significant impact on Nonlinear
Analysis as a whole. Nonsmooth Analysis provides an important new per-
spective: useful properties of functions, even differentiable functions, can
be proved by examining related nondifferentiable functions, in the same
way that trigonometric identities relating to real numbers can sometimes
simply be derived by a temporary excursion into the field of complex num-
bers. Viscosity Methods, on the other hand, provide a fruitful approach to
studying generalized solutions to broad classes of nonlinear partial differen-
tial equations which extend beyond Hamilton-Jacobi equations of Optimal
Control and their approximation for computational purposes.
The Calculus of Variations (in its modern guise as Optimal Control)
continues to uphold a long tradition, as a stimulus to research in other
fields of mathematics.

Clarke's influential book, Optimization and Nonsmooth Analysis, [38] of


1982 covered many important advances of the preceding decade in Non-
smooth Analysis and its applications to the derivation of necessary con-
ditions in Optimal Control. Since then, Optimal Control has remained an
active area of research. In fact, it has been given fresh impetus by a number
of developments. One is widespread interest in nonlinear controller design
methods, based on the solution of an optimal control problem at each con-
troller update time and referred to as Model Predictive Control [117],[103].
Another is the role of Optimal Control and Differential Games in general-
izations of H-infinity controller design methods to nonlinear systems [90].
There has been a proliferation of new nonsmooth necessary conditions,
distinguished by the hypotheses under which they are valid and by their
ability to eliminate from consideration certain putative minimizers that
are not excluded by rival sets of necessary conditions. However recent work
has helped to clarify the relationships between them. This has centered on
the Extended Euler-Lagrange Condition, a generalization of Euler's Equa-
tion of the classical Calculus of Variations. The Extended Euler-Lagrange
Condition subsumes the Hamilton Inclusion, a key early nonsmooth neces-
sary condition, and is valid under greatly reduced hypotheses. A number
of other necessary conditions, such as the Nonsmooth Maximum Principle,
follow as simple corollaries. Necessary conditions for problems involving
pathwise state constraints have been significantly improved, and degen-
erate features of earlier pecessary conditions for such problems have been
eliminated. Sharper versions of former nonsmooth transversality conditions
have been introduced into the theory. Techniques for deriving improved nec-
essary conditions for free time problems are now also available. Extensions
of Tonelli's work on the structure of minimizers have been carried out and
used to derive new necessary conditions for variational problems in cases
when traditional necessary conditions do not even make sense. In fact, there
have been significant advances in virtually all areas of nonsmooth Optimal
Control since the early 1980s.

Over the last two decades, viscosity techniques have had a growing fol-
lowing. Applications of Viscosity Methods are now routine in Stochastic
Control, Mathematical Finance, Differential Games and other fields be-
xvi Preface

sides Optimal Control.

Dynamic Programming is well served by a number of up-to-date expos-


itory texts, including Fleming and Soner's book Controlled Markov Pro-
cesses and Viscosity Solutions [66], Barles's lucid introductory text, Solu-
tions de Viscosity des Equations de Hamilton-Jacobi, and Bardi and Ca-
puzzo Dolcetta's comprehensive monograph, Optimal Control and Viscos-
ity Solutions of Hamilton-Jacobo-Bellman Equations [14] . This cannot be
said of applications of Nonsmooth Analysis to the derivation of necessary
conditions in Optimal Control. Expository texts such as Loewen's Optimal
Control Via Nonsmooth Analysis [94] and Clarke's Methods of Dynamic
and Nonsmooth Optimization [38] give the flavor of contemporary thinking
in these areas of Optimal Control, as do the relevant sections of Clarke et
al.'s recent monograph, Nonsmooth Analysis and Control Theory [54]. But
details of recent advances, dispersed as they are over a wide literature and
written in a wide variety of styles, are by no means easy to follow.

The main purpose of this book is to bring together as a single publica-


tion many major developments in Optimal Control based on Nonsmooth
Analysis of recent years, and thereby render them accessible to a broader
audience. Necessary conditions receive special attention. But other topics
are covered as well. Material on the important, and unjustifiably neglected,
topic of minimizer regularity provides a showcase for the application of
nonsmooth necessary conditions to derive qualitative information about
solutions to variational problems. The chapter on Dynamic Programming
stands a little apart from other sections of the book, as it is complementary
to recent mainstream research in the area based on Viscosity Methods (and
which in any case is already the subject matter of substantial expository
texts). Instead we concentrate on aspects of Dynamic Programming well
matched to the analytic techniques of this book, notably the characteriza-
tion (in terms of the Hamilton-Jacobi Equation) of extended-valued value
functions associated with problems having endpoint and state constraints,
inverse verification theorems, sensitivity relationships and links with the
Maximum Principle.

A subsidiary purpose is to meet the needs of readers with little prior ex-
posure to modern Optimal Control who seek quick answers to the questions:
what are the main results, what were the deficiencies of the "classical" the-
ory and to what extent have they been overcome? Chapter 1 provides, for
their benefit, a lengthy overview, in which analytical details are suppressed
and the emphasis is placed instead on communicating the underlying ideas.

To render this book self-contained, preparatory chapters are included


on Nonsmooth Analysis, measurable multifunctions, and differential in-
clusions. Much of this material is implicit in the recent books of R. T.
Preface xvii

Rockafellar and J. B. Wets [125] and Clarke et al. [53], and the somewhat
older book by J.-P. Aubin and H. Frankowska [12]. It is expected, how-
ever, that readers, whose main interest is in Optimization rather than in
broader application areas of Nonsmooth Analysis, which require additional
techniques, will find these chapters helpful, because of the strong focus on
topics relevant to Optimization.

Optimal Control is now a large field, and our choice of material for in-
clusion in this book is necessarily selective. The techniques used here to
derive necessary conditions of optimality are within a tradition of research
pioneered and developed by Clarke, loffe, Loewen, Mordukhovich, Rockafel-
lar, Vinter, and others, based on perturbation, elimination of constraints
and passage to the limit. The necessary conditions are "state of the art,"
as far as this tradition is concerned. Alternative approaches are not ad-
dressed, such as that of H. Sussmann [132], a synthesis of traditional ideas
for approximating reachable sets and of extensions to the Warga's theory of
derivate containers, which permit a relaxation of hypotheses under which
the Maximum Principle is valid in some respects and leads to different kinds
of necessary conditions for problems in which the dynamic constraint takes
the form of a differential inclusion. The topic of higher order necessary con-
ditions, addressed for example in [155] and [102], is not entered into, nor
are computational aspects, examined for example in [113], discussed.

I wish to thank Dorothee Bessis, Fernando Fontes, Thomas Stadler, and


Harry Zheng for their assistance in preparing the manuscript, and my
colleagues at Imperial College, including John Allwright, Alessandro As-
tolfi, Martin Clark, Mark Davis, Imad Jaimoukha, David Limebeer, Sandro
Machietto, David Mayne, John Perkins, Roger Sargent, and George Weiss
for their part in creating a working environment in which writing this book
has been possible. Many people have influenced my thinking on contents
and presentation. I offer special thanks to Francis Clarke, for generously
sharing his insights and ideas with me and for making our research col-
laboration, the fruits of which feature prominently in this publication, so
rewarding and enjoyable.

Above all, I express my profound gratitude to my wife, Donna, for her en-
couragement and support throughout this seemingly endless writing project
and for helping me, finally, bring it to a close.

The webpage http.//www.ps.ic.ac.uk/-rbv/oc.html will record errors, am-


biguities, etc., in the book, as they come to light. Readers' contributions,
via the e-mail address [email protected], are welcome.

London, United Kingdom Richard Vinter


Notation
B Closed unit ball in Euclidean space
lxI Eulidean norm of x
dc(x) Euclidean distance of x from C
int C Interior of C
bdy C Boundary of C
C Closure of C
Nc (x) Proximal normal cone to C at x
Nc(x) Strict normal cone to C at x
Nc(x) Limiting normal cone to C at x
Tc (x) Bouligand tangent cone to C at x
Tc (x) Clarke tangent cone to C at x
epi f Epigraph of f
aP f (x) Proximal subdifferential of f at x
af(x) Strict subdifferential of f at x
af(x) Limiting subdifferential of f at x
Op f (x) Asymptotic proximal subdifferential
off at x
Op- f (x) Asymptotic strict subdifferential
of f at x
c9' f (x) Asymptotic Limiting Subdifferential
of f at x
dom f (Effective) domain of f
Gr F Graph of F
epi f Epigraph of f
f ° (x; v) Geperalized directional derivative of f
at x in the direction v
llc(x) Indicator function of the set C
at the point x
V f (x) Gradient vector off at x
xi4x xi -+xandxiEC Vi
xi 4 x xi -+ x and f (xi) -> f (x) Vi
supp p Support of the measure p
W1"1(I; R') Absolutely continuous functions f : I -* R'°
H, 9i Hamiltonian, Unmaximized Hamiltonian
Chapter 1
Overview
Everything should be made as simple as possible, but not any
simpler.
- Albert Einstein

1.1 Optimal Control


Optimal Control emerged as a distinct field of research in the 1950s, to ad-
dress in a unified fashion optimization problems arising in scheduling and
the control of engineering devices, beyond the reach of traditional analyt-
ical and computational techniques. Aerospace engineering is an important
source of such problems, and the relevance of Optimal Control to the Amer-
ican and Russian space programs gave powerful initial impetus to the field.
A simple example is:

The Maximum Orbit Transfer Problem. A rocket vehicle is in a circu-


lar orbit. What is the radius of the largest possible coplanar orbit to which
it can be transferred over a fixed period of time? See Figure 1.1.

FIGURE 1.1. The Maximal Orbit Transfer Problem.

The motion of the vehicle during the maneuver is governed by the rocket
thrust and by the rocket thrust orientation, both of which can vary with

R. Vinter, Optimal Control, Modern Birkhäuser Classics, 1


DOI 10.1007/978-0-8176-8086-2_1, © Springer Science+Business Media, LLC 2010
2 1. Overview

time. The variables involved are

r = radial distance of vehicle from attracting center,


u = radial component of velocity,
v = tangential component of velocity,
m = mass of vehicle,
T,, = radial component of thrust, and
Tt = tangential component of thrust.
The constants are
ro initial radial distance,
mo initial mass of vehicle,
'Ymax maximum fuel consumption rate,
Tmax maximum thrust,
p gravitational constant of attracting center, and
tf duration of maneuver.

A precise formulation of the problem, based on an idealized point mass


model of the space vehicle is as follows.

Minimize - r(tf)
over radial and tangential components of the thrust history,
(T,.(t),Tt(t)), 0 < t < t1, satisfying
r(t) - u,
it (t) = v2(t)/r(t) - p/r2(t) +Tr(t)/m(t),
v(t) _ -u(t)v(t)/r(t) + Tt(t)/m(t),
m(t) _ -('Ymax/Tmax)(T, (t) +T2(t))1I2,
(T, 2(t)
m t2(t))1/2 < Tmax,
m(0) = mo, r(0) = ro, u(0) = 0, v(0) = µ/ro,
u(tf) = 0, v(t f) = p/r(tf).

Here r(t) denotes dr(t)/dt, etc. It is standard practice in Optimal Control


to formulate optimization problems as minimization problems. Accordingly,
the problem of maximizing the radius of the terminal orbit r(t f) is replaced
by the equivalent problem of minimizing the "cost" -r(tf). Notice that
knowledge of the control function or strategy (T,. (t), Tt (t)), 0 < t < t f per-
mits us to calculate the cost -r(t f): we solve the differential equations, for
the specified boundary conditions at time t = 0, to obtain the correspond-
ing state trajectory (r(t), u(t), v(t), m(t)), 0 < t < t1, and thence determine
-r(t f). The control strategy therefore has the role of choice variable in the
optimization problem. We seek a control strategy that minimizes the cost,
from among the control strategies whose associated state trajectories sat-
isfy the specified boundary conditions at time t = tf.
1.1 Optimal Control 3

-1.5 -0.5 0.5 1.5

FIGURE 1.2. An Orbit Transfer Strategy.

Figure 1.2 shows a control strategy for the following values of relevant
dimensionless parameters:

Tmax/1?0 = 0.1405,
7ma / a µ/r0 0.07487, 3.32 .
µ/ro ro/
For this strategy, the radius of the terminal circular orbit is

r(tf) = 1.5ro.

The arrows indicate the magnitude and orientation of the thrust at times
t = 0, 0.1tf, 0.2t1,... , t f. As indicated, full thrust is maintained. The
thrust is outward for (approximately) the first half of the maneuver and
inward for the second.
Suppose, for example,,that the attracting center is the Sun, the space
vehicle weighs 10, 000 lb, the initial radius is 1.50 million miles (the radius
of a circle approximating the Earth's orbit), the maximum thrust is 0.85 lb
(i.e., a force equivalent to the gravitational force on a 0.85 lb mass on the
surface of the earth), the maximum rate of fuel consumption is 1.81 lb/day,
and the transit time is 193 days. Corresponding values of the constants are

Tmax = 3.778 N, mo = 4.536 x 103 kg,


ro = 1.496 x 1011 m, 'A'max = 0.9496 x 10-5 kg s-1,
tf = 1.6675 x 107 s, p = 1.32733 x 1020 m3s-2.

Then the terminal radius of the orbit is 2.44 million miles. (This is the
radius of a circle approximating the orbit of the planet Mars.)
4 1. Overview

Numerical methods, inspired by necessary conditions of optimality akin


to the Maximum Principle of Chapter 6, were used to generate the above
control strategy.
Optimal Control has its origins in practical flight mechanics problems.
But now, 40 years on, justification for research in this field rests not only
on aerospace applications, but on applications in new areas such as process
control, resource economics, and robotics. Equally significant is the stimulus
Optimal Control has given to research in related branches of mathematics
(convex analysis, nonlinear analysis, functional analysis, and dynamical
systems).
From a modern perspective, Optimal Control is an outgrowth of the Cal-
culus of Variations which takes account of new kinds of constraints (dif-
ferential equation constraints, pathwise constraints on control functions
"parameterizing" the differential equations, etc.) encountered in advanced
engineering design. A number of key recent developments in Optimal Con-
trol have resulted from marrying old ideas from the Calculus of Variations
and modern analytical techniques. For purposes both of setting Optimal
Control in its historical context and of illuminating current developments,
we pause to review relevant material from the classical Calculus of Varia-
tions.

1.2 The Calculus of Variations


The Basic Problem in the Calculus of Variations is that of finding an arc
x which minimizes the value of an integral functional

J(x) = fST L (t, x(t), i(t))dt

over some class of arcs satisfying the boundary condition

x(S) = xo and x(T) = xi.


Here [S, T] is a given interval, L : [S, T] x RI x Rn -> R is a given function,
and x0 and xl are given points in R'.

The Brachistochrone Problem. An early example of such a problem


was the Brachistochrone Problem circulated by Johann Bernoulli in the late
17th century. Positive numbers s f and x f are given. A frictionless bead,
initially located at the point (0, 0), slides along a wire under the force of
gravity. The wire, which is located in a fixed vertical plane, joins the points
(0, 0) and (s f, x1). What should the shape of the wire be, in order that the
bead arrive at its destination, the point (s, x f), in minimum time?
1.2 The Calculus of Variations 5

FIGURE 1.3. The Brachistochrone Problem.


There are a number of possible formulations of this problem. We now
describe one of them. (See Figure 1.3.) Denote by s and x the horizontal
and vertical distances of a point on the path of the bead (vertical distances
are measured downward). We restrict attention to wires describable as the
graph of a suitably regular function x(s), 0 < s < s f. For any such function
x, the velocity v(s) is related to the downward displacement x(s), when the
horizontal displacement is s, according to

mgx(s) = 1mv2(s)

("loss of potential energy equals gain of kinetic energy"). But, denoting the
time variable as t, we have
1 + jdx(s)/dsI2
v(s) =
dt(s)/ds

It follows that the transit time is

J0
Bf
dt = f 1 + Idx(s)/dsl2
v(s)
ds.

Eliminating v(s) from the preceding expressions, we arrive at a formula for


the transit time: rsf
J(x) = L(s, x(s), e(s))ds,
J0
in which
1 + 1wr
L( w)
VfEg-x
The problem is to minimize J(x) over some class of arcs x satisfying
x(0) = 0 and x(sf) = xf.
This is an example of the Basic Problem of the Calculus of Variations, in
which (S, xo) = (0, 0) and (T, xl) = (s f, x f).
6 1. Overview

Suppose that we seek a minimizer in the class of absolutely continuous


arcs. It can be shown that the minimum time t* and the minimizing arc
(x(t), s(t)), 0 < t < t* (expressed in parametric form with independent
variable time t) are given by the formulae

x(t) = a (I - cos and s(t) = a (t%/ t - sin


a
Here, a and t* are constants that uniquely satisfy the conditions

The minimizing curve is a cycloid, with infinite slope at the point of


departure: it coincides with the locus of a point on the circumference of a
disc of radius a, which rolls without slipping along a line of length tf.
Problems of this kind, the minimization of integral functionals, may
perhaps have initially attracted attention as individual curiosities. But
throughout the 18th and 19th centuries their significance became increas-
ingly evident, as the list of the laws of physics that identified states of nature
with minimizing curves and surfaces lengthened. Examples of "Rules of the
Minimum" are as follows.
Fermat's Principle in Optics. The path of a light ray achieves a local
minimum of the transit times over paths between specified endpoints that
visit the relevant reflecting and refracting boundaries. The principle pre-
dicts Snell's Laws of Reflection and Refraction, and the curved paths of
light rays in inhomogeneous media. See Figure 1.4.

FIGURE 1.4. Fermat's Principle Predicts Snell's Laws.

Dirichlet's Principle. Take a bounded, open set 52 C R2 with boundary


852, in which a static two-dimensional electric field is distributed. Denote
by V (x) the voltage at point x E Q. Then V (x) satisfies Poisson's Equation
AV (x) = 0 for x E 52
V (x) = V (x) for x E 852.
1.2 The Calculus of Variations 7

Here, V : a12 -- R is a given function, which supplies the boundary data.


Dirichlet's Principle characterizes the solution to this partial differential
equation as the solution of a minimization problem
Minimize fo VV(x) VV(x)dx
over surfaces V satisfying V (x) = IV(x) on a52.
This optimization problem involves finding a surface that minimizes a given
integral functional. See Figure 1.5.

FIGURE 1.5. A Mimimizer for the Dirichlet Integral.

Dirichlet's Principle and its generalizations are important in many re-


spects. They are powerful tools for the study of existence and regularity of
solutions to boundary value problems. Furthermore, they point the way to
Galerkin methods for computing solutions to partial differential equations,
such as Poisson's equation: the solution is approximated by the minimizer
of the above Dirichlet integral over some finite-dimensional subspace SN
of the domain of the original optimization problem, spanned by a finite
collection of "basis" functions {Oi}N 1,

SN = { ai0i(x) : a E RN} .
i=1
The widely used finite element methods are modern implementations of
Galerkin's method.
The Action Principle. Let x(t) be the vector of generalized coordinates
of a conservative mechanical system. The Action Principle asserts that x(t)
8 1. Overview

evolves in a manner to minimize (strictly speaking, to render stationary)


the "action," namely,

f [T(x(t), x(t)) - V (x(t))]dt.

Here T (x, a) is the kinetic energy and V (x) is the potential energy. Suppose,
for example, x = (r, B), the polar coordinates of an object of mass m moving
in a plane under the influence of a radial field (the origin is the center of
gravity of a body, massive in relation to the object). See Figure 1.6. Then

2 m(r2 + r2e2)

and
V (r) = -K/r,
for some constant K. The action in this case is
(m2t) + r2(t)82(t)] - K/r(t) I dt.
J

r motion

attractive center

FIGURE 1.6. Motion in the Plane for a Radial Field.

The Action Principle has proved a fruitful starting point for deriving the
dynamical equations of complex interacting systems, and for studying their
qualitative properties (existence of periodic orbits with prescribed energy,
etc.).

Necessary Conditions
Consider the optimization problem

Minimize J(x) fs L(t, x(t), x(t))dt


(CV) over arcs x satisfying
I x(S) = xo and x(T) = xl,
1.2 The Calculus of Variations 9

in which [S, T] is a fixed interval, L : R x RI x RI - R is a given C2


function, and x0 and x1 are given points in R. Precise formulation of
problem (CV) requires us to specify the domain of the optimization prob-
lem. We take this to be W1"1([S,T]; R"), the class of absolutely continuous
R"-valued arcs on [S, T], for reasons that are discussed presently.
The systematic study of minimizers x for this problem was initiated by
Euler, whose seminal paper of 1744 provided the link with the equation:

x(t), x(t)) = x(t), x(t)). (1.1)

(In this equation, Lx and L are the gradients of L(t, x, v) with respect to
the second and third arguments, respectively.)
The Euler Equation (1.1) is, under appropriate hypotheses, a necessary
condition for an arc x to be a minimizer. Notice that, if the minimizer
± is a C2 function, then the Euler Equation is a second-order, n-vector
differential equation:
Lvt (t, x(t), x(t)) + Lvy (t, x(t), x(t)) x(t)
+Lvv(t,x(t),x(t)) X (t) = Lx(t,x(t),x(t)).
A standard technique for deriving the Euler Equation is to reduce the
problem to a scalar optimization problem, by considering a one-parameter
family of variations. The Calculus of Variations, incidentally, owes its name
to these ideas. (Variations of the minimizing arc cannot reduce the cost;
conditions on minimizers are then derived by processing this information,
with the help of a suitable calculus to derive necessary conditions of op-
timality.) Because of its historical importance and its continuing influence
on the derivation of necessary conditions, we now describe the technique
in detail.
Fix attention on a minimizer x. Further hypotheses are required to derive
the Euler Equation. We assume that there exists some number K such that
jL(t, x, v) - L(t, y, w) < K(lx - yI + Iv - wl) (1.2)

for all x, y E R" and all v, w E R".


Take an arbitrary C1 arc y, which satisfies the homogeneous boundary
conditions
y(S) = y(T) = 0.
Then, for any e > 0, the "variation" x + ey, which satisfies the endpoint
constraints, must have cost not less than that of x. It follows that
C-1 [Ax + ey) - J(x)] > 0.

Otherwise expressed
T

f e-1 [L(t, x(t) + ey(t), (t) + ey(t)) - L(t, x(t), x(t))]dt >0.
10 1. Overview

Under Hypothesis (1.2), the Dominated Convergence Theorem permits us


to pass to the limit under the integral sign. We thereby obtain the equation
ffT
[L.(t, x(t), x(t)) y(t) + x(t), X(t)) y(t)]dt > 0.
s
This relationship holds, we note, for all continuously differentiable func-
tions y satisfying the boundary conditions y(S) = 0 and y(T) = 0. By
homogeneity, the inequality can be replaced by equality.
Now apply integration by parts to the first term on the left. This gives

IsT [- st Lx(s, x(s), x(s))ds + x(t), X* (t))] y(t)dt = 0.


J
Take any continuous function w : [S, T] -3 R' that satisfies
fT
w(t)dt = 0. (1.3)
s

Then the continuously differentiable are y(t) fs w(s)ds vanishes at the


endtimes. Consequently,
fT[- '
S Lx(s, x(s), x(s))ds + x(t), x(t))] w(t)dt = 0, (1.4)

a relationship that holds for all continuous arcs w satisfying (1.3). To ad-
vance the analysis, we require

Lemma (Raymond Dubois) Take a function a E L2([S,T]; R''2). Suppose


that
rT
a(t) w(t) dt = 0 (1.5)
is
for every continuous function w that satisfies
fT
w(t)dt = 0. (1.6)
s
Then there exists some vector d E R' such that
a(t) = d for a.e. t E [S, TI.

Proof. We give a contemporary proof (based on an application of the


Separation Principle in Hilbert space) of this classical lemma, a precursor
of 20th century theorems on the representation of distributions.
By hypothesis, for any continuous function w, (1.6) implies (1.5). By
using the fact that the continuous functions are dense in L2([S,T];R"`)
1.2 The Calculus of Variations 11

with respect to the strong L2 topology, we readily deduce that (1.6) implies
(1.5) also when w is a L2 function. This fact is used shortly.
Define the constant elements ej E L2 ([S, T]; Rn), j = 1, . . . , n to have
components:
ei (t) _
1 if i = j i = 1, ... , n.
0 ifi#j
Since a function is a constant function if and only if it can be expressed
as a linear combination of the e2(t)s, the properties asserted in the lemma
can be rephrased:
a E V,

where V is the (closed) n-dimensional subspace of L2 ([S, T]; Rn) spanned


by the e3s. Suppose that the assertions are false; i.e., a V V. Since V is a
closed subspace, we deduce from the Separation Theorem that there exists
a nonzero element w E L2 and e > 0 such that

< a, w >L2 < < V, W >L2 -e.


for all v E V. Because V is a subspace, it follows that

< V, W >L2 = 0

for all v E V. But then


< a,w >L2 < -e. (1.7)

Observe that, for each j,

<e3 ,w>L2=0,
since ej E V. This last condition can be expressed as
fT
w(t)dt = 0.
s
In view of our earlier comments,

<a,w>L2= 0.
This contradicts (1.7). The lemma is proved.

Return now to the derivation of the Euler Equation. We identify the


function a(.) of the lemma with

-f L(s, (s), (s))ds + L(t, (t), (t)).


12 1. Overview

In view of (1.4), the lemma informs us that there exists a vector d such
that
- J t L (s, x(s), i(s))ds + x(t), x(t)) = d a.e. (1.8)
s
Since Lx (t, x(t), x(t)) is integrable, it follows that t -+ L (t, x(t), x(t)) is
almost everywhere equal to an absolutely continuous function and

-T- (t), x(t)) = Lx(t,x(t),x(t)) a.e.

We have verified the Euler Equation and given it a precise interpretation,


when the domain of the optimization problem is the class of absolutely
continuous arcs.
The above analysis conflates arguments assembled over several centuries.
The first step was to show that smooth minimizers x satisfy the pointwise
Euler Equation. Euler's original derivation made use of discrete approxi-
mation techniques. Lagrange's alternative derivation introduced variational
methods similar to those outlined above (though differing in the precise na-
ture of the "integration by parts" step). Erdmann subsequently discovered
that, if the domain of the optimization problem is taken to be the class of
piecewise C' functions (i.e., absolutely continuous functions with piecewise
continuous derivatives) then
"the function t -* L (t, x(t), X(t)) has removable discontinuities."
This condition is referred to as the First Erdmann Condition.
For piecewise Cl minimizers, the integral version of the Euler Equa-
tion (1.8) was first regarded as a convenient way of combining the point-
wise Euler Equation and the First Erdmann Condition. We refer to it as
the Euler-Lagrange Condition. An analysis in which absolutely continuous
minimizers substitute for piecewise Cl minimizers is an early 20th century
development, due to Tonelli.
Another important property of minimizers can be derived in situations
when L is independent of the t (write L(x, v) in place of L(t, x, v)). In this
"autonomous" case the second-order n-vector differential equation of Euler
can be integrated. There results a first-order differential equation involving
a "constant of integration" c:
L(x(t), x(t)) c. (1.9)

This condition is referred to as the Second Erdmann Condition or Con-


stancy of the Hamiltonian Condition. It is easily deduced from the Euler-
Lagrange Condition when _Jr- is a C2 function. Fix t. We calculate in this
case

dt
(L x(t) - L) = d L i(t) + L x (t) - Lx X (t) - L x (t)
1.2 The Calculus of Variations 13

dtLv - Lx) x(t)


0.

(In the above relationships, L, Lv, etc., are evaluated at (.t (t), xk (t)).) We
deduce (1.9).
A more sophisticated analysis leads to an "almost everywhere"' version of
this condition for autonomous problems, when the minimizer x in question
is assumed to be merely absolutely continuous.
Variations of the type x(t) +ey(t) lead to the Euler--Lagrange Condition.
Necessary conditions supplying additional information about minimizers
have been derived by considering other kinds of variations. We note in
particular the Weierstrass Condition or the Constancy of the Hamiltonian
Condition:

Lv (t, x(t), x(t)) (t) - L(t, x(t), i(t))


.

= max {L,, (t, x(t), i(t)) v - L(t, x(t), v)}.


VER'

Suppressing the (t, x(t)) argument in the notation and expressing the Weier-
strass Condition as
L(v) - L(i(t)) > L,, (i) (v - i(t)) for all v E Rn,
we see that it conveys no useful information when L(t, x, v) is convex with
respect to the v variable: in this case it simply interprets Lv as a subgradient
of L in the sense of convex analysis. In general however, it tells us that L
coincides with its "convexification" (with respect to the velocity variable)
along the optimal trajectory; i.e.,
L(t, x(t), (t)) = L(t, x(t), i(t)).
Here L(t, x,.) is the function with epigraph set co {epi L(t, x,.)}. See Figure
1.7.

The above necessary conditions (the Euler-Lagrange Condition, the Weier-


strass Condition, and the Second Erdmann Condition) have convenient
formulations in terms of the "adjoint arc"

p(t) = Lv(t, x(t), x(t))


They are
(p(t),p(t)) = Vx,vL(t,2(t),x(t)), (1.10)
p(t) i(t) - L(t, x(t), (t)) = mR [p(t) v - L(t, x(t), v)],
and, in the case when L does not depend on t,
p(t) i(t) - L(t, x(t), i(t)) = c
14 1. Overview

L(t,X(t),v)

convex hull of
epi

L(t,X(t),X(t))

FIGURE 1.7. The Weierstrass Condition.


for some constant c.
To explore the qualitative properties of minimizers it is often helpful to
reduce the Euler-Lagrange Condition to a system of specially structured
first-order differential equations. This was first achieved by Hamilton for
Lagrangians L arising in mechanics. In this analysis the Hamiltonian,
H(t,x,p) := ma
VERn

has an important role.


Suppose that the right side has a unique maximizer vrnax. This will de-
pend on (t, x, p) and so we can use it to define a function x : [S, T] x R" x
Rn-3Rn:
X(t,x,p) := vmax
Then
H(t, x, p) = p - X (t, x, p) - L (t, x, X (t, x)p) )
and
V v (p v - L(t, x, v)) (v=x(t,X,P) = 0.
The last relationship implies
p=Lv(t,x,X(t,x,p))
(The mapping from x vectors to p vectors implicit in this relationship, for
fixed t, is referred to as the Legendre Transformation.)
Now consider an arc t and associated adjoint are p that satisfy the
Euler-Lagrange and Weierstrass Conditions, namely
(p(t),p(t)) = x,vL(t,x(t),Wi(t)) (1.12)
and

p(t) x(t) - L(t, x(t), x(t)) = maXVER^ {p(t) . v - L(t, x(t), v)} .
1.2 The Calculus of Variations 15

Since it is assumed that the "maximizer" in the definition of the Hamilto-


nian is unique, it follows from the Weierstrass condition that

Wi(t) = X(t, -t M, P(t)) -

Fix t. Let us assume that X(t,. , .) is differentiable. Then we can calculate


the gradients of H(t, . , .):

vxH(t, x, p) I x=.t(t),p=p(t)
= vx (p . X(t, x, p) - L(t, x, x(t, x, p))) I x=2(t),p=p(t)
= p - X. (t, x, p) - Lx (t, x, x(t, x, p))
- LT, (t, x, X (t, x, P)) - Xx (t, x, p) I x=z(t),p=p(t)
= (p - Lv (t, x, X (t, x, p))) . xx (t, x, p) - Lx (t, x, X (t, x, P)) I x=t (t),p=p(t)
= 0 -,ro(t).

(The last step in the derivation of these relationships makes use of (1.11)
and (1.12).) We have evaluated the x-derivative of H:

vxH(t,x(t),P(t)) = -P(t).
As for the p-derivative, we have

V pH(t, x)p) I x=x(t),p=p(t)


= vp(p X(t, x, p) - L(t, x, X(t, x, p))) I x=x(t),p=p(t)
= X (t, x, p) +p. xp (t, x, p) - Lv (t, x, X (t, x, p)) - Xp (t, x, p) I x=I(t),p=p(t)
=X(t) + (p(t) - x(t), x(t))) - xp(t, x(t), p(t))
= x(t) + 0.
Combining these relationships, we arrive at the system of first-order differ-
ential equations of interest, namely, the Hamilton Condition
(-p(t),Wi(t)) = VH(t,Wi(t),p(t)), (1.13)

in which VH denotes the gradient of H(t, . , .).


So far, we have limited attention to the problem of minimizing an integral
functional of arcs with fixed endpoints. A more general problem is that in
which the arcs are constrained to satisfy the boundary condition
(x (S), x(T)) E C, (1.14)

for some specified subset C C Rn x R". (The fixed endpoint problem is a


special case, in which C is chosen to be {(xo,x1)}. The above necessary
conditions remain valid when we pass to this more general endpoint con-
straint, since a minimizer x over arcs satisfying (1.14) is also a minimizer
for the fixed endpoint problem in which C = {±(S),.t(T)}. But something
16 1. Overview

more is required: replacing the fixed endpoint constraint by other types


of endpoint constraint that allow endpoint variation introduces extra de-
grees of freedom into the optimization problem, which should be reflected
in supplementary necessary conditions. These conditions are conveniently
expressed in terms of the adjoint arc p:

(p(S), -p(T)) is an outward normal to C at (x(S), x(T)).

They are collectively referred to as the Transversality Condition, because


they assert that a vector composed of endpoints of the adjoint arc is or-
thogonal or "transverse" to the tangent hyperplane to C at (x(S), x(T)).
Figure 1.8 illustrates the transversality condition, in relation to an arc x
and adjoint are p.

direction (p(S),-p(T))

FIGURE 1.8. The Transversality Condition.

The Transversality Condition too can be derived by considering suit-


able variations (variations which, on this occasion, allow perturbations of
the endpoint values of the minimizer x). In favorable circumstances the
Transversality Condition combines with the boundary condition (1.14) to
supply the appropriate number of 2n boundary conditions to accompany
the system of 2n equations (1.13).
The following theorem brings together these classical conditions and gives
precise hypotheses under which they are satisfied.
Theorem 1.2.1 Let x be a minimizer for

Minimize J(x) := f S L(t, x(t), i(t))dt


over absolutely continuous arcs x satisfying
(x(S),x(T)) E C,
for some given interval [S, T], function L : [S, T] x Rn x Rn -4 R, and
closed set C C Rn x Rn. We list the hypotheses:
1.2 The Calculus of Variations 17

(i) L(. , x, v) is measurable for each (x, v) and L(t,. , .) is continuously


differentiable for each t E [S, T];
(ii) there exist k(.) E L1, 0 > 0 and e > 0 such that

IL(t,x,v) - L(t,x',v)I << (k(t) +alvl)lx - x'I


for all x, x' E x(t) + EB, v E Rn;

(iii) L(t, x, .) is convex for all (t, x) E [S, T] x Rn;


(iv) H(t, . , .) is continuously differentiable on a neighborhood of (x(t), x(t))
for each t E [S, T] where
H(t, x, p) := sup {p v - L(t, x, v)}.
VCR,

Suppose that Hypotheses (i) and (ii) are satisfied. Then there exists an ab-
solutely continuous arc p : [S, T] -* Rn satisfying the following conditions.

The Euler-Lagrange Condition


(p(t), p(t)) = V.,vL(t, x(t), x(t)) a.e.
The Weierstrass Condition
p(t) i(t) - L(t, x(t), x(t)) = ma {p(t) v - L(t, x(t), v)} a.e.
vERn

The Transversality Condition


(p(S), -p(T)) E Nc(x(T), x(t)).
If additionally Hypotheses (iii) and (iv) are satisfied, then p can be chosen
also to satisfy

The Hamilton Condition


(-p(t), x(t)) = V ,PH(t, x(t), p(t)) a.e.
Finally, suppose that the hypothesis
(v) L(t, x, v) is independent of t
is satisfied. Then all the above assertions can be strengthened to require that
p also satisfies

Constancy of the Hamiltonian Condition


p(t) x(t) - L(t, x(t), x(t)) = c a.e.
for some constant c.
18 1. Overview

The theorem is a special case of general necessary conditions which are


proved in Chapter 7.
In this theorem, the Transversality Condition is expressed in a way that
makes sense when C is assumed to be a closed set. It makes reference to
the "limiting normal cone" NC(x(S), x(T)). This is defined in the overview
of Nonsmooth Analysis provided in Section 1.8.

1.3 Existence of Minimizers and Tonelli's Direct


Method
Deriving the Euler-Lagrange Condition and related conditions would ap-
pear to reduce the problem of finding a minimizer to one merely of solving
a differential equation. But this overlooks the fact that satisfaction of the
Euler-Lagrange Condition is merely a necessary condition for optimality;
we cannot be certain, without further analysis, that an arc satisfying these
conditions is truly a minimizer.
Earlier this century, Tonelli recognized the significance of establishing
the existence of minimizers before using necessary conditions of optimality
to try to identify them. Tonelli's Direct Method for obtaining a minimizer
consists of the following steps.
1: Show that a minimizer exists.
2: Search among arcs satisfying the necessary conditions for an arc with
lowest cost.
These steps, when successfuly carried out, are guaranteed to yield a mini-
mizer. Notice however that, if we neglect Step 1, then Step 2 alone can be
positively misleading. The point is illustrated by the following example.

Minimize f-1 x(t)x2(t)dt


{ over absolutely continuous arcs x satisfying
x(O) = 0 and x(1) = 0.
This example is not precisely matched to the necessary conditions of The-
orem 1.2.1 because the integrand L fails to satisfy Hypothesis (ii). It can
nevertheless be shown that any Lipschitz continuous minimizer x satisfies
the Euler-Lagrange, Weierstrass, and Transversality Conditions. In this
case the Euler-Lagrange Condition takes the form:

2dt {x(t)x(t)} = x2(t) a.e.

Obviously the are


1.3 Existence of Minimizers 19

satisfies the Euler Equation. It is in fact the unique Lipschitz continuous


function so doing. To see this note that, for any Lipschitz continuous arc y
which vanishes at the endtimes and satisfies the Euler-Lagrange Condition,
we have

d 2t2t
y2(t) = y2(t) a.e.
It follows that, for some constant c,
/'
dt y2 (t) = c + J y2 (s)ds for all t.
0

Since y vanishes at the left endpoint and is assumed to have bounded


slope, we deduce from this last condition that c = 0. If y 0- 0 then y must
be nonzero on a set of positive measure, in order to satisfy the endpoint
constraints. But then
o8
r01
y(1)2 = 0 + y2(t) dsdt > 0,
JJ
in violation of the endpoint constraints. We have confirmed that :k - 0 is the
unique Lipschitz continuous arc satisfying the Euler-Lagrange Condition.
Notice, however, that x is not a minimizer, not even in a local sense.
Indeed by choosing the parameter e > 0 sufficently small we can arrange
that the arc t -p -et(1-t), which has cost strictly less than 0, is arbitrarily
close to x with respect to the supremum norm.
To summarize, in this example there is a unique Lipschitz continuous
arc satisfying the constraints of the problem and also the Euler-Lagrange,
Weierstrass, and Transversality Conditions. Yet it is not a minimizer. This
is a case in which a naive belief in the power of necessary conditions to
identify minimizers leads us astray. The pathological feature of this ex-
ample is, of course, that there are no minimizers in the class of Lipschitz
continuous functions. The collection of arcs satisfying necessary conditions
for a Lipschitz continuous arc to be a minimizer will not therefore contain
a minimizer, even though it is nonempty.

Clearly, it makes sense to speak of a minimizer only if we specify the class


of functions from which minimizers are chosen. Unfortunately, hypotheses
under which we can guarantee existence of minimizers in the "elementary"
function spaces (C' functions, piecewise C1 functions, etc.) are, for many
purposes, unacceptably restrictive. One of Tonelli's most significant contri-
butions in the Calculus of Variations was to show that, by enlarging the
domain of (CV) to include W1"1([S,T];R") arcs, the existence of minimiz-
ers could be guaranteed under broad, directly verifiable hypotheses on the
Lagrangian L. An existence theorem, in the spirit of Tonelli's pioneering
work in this field, is:
20 1. Overview

Theorem 1.3.1 Consider the minimization problem

Minimize ff L(t, x(t), b(t))dt over x E W1'1([S,T]; Rn)


satisfying
x(S) = xo and x(T) = xi,

for given [S,T] C R, L : [S,T] x RI x Rn -4 R, xo E R', and x1 E Rn.


Assume that

(i) L is continuous,

(ii) L(t, x,.) is convex for all t E [S, T] and x E Rn, and

(iii) there exist constants c > 0, d > 0, and a > 0 such that

L(t, x, v) > clvll+a - d for all t E [S, T], x E Rn, and v E Rn.

Then there exists a minimizer (in the class W1'1([S,T];Rn).


A more general version of this theorem is proved in Chapter 2.
Nowadays, existence of minimizers is recognized as an important topic
in its own right, which gives insights into whether variational problems
have been properly formulated and also into the regularity properties of
minimizers. But, as we have already observed, existence of minimizers was
first studied as an ingredient in the Direct Method, i.e., to justify seeking
a minimizer from the class of arcs satisfying the necessary conditions.
Unfortunately, existence theorems such as Theorem 1.3.1 are not en-
tirely adequate for purposes of applying the Direct Method. The difficulty
is this: implicit in the Direct Method is the assumption that a single set of
hypotheses on the data for the optimization problem at hand simultane-
ously guarantees existence of minimizers and also validity of the necessary
conditions. Yet for certain variational problems of interest, even though
existence of a minimizer 2 is guaranteed by Theorem 1.3.1, it is not clear a
priori that the hypotheses are satisfied under which necessary conditions,
such as those listed in Theorem 1.2.1, are valid at X. For these problems
we cannot be sure that a search over arcs satisfying the Euler-Lagrange
Condition and related conditions will yield a minimizer.
There are two ways out of this dilemma. One is to refine the existence
theory by showing that W1,1 minimizers are confined to some subset of
W1,1 for which the standard necessary conditions are valid. The other is to
replace the standard necessary conditions by conditions that are valid under
hypotheses similar to those of existence theory. Tonelli's achievements along
these lines, together with recent significant developments, provide much of
the subject matter for Chapter 11.
1.4 Sufficient Conditions 21

1.4 Sufficient Conditions and the Hamilton-Jacobi


Equation
We have discussed pitfalls in seeking minimizers among arcs satisfying the
necessary conditions. The steps initiated by Tonelli for dealing with them,
namely, investigating hypotheses under which minimizers exist, is a rela-
tively recent development. They were preceded by various procedures, some
of a rather ad hoc nature, for testing whether an arc satisfying known nec-
essary conditions, which we have somehow managed to find, is truly a
minimizer. For certain classes of variational problems, an arc t satisfying
the necessary conditions of Theorem 1.2.1 is a minimizer (at least in some
"local" sense) if it can be shown to satisfy also certain higher-order condi-
tions (such as the strengthened Legendre or Jacobi Conditions). In other
cases we might hope to carry out calculations, "completing the square" or
construction of a "verification function" for example, which make it obvi-
ous that an arc we have obtained and have reason to believe is a minimizer,
is truly a minimizer. This ragbag of "indirect" procedures, none of which
may apply, contrasts with Tonelli's Direct Method, which is in some sense
a more systematic approach to finding minimizers.
However one "indirect" method for confirming that a given arc is a min-
imizer figures prominently in current research and we discuss it here.
It is based on the relationship between the variational problem
Minimize fs L(t, x(t), x(t))dt + g(x(T))
(Q) over Lipschitz continuous arcs x satisfying
x(S) = xo and x(T) E Rn
(in which L : [S, T] x Rn x Rn -> R and g : Rn -* R are given contin-
uous functions and xo is a given n-vector) and Hamilton-Jacobi partial
differential equation:
Ot (t, x) + minVERn {fix (t, x) v + L(t, x, v)} = 0
(HJE) for all t E (S, T), X E Rn
O(T,x)=g(xjforallxER'.
We sometimes find it convenient to express the first of these equations in
terms of the Hamiltonian H:
H(t, x, p) := sup {p v - L(t, x, v)};
VER"

thus
Ot(t, x) - H(t, x, -4' (t, x)) = 0 for all t E (S, T), X E Rn.
(It is helpful in the present context to consider a variational problem in
which the right endpoint is unconstrained and a function of the terminal
value of the arc is added to the cost.)
22 1. Overview

The following proposition summarizes the approach, which is referred to


as Caratheodory's Verification Technique.
Proposition 1.4.1 Let x be a Lipschitz continuous are satisfying x(S)
x0. Suppose that a continuously differentiable function 0 : R x R"` -* R can
be found satisfying (HJE) and also
Ox (t, x(t)) x(t) + L(t, x(t), i(t))
= {qX (t, x(t)) v + L(t, x(t), v)} a.e. t E [S, T].

Then x is a minimizer for (Q) and


q5(S, xo) = inf (Q).

Proof. Take any Lipschitz continuous arc x : [S, T] -* R" for which x(S)
x0. Then t -* 0(t, x(t)) is a Lipschitz continuous function and

dt (t,x(t)) _ Ot(t,x(t)) +Ox(t,x(t)) . i(t) a.e.


Now express q(t, x(t)) as the integral of its derivative:

O(S, xo) f T d

dt
0(t, x(t))dt + g(x(T))

(we have used the facts that x(S) = xo and 4(T, x) = g(x))

f [ot (t, x(t)) + 0. (t, x(t)) . ±(t) + L(t, x(t), x(t))]dt


+ f L(t, x(t), i(t))dt + g(x(T)).
T

s
But the first term on the right satisfies

- f T [ot (t, x(t)) + Ox (t, x(t)) . i(t) + L(t, x(t), i (t))]dt


< -f
T
[ot (t, x(t)) + inf{ox(t, x(t)) v + L(t, x(t), v)}]dt
S
0. (1.15)

It follows that

¢(S, xo) < fis L(t, x(t), x(t))dt + g(x(T)).


T
(1.16)

Now repeat the above arguments with x in place of x. Since, by hypothesis,


Ox (t, x(t)) . (t) + L(t, x(t), i(t)) = {Qix (t, x(t)) v + L(t, x(t), v)},
1.4 Sufficient Conditions 23

we can replace inequality (1.15) by equality:


T[-Ot(t,
x(t)) + q.(t, x(t)) - x(t) + L(t, x(t), .(t))]dt = 0.
J
Consequently

q(S, xo) = T L (t, x(t), x(t))dt + g(x(T)). (1.17)


IS-

We see from (1.16) and (1.17) that :x has cost ¢(S, xo) and, on the other
hand, any other Lipschitz continuous arc satisfying the constraints of prob-
lem (Q) has cost not less than t(S, xo). It follows that O(S, xo) is the min-
imum cost and x is a minimizer.

In ideal circumstances, it is possible to find a continuously differentiable


solution 0 to the Hamilton-Jacobi Equation such that the function
v -> 0x (t, x) v + L(t, x, v)
has a unique minimizer (write it d(t, x) to emphasize that it will depend
on (t, x)) and the differential equation
r(t) = d(t,x(t)) a.e.
x(0) = xo
has a Lipschitz continuous solution Then, Proposition 1.4.1 informs us,
x must be a minimizer.
Success of this method for finding minimizers hinges of course on find-
ing a solution 0 to the Hamilton-Jacobi Equation. Recall that, if a solu-
tion 0 confirms the optimality of some putative minimizer, then c(S,xo)
coincides with the minimum cost for (Q). (This is the final assertion of
Proposition 1.4.1.) It follows that the natural candidate for solution to the
Hamilton-Jacobi Equation is the value function V : [S, T] x R" -+ R for
(Q):
V (t, x) = inf (Qt,.,) .
Here, the right side denotes the infimum cost of a variant of problem (Q),
in which the initial data (S, xo) is replaced by (t, x):
r Minimize ft T L(s, y(s), y(s)) ds + g(y(T))
(Qt,x) over Lipschitz continuous functions y satisfying
Y(t) = X.
The following proposition gives precise conditions under which the value
function is a solution to (HJE).
24 1. Overview

Proposition 1.4.2 Let V be the value function for (Q). Suppose that
(i) V is a continuously differentiable function;
(ii) for each (t, x) E [S, T] x R'n, the optimization problem (Qt,.,) has a
minimizer that is continuously differentiable.
Then V is a solution to (HJE).
Proof. Take (t, x) E [S, T] x R'n, T E [t, T], a continuously differentiable
function y : [t, T] -4 Rn, and a continuously differentiable minimizer y for
(Qt,x). Simple contradiction arguments lead to the following relationships.

V (t, x) < fT L(s, y(s), y(s)) ds + V (T, y(T)),


V (t, x) = L(s, y(s), y(s))ds + V (r, y(r)).
Jt
Take arbitrary v E Rn, E E (0,T - t), and choose y(s) = x + (s - t)v.
Then
t+
E1[V(t+E,x+cv) -V(t,x)] +E1
f L(s,x+sv,v)ds 10

and
t+E t+
E-1 [V (t + E, X + y(s)ds) - V (t, x)] + e-1 J L(s, V(s), y(s))ds = 0.
Jt t

Passage to the limit as c J 0 gives


Vt(t,x) +Vx(t,x) v+L(t,x,v) > 0
and
Vt (t, x) + V. (t, x) y(t) + L(t, x, y(t)) = 0.
Bearing in mind that v is arbitrary, we conclude that
Vt(t, x) + mien {Vx(t, x) v + L(t, x, v)} = 0.
VE,
We have confirmed that V satisfied (HJE). The fact that V also satisfies
the boundary condition V (T, x) = g(x) follows directly from the definition
of the value function.

How might we find solutions to the Hamilton-Jacobi Equation? One


approach, "construction of a field of extremals," is inspired by the above
interpretation of the value function. The idea is to generate a family of
1.5 The Maximum Principle 25

continuously differentiable arcs {yt,x}, parameterized by points (t, x) E


[S, T] x R', satisfying the necessary conditions. We then choose
fT
0(t, x) = L( t,yt,x(s),(s))ds+g(yt,x(T)).

If it turns out that 0 is a continuously differentiable function and the


extremal yt,z really is a minimizer for (Qt,x) (a property of which we have
no a priori knowledge), then V and 0 is a solution to the Hamilton-
Jacobi Equation.
The Hamilton-Jacobi Equation is a nonlinear partial differential equa-
tion of hyperbolic type. From a classical perspective, constructing a field
of extremals (a procedure for building up a solution to a partial differential
equation from the solutions to a family of ordinary differential equations)
amounts to solving the Hamilton-Jacobi Equation by the method of char-
acteristics.

1.5 The Maximum Principle


A convenient framework for studying minimization problems, encountered
in the optimal selection of flight trajectories and other areas of advanced
engineering design and operation, is to regard them as special cases of the
problem:
Minimize g(x(S),x(T))
over measurable functions u : [S, T] -+ RI and arcs
(P) x E W1'l([S,T]; Rn) satisfying
x( t) = f (t, x(t), u(t)) a.e.,
u(t) E U a.e.,
(x(S), x(T)) E C,
the data for which comprise an interval [S, T], functions f : [S, T] x Rn X
R" -+ Rnandg:RnxRI->R,andsets CCRnandUCR'".
This formulation of the Optimal Control Problem (or variants on it in
which, for example, the cost is augmented by an integral term, or the end-
points of the time interval [S, T] are included among the choice variables,
or pathwise constraints are imposed on values of x) is referred to as the
Pontryagin formulation. The importance of this formulation is that it em-
braces a wide range of significant optimization problems which are beyond
the reach of traditional variational techniques and, at the same time, it is
very well suited to the derivation of general necessary conditions of opti-
mality.
In (P), the n-vector dependent variable x is called the state. The function
describing its time evolution, x(t), S < t < T, is called the state trajectory.
26 1. Overview

The state trajectory depends on our choice of control function u(t), S < t <
T, and the initial state x(S). The object is to choose a control function u(.)
and initial state to minimize the value of the cost g(x(S),x(T)) resulting
from our choice of u(.) and x(S).
Frequently, the initial state is fixed; i.e., C takes the form
C = {xo} x C1 for some xo and some Cl C Rn.
In this case, (P) is a minimization problem over control functions. However,
allowing freedom in the choice of initial state introduces a flexibility into
the formulation that is useful in some applications. A case in point is ap-
plication of Optimal Control to some problems of exploitation of renewable
resources, where an endpoint constraint set of interest is
C = { (xo, xl) : xo = x1 j.
This "periodic" endpoint constraint corresponds to the requirement that
the population which is being harvested is the same at the beginning and
the end of the harvesting cycle ("sustainable harvesting"). The size of the
initial population is determined by optimization.
There follows a statement of the Maximum Principle, whose discovery
by L. S. Pontryagin et al. in the 1950s was an important milestone in the
emergence of Optimal Control as a distinct field of research.
Theorem 1.5.1 (The Maximum Principle) Let (u, x) be a minimizer
for (P). Assume that
(i) g is continuously differentiable,
(ii) C is a closed set,
(iii) f is continuous, f (t, . , u) is continuously differentiable for each (t, u),
and there exist e > 0 and k(.) E Ll such that
If(t,x,u)- f(t,x',u)J < k(t)lx-x'l
for all x, x' E x(`t) + eB and u E U, a.e., and
(iv) Gr U is a Borel set.
Then there exist an arc p E W',' ([S, T]; Rn) and A > 0, not both zero, such
that the following conditions are satisfied:

The Adjoint Equation:


-p(t) = p(t) . fx (t, x(t), fi(t)), a.e.;
The Generalized Weierstrass Condition:
p(t) f (t, x(t), u(t)) = ma J p(t) . f (t, x(t), u) a.e.;
UC-
1.5 The Maximum Principle 2

The Transversality Condition:


(p(S),-p(T)) = A9x(x(S),x(T))+77
for some 77 E NC(x(S),x(T)).
The limiting normal cone NC of the endpoints constraint set C featured
in the above Transversality Condition is defined presently; in the case that
C is a smooth manifold it reduces to the set of outward-pointing normals.
An alternative statement of the Maximum Principle is in terms of the
Unmaximized Hamiltonian

'H (t, x, u) p . f (t, x, u)


The adjoint equation (augmented by the state equation x = f) and the
Generalized Weierstrass Condition can be written

(x(t), -p(t)) = Vx,pH(t, x(t), p(t), u(t)) a.e.

and
R(t, x(t), p(t), u(t)) = maUx7-l(t, x(t), p(t), u) a.e.,

a form of the conditions that emphasizes their affinity with Hamilton's


system of equations in the Calculus of Variations.
In favorable circumstances, we are justified in setting the cost multiplier
A = 1, and the Generalized Weierstrass Condition permits us to express u
as a function of x and p
u = u*(x,p)
The Maximum Principle then asserts that a minimizing arc x is the first
component of a pair of absolutely continuous functions (x, p) satisfying the
differential equation

(-p(t),x(t)) = Vx,rW(t,x(t),p(t),u*(x(t),p(t))) a.e. (1.18)

and the endpoint conditions


(x(S), x(T)) E C and (p(S), -p(T)) E NC(x(S), x(T)).
The minimizing control is given by the formula
u(t) = u*(x(t), p(t)).
Notice that the (vector) differential equation (1.18) is a system of 2n scalar,
first-order differential equations. If C is a k-dimensional manifold, speci-
fied by k scalar functional constraints on the endpoints of x, then the
Transversality Condition imposes 2n - k scalar functional constraints on
the endpoints of p. Thus the "two-point boundary value problem" which
we must solve to obtain x has the "right" number of endpoint conditions.
28 1. Overview

To explore the relationship between the Maximum Principle and classi-


cal conditions in the Calculus of Variations, let us consider the following
refinement of the Basic Problem of the Calculus of Variations, in which a
pathwise constraint on the velocity variable x(t) E U is imposed. (U is a
given Borel subset of Rn'.)

Minimize fs L(t,x(t),i(t))dt
over arcs x satisfying (1.19)
i(t) E U,
(x(S),x(T)) E C .

The problem can be regarded as the following special case of (P):


Minimize z(T)
over measurable functions u
and arcs (x, z) E W','([S,T]; Rn+l) satisfying
(1.20)
(x(t), z(t)) = (u, L(t, x, u)) a.e.,
u(t) E U a.e.,
z(S) = 0, (x (S), x(T)) E C.

To be precise, x is a minimizer for (1.19) if and only if

=_ X, x, 2(t) L(T, _t (-r), x(T))dr)


\U JS
is a minimizer for (1.20).
Let x be a minimizer for (1.19). Under appropriate hypotheses, the Max-
imum Principle, applied to Problem (1.20), supplies A > 0 and an adjoint
arc (which, for convenience, we partition as (q, -a)) satisfying
(q, a, A) #0 (1.21)
= -a(t)L. (t, x(t), x(t)), a(t) = 0,
-cj(t) (1.22)
(q(S), -q(T)) = ', a(T) = A,
for some ' E NC(x(S),x(T)), and
q(t) x(t) - a(t)L(t, x, z(t)) = maUx{q(t) v - a(t)L(t, x, v)}, (1.23)

We now impose the "constraint qualification" :


(CQ): A can be chosen strictly positive.
Two special cases, in either of which (CQ) is automatically satisfied, are
(i) x(t) E int U on a subset of [S, T] having positive measure;
(ii) C = Co x Rn for some subset Co C Rn.
1.5 The Maximum Principle 29

(It is left to the reader to check that, if the condition A > 0 is violated,
then, in either case, (1.22) and (1.23) imply (q, a, A) = 0, in contradiction
of (1.21).)
It is known that a - 0. It follows that a(.) - A. In terms of p(t)
A-lq(t) and these conditions can be expressed
p(t) = L.(t, x(t), x(t)) a.e. (1.24)
(p(S), -p(T)) = , (1.25)

for some E NC(x(S),x(T)) and


p(t) x(t) - L(t, x, x(t)) = maUx{p(t) v - L(t, x, v)}. (1.26)

Notice that, when U = Rn, the Generalized Weierstrass Condition (1.26)


implies
p(t) E Lv(t, x(t), X(t)) a.e.
(Lv denotes the derivative of L(t, x,.).) This combines with (1.24) to give

(P(t), p(t)) = V ,vL(t, x(t), x(t)) a.e.


We have shown that the Maximum Principle subsumes the Euler-Lagrange
Condition and the Weierstrass Condition. But is has much wider implica-
tions, because it covers problems with pathwise velocity constraints.
The innovative aspects of the Maximum Principle, in relation to classical
optimality conditions, are most clearly revealed when it is compared with
the Hamilton Condition.
For Problem (1.19), the Maximum Principle associates with a minimizer
x an adjoint arc p satisfying relationships (1.24) to (1.26) above. These can
be expressed in terms of the Unmaximized Hamiltonian
71(t, x, p, v) = p v - L(t, x, v)
as

(-p(t),x(t)) = VX,vfl(t,x,p(t),x(t))
and
W(t, x(t), p(t), Xk(t)) = maUx{?-1(t, x(t), p(t), v)}.

On the other hand, classical conditions in the form of Hamilton's equations,


which are applicable in the case U = R' are also a set of differential
equations satisfied by x and an adjoint arc p,

(-1.(t), x(t)) = V ,vH(t, x(t),p(t)),


where H is the Hamiltonian
H(t, x, p) = max{p v - L(t, x, v)}.
vER'
Another random document with
no related content on Scribd:
sirommassa kaaressa vie baretin rinnalleen ja ylevämmin tekee
kumarruksensa?

Mutta vaikkapa myöntäisikin Oscar Wilden olevan jonkin verran


oikeassa hänen väittäessään herrasmiehen kaulaliinan solmun
osoittavan asianomaisen sivistystasoa, niin tekisi sekä
kulttuurikäsitteelle että Axel Ahlbergille veristä vääryyttä, jos tyytyisi
vain tuollaisiin mittapuihin. Ken on kuullut näyttelijän tarinoivan
taiteesta ja taiteilijoista, runoista ja runoilijoista, elämästä ja
ihmisistä sillä hienolla ja ymmärtäväisellä tavalla, joka on Ahlbergille
ominainen, hän käsittää täydelleen, että paljon syvemmistä hetteistä
kumpuaa hänen sivistyneisyytensä. Mutta perintönä saadut, kodissa
ja koulussa varttuneet, elämän varrella kartutetut "sisäiset aarteet"
hiovat ja jalostavat käytöstavan ulkonaisetkin ilmaisumuodot.

*****

Axel Ahlberg oli kauan Suomalaisen teatterin primo amoroso,


"ensimmäinen rakastaja." Melkein lukemattomat ovat ne hahmot,
joissa hän on lempensä sanoja kuiskaillut. Hän on vannonut Romeon
valoja ja Perdicanina tulkinnut Musset’n "katkeransuloista runoutta"
("Ei lempi leikin vuoksi"). — Mutta myöskin toisellaisia tehtäviä hän
on loistolla ja kunnialla edellisten rinnalla hoidellut; hänen
voittoretkiensä reittien viittoina ovat m.m. Hamlet, Othello, Antonius;
Sigismund, (Calderón); Gessler, Posa (Schiller); Oidipus, Kreon
(Sofokles); Örnulf, Nils Lykke, Rank, Helmer, Lövborg (Ibsen); Daniel
Hjort, Kustaa Aadolf, Klaus Kurki.

Eihän Ahlberg enää pitkiin aikoihin ole kauniina kuutamoöinä


kavunnut köysiportaita Capuletin taloon, ei muutamaan vuoteen
tulkinnut Hamletin nerokasta hulluutta — Romeot ja Tanskan prinssit
ovat vaihtuneet samoin kuin Juliat ja Ofeliat —, mutta yhä edelleen
hän on yhtä suuremmoinen, — ei, vaan vielä suuremmoisempi —
Klaus Kurki kuin viisikolmatta vuotta sitten, jolloin hänen vierellään
Kirstinä seisoi Ida Aalberg, ja Kustaa Aadolfina ei hänen kanssaan
tänäkään päivänä kukaan kilpasille kykene. Daniel Hjortin hän on
vaihtanut Stålarmiin — eikä yhtä ylevän jaloryhtistä Arvid Stålarmia
vielä näyttämöllä ole ollut.

*****

Axel Ahlbergin taide ei ole vivahdusrikasta — hän on kerta


kaikkiaan suurien piirteiden ja suorien viivojen mies —; se ei yllätä
meitä uusilla otteilla eikä odottamattomilla käänteillä, siinä eivät
väiky alati uudet vaihtelut. Mutta siitä huolimatta on näyttelijän
liikunta-ala sangen laaja. Takavuosina Ahlberg suorastaan
hämmästytti meidät Bahrin "Mestarissa": emme osanneet odottaa,
että hänen dekoratiivisella taiteellaan voisi luoda niin ilmielävää ja
suorastaan rehevää luonnekuvaa nykypäiväisestä ihmisestä, vaikka
olimmekin niin monet kerrat nähneet taiteilijan liukkaasti liikkuvan
kepeäpoljentoisen salonkinäytelmän parketeilla. Ja aivan äskettäin oli
meillä Wilden "näytelmässä hyvästä vaimosta" vereksin tilaisuus
näyttelijän huumorin hienouden toteamiseen.

Mitä tämä todistaa? Mitäpä muuta kuin sitä, että Axel Ahlbergin
taide alunpitäen tavallaan on ollut lähempänä niitä periaatteita ja
ihanteita, jotka ovat nykyaikaisen teatteritaiteen uudistuspyrintöjen
pohjana, lähempänä kuin monien sellaisten, jotka
todellisuustuntuisen realismin merkeissä ovat taidettaan kehitelleet.
Mutta ennen kaikkea se todistaa, että Axel Ahlberg on oikea
näyttelijä. "Schauspielen heisst: Menschen darstellen, dichterisch
wahre Menschen" — näytteleminen on ihmisten esittämistä,
runoilijan silmillä nähtyjen tosi-ihmisten. Näitä tosi-ihmisiä on
Ahlbergin näyttämökuvain galleriassa pitkä ja unohtumaton sarja.

*****

Axel Ahlbergin taide on vakaumuksen taidetta; siinä ei ole


humbuugia nimeksikään. Kuulee toisinaan sanottavan, että se
"vaikuttaa teatterilta" — sellainenhan lausetapa on yleinen. Eräissä
suhteissa se Ahlbergin taiteeseen nähden on paikallaankin. Siinä on
nimittäin teatteria, mutta — ja se on huomattava! — hyvää teatteria.
Sitä teatteria, jota Axel Ahlbergin vakaumuksen ja taiteellisten
ihanteiden mukaan hyvässä näyttämöluomassa tuleekin olla.

— — — Minusta on aina tuntunut siltä, ettei kenenkään


suomalaisen näyttelijän kilpilauseeksi sovi niin hyvin kuin Axel
Ahlbergin Cyranon uljaat repliikit:

"Yks' sentään mulle jäi, vaikk' köyhäksi mun teitte, se kirkas,


kaunis on ja vailla vammaa, vikaa, ei siinä pienintä oo tahraa eikä
likaa, — — — — — — — — — — ja liikkein kaunehin ja siroin
uljahasti se kaartuu alas auerkynnyksehen asti; ja korvissani kun
soi kutsu: sisään tulkaa, se mull' on… kannan…

ROXANE

Mitä?

CYRANO

Hatun töyhtösulkaa."
3.

Adolf Lindfors.

[Adolf Erik Lindfors syntyi Porvoossa toukokuun 18 p:nä 1857


(vanhemmat: kirjanpainaja Gustaf Adolf Lindfors ja Ulrika Gustava
Berg), kävi Helsingissä alkeiskoulua ja palveli Hofferin
valokuvaamossa, tuli sitten oppilaaksi Mallanderin perustamaan
yksityiseen teatterikouluun; oli Ruotsalaisen teatterin palveluksessa
1873—81, jonka jälkeen yritti merimieheksi, mutta liittyi jo syksyllä
1881 Fiorinin italialaiseen oopperaan Helsingissä, esiinnyttyään sitä
ennen aivan lyhyen ajan kiertävässä seurueessa Göteborgissa; kuului
Suomalaiseen teatteriin 1882—92; oli Helsingin dramaattisen
yhdistyksen oppilaskoulun johtajana 1892—94; sittemmin toiminut
Kansallisteatterissa, jonka johtajana oli 1907-09.]

Tässä on otettava puheeksi Adolf Lindforsin näyttelijätaide. Mutta


mistä alottaa ja miten määritellä sen olemusta? Ei ole helppoa keksiä
iskusanaa eikä kaavaa, johon tämän etevän näyttelijän taiteen
ytimen saisi keskitetyksi. Se liikkuu niin rikkaissa poljennoissa,
etteivät sen olennaisimmatkaan piirteet mahdu yhteen
määritelmään; sen heijastuspinnat ovat niin monisärmäisiä, että
ylimalkaiset laatusanat kutistuvat köyhiksi ja vähän sanoviksi. Ja niin
ollen jää ainoaksi keinoksi näyttelijän luomien muistiinpalauttaminen
yksitellen yhtenäisten säikeitten ja yhteisten ominaisuuksien
poimimista ja muistiinmerkitsemistä varten.

*****

Adolf Lindfors on suurin Molièretulkitsijana; hän on sillä alalla


aikansa kaikkein ensimmäisiä ja loistavimpia. Olkootpa esitettävänä
suuren mestarin, vallattominta farssia lähentelevät huvinäytelmät —
"Scapinin vehkeilyt", "Lääkäri vastoin tahtoansa", "Sievistelevät
hupsut" — tahi suuret komediat — esim. "Saituri" —, samalla
ylevällä taidolla ja samalla terävällä älykkyydellä Lindfors niiden
päähenkilöt hahmottelee ja elävöittää. Jok'ikisestä hänen
tulkitsemastaan suuren ranskalaisen luonnekuvasta näkee, kuinka
antautuvalla hartaudella esittäjä mestaria lähestyy. Eipä satu juuri
usein, että näyttelijä, joka muissa osissa kieltämättä toisinaan
lankeaa pieneen kiusaukseen ja helposti uhraa hivenen luontevata
inhimillisyyttä liioittelun alttarille — rivien huvittamisen luvattomassa
tarkoituksessa, että hän Molière-tehtävissä tinkisi tyylitajunsa ja
taiteellisen omantuntonsa vaatimuksista.

Varsinkin kaksi Lindforsin Molière-luomaa on muita kirkkaampina


muistiini jäänyt: saituri Harpagon ja luulosairas Argan.

Monien pitkien vuosien takaa nousee vieläkin silmiini ilmielävänä


Harpagonin hahmo vanhan Arkadian näyttämöllä, jolta sen
ensikerran näin. Kuinka suorastaan tärisyttävä olikaan Lindfors
neljännen näytöksen loppukohtauksessa, kun Harpagon aarteensa
menetettyään linkoaa vihlovan tuskansa kysymykset katsomoon:
"Missä minä olen? Kuka minä olen?" Näyttelijä astui askeleen
ramppia kohti, koko hänen olentonsa värisi tukahduttavan intohimon
paineen alla, sitten hän onnettomuutensa äärettömän taakan
taittamana vaipui maahan, ikään kuin sielunsa kadottaneena: "Minä
olen kuollut!" — Ja kun sitten — v. 1912 — uudelleen näin "Saiturin",
vaikutti tämä komedian tragediaa lähentelevä huippukohta vähintään
yhtä voimakkaasti, ehkäpä sen teho oli vielä entisestään
terästynytkin.
Entä "Luulosairas", joka viisitoistavuotisen väliajan jälkeen
uusintona esitettiin Lindforsin 60-vuotispäivänä! Millä verrattomalla
mestaruudella juhlija elävöittikään hypokondrisen Arganin hahmon ja
vuorosanat! — Tämä Argan on todella aivan ihmeteltävä
näyttämöluoma, täynnä eloa ja nerokkaasti keksittyjä detaljeja. Ja
kuitenkin ehyt ja johdonmukainen. Notkeasti taipuilee näyttelijän
lausunta — vierasperäisestä poljennostaan huolimatta — kaikkia
pienimpiäkin mielentilojen ja päähänpistojen heijastuksia
kuvastamaan mitä rikasvivahteisimman mimiikin säestämänä.
Asianlaita on niinkuin eräs arvostelija sanoo: "Monessa muussa
osassa näyttelijä ei karta kaikkein suurimpiakaan keinoja eikä
teatterimaisimpiakaan temppuja — aina asianhaarojen mukaan.
Tässä istuu hän enimmän aikansa rauhallisesti nojatuolissaan, vain
pienillä vivahduksilla, äänenväreillä ja kasvojeneleillä merkiten
henkilönsä eri mielentiloja. Ellei ohjelmassa olisi Molièren nimeä ja
ellei kappaleen sommittelu sekä ulkonainen asu toisi mieleemme
entisiä aikoja, luulisi näkevänsä jotakin modernista salonkinäytelmää,
mitä modernisimmän salonkinäyttelijän esittämänä."

*****

Lindforsin taide on detaljityötä. Luonteenomaiset pikkupiirteet ja


kuvaavat, mitä terävimmällä havaintokyvyllä huomatut yksityiskohdat
ovat rakennusaineina hänen luomissaan. Viiva viivalta, väri väriltä
hän tutkii ja punnitsee kuvattavan luonteen ominaisuudet ja kohta
kohdalta, käänne käänteeltä hän valmistaa tulkintansa. Hän on
sukua maalaustaiteen pointilisteille, jotka panevat kokoon taulunsa
vieri viereen asetetuista väripilkuista, mitkä katsojan silmissä sulavat
toisiinsa kiinteäksi kokonaisuudeksi. Niin ovat Lindforsinkin
näyttämöluomat koottuja miltei lukemattomista pilkuista ja pisteistä.
— On luonnollista, että tällaisen taiteen edellytyksenä täytyy olla
keinojen suvereeninen hallinta. Lindforsissa onkin teknillinen
taituruus huippuunsa kohonnut, hän on meidän näyttämömme
suurin taitaja; eikä vain meidän, sillä siinä suhteessa hänen
vertaisiaan täytyy etäämpääkin etsimällä etsiä. Mutta tämä
menetelmä edellyttää myöskin, että sen käyttäjällä on kylliksi
teatterikulttuuria voidakseen viljellyn maun ja erehtymättömän
asiantuntemuksen avulla ratkaista ja ennakolta arvioida laskelmiensa
pätevyyttä. Tiedämme, että Adolf Lindfors tässäkin suhteessa mitan
täyttää. Ja vielä kolmanneksi: detaljinäyttelijän täytyy onnistuakseen
ymmärtää ja osata yhdistää yksityiskohdat ja hajapiirteet kireäksi ja
eheäksi kokonaisuudeksi. Lindforsin Harpagon ja Argan ovat loistavia
todistuksia tällaisesta avusta.

*****

Sellainen taide, jolla Adolf Lindforsissa on etevin ja loistavin


edustajansa meillä, on järkeilevää taidetta. Se on kotoisin hyvästä
päästä eikä hellästä sydämestä. Ja sellaista taidetta vaanii hyvin
usein kuivakiskoisuuden vaara. Lindfors tuon vaaran välttää eloisalla
tempperamentillään — pienempilahjainen epigooni joutuisi
auttamattomasti karille.

Järkiperäisyys, sepä selittääkin, miksi Adolf Lindfors juuri


klassillisen komedian alalla — ei vain Molièren, vaan myöskin
Beaumarchais'n ja Holbergin näytelmissä — on kauneimmat
laakerinsa niittänyt. Hän on klassillisten komediarunoilijain läheinen
henkinen sukulainen. Hän on taipumuksiltaan yhtä intellektualistinen
kuin hekin. Ja siksi sopii juuri heidän tekstinsä paremmin kuin muut
hänen näyttämöluomiensa pohjaksi.

Mainitsin ohimennen Holbergin nimen. "Jeppe Niilonpojassa"


pitäisi kuitenkin hetkinen viivähtää, sillä se on myöskin luettava
Lindforsin kaikkein loistavimpiin saavutuksiin. Siinä on niin
ylenpalttinen pikkupiirteiden ja karakterististen mielijohteiden
rikkaus, että katsojan ihmettely muuttuu ihailuksi, etenkin kun joka
hetki havaitsee, kuinka terävään erittelyyn tuo runsaus pohjautuu ja
kuinka loogillinen johdonmukaisuus sitä aina ohjaa ja pitää koossa.
Ja senpä vuoksi katsoja niin mielellään on sulkevinaan silmänsä
niissä muutamissa kohdissa, joissa näyttelijä ei malta olla
pirahuttamatta pientä liikapisaraa ilomaljansa läikkyvään pikariin…

Klassillisesta repertoarista puhuttaessa ei tietenkään saa syrjäyttää


kansallisen päänäyttämömme runsasta Shakespeare-ohjelmistoa,
jossa Adolf Lindfors monta tärkeätä tehtävää on suorittanut,
mainitsen vain Fallstafin, "Myrskyn" Trinculon, kankuri Pulman
"Kesäyön unelmassa", Autolycuksen "Talvisessa tarinassa" ja
"Kuningas Learin" narrin esimerkkeinä Lindforsin mestarillisesta
osuudesta suuren brittiläisen henkilökuvien esittämisessä.
Shakespearen narrien älykkään leikinlaskun ja säkenöivien
pistosanojen maailmasta on jokseenkin pitkä harppaus siihen
sisällyspitoisuudeltaan varsin vaatimattomaan, mutta silti rehevään
ja mehukkaaseen porvarilliseen ilonpitoon, jota Fritz Reuter edustaa.
Adolf Lindfors osaa kuitenkin, jos tarvitaan, vieläkin pitempiä
askeleita ottaa. Hän maalailee oivallisin värein, joskaan ei ylen levein
siveltimenvedoin — viimeksi mainituthan eivät olekaan hänen
luonteensa mukaisia — inspehtorin herttaista naurettavuutta "Setä
Bräsigissä." Hän osaa temppuilla ja naurattaa; hän ei halveksi
minkäänlaisia keinoja, sillä hänen mielestään sopivat tämänlaisiin
tehtäviin yksinpä areenankin ilonpitäjien otteita muistuttavat
vivahteet.

*****
On luonnollista, että pienimpiä yksityiskohtia ja hiuksenhienoja
vivahteita hyväkseen käyttävä ja niihin perustuva "siselöimistyö"
vaatii laajaa esivalmistelua. Niinpä saattaakin jokaisesta Lindforsin
luomasta lukea — ikäänkuin rivien välistä —, että luontevan
liukkaitten liikkeitten, välähdyksinä vaihtuvien ilmeitten ja mainioitten
äänenpainojen takana on mitä tarkin harkinta ja huolellisin
valmistelu. Lindfors perustaa hahmottelunsa tunnollisiin tutkimuksiin.
Hän käyttää hyväkseen tarjolla olevat kirjalliset lähteet
syventyäkseen henkilönsä aikakauteen, hän miettii ja punnitsee sen
ajatusmaailman rakennetta, jossa kuvattava liikkuu, ja vasta tällaisen
tarkan analyysin jälkeen hän ryhtyy muovailutyöhönsä, jossa ei
mikään saa jäädä sattuman varaan. Niinpä saakin hän aikaan sen,
että esim. hänen verraton Napoleoninsa Sardoun "Rouva
Suorasuussa" virittää katsojaan oikean historiallisen tunnelman.
Hänen liikkeensä, asemansa, eleensä, ryhtinsä — koko kuva on aivan
niiden vaikutelmien mukainen, jotka muotokuvat, elämäkerrat ja
anekdootit ovat mieliimme syövyttäneet.

Välitöntä ei Lindforsin taide koskaan ole. Aina se on mietiskelyn ja


tarkkojen laskelmien tulosta. Siitä eivät soi syvät ja väräjävät
rintaäänet, siinä eivät tunnu elämän lämpimät ja suuret sykähtelyt
sellaisina, jollaisiksi me ne herkkinä hetkinämme havaitsemme. Siinä
ei ole "iäisyyspiirrettä", der Ewigkeitszug. Ja siksipä saattaakin
tapahtua, ettei tämä taide, loistavasta ja täydellisestä
muotovalmiudestaan huolimatta, meitä vakuuta. Tavallisesti Lindfors
kuitenkin tempaa katsojan niin kokonaan mukaansa, että kaikki
vastaväitteet vaikenevatpa me otamme taideluoman vastaan juuri
sellaisena kuin se meille tarjotaan. Kuinka monet kuvat
kirjallisuudesta ovatkaan meissä aivan kiinnikasvaneina Lindforsin
luomiin!
Kävisi tietysti mahdottomaksi luetella kaikkien — edes
suurempienkaan — näytelmähenkilöiden nimet, jotka Adolf Lindfors
on meille välittänyt. Niitä on niin paljon. On ranskalaisia
salonkikomediojen herroja — sillä alalla Lindfors, ranskalaisine
kouluineen ja parisilaisimpana kaikista Talian palvelijoistamme,
liikkuu yhtä varmasti kuin muut virkatoverit supisuomalaisessa
kansannäytelmässä —; on vakavampiakin miehiä, sekä hyviä että
huonoja ihmisiä — mutta miltei aina häikäisevän taidokkaita
näyttämöluomia.

Adolf Lindfors kuuluu kansallisen näyttämötaiteemme


uranuurtajapolveen; hänen hiihtämänsä latu kohti taitamisen ja
muotovalmiuden päämaaleja on ollut ja on edelleenkin oleva mitä
merkityksellisin näyttelijätaiteemme kehitykselle.

4.

Otto Närhi.

[Otto Närhi syntyi Rautalammilla helmikuun 28 p:nä v. 1864, oli


soittajana Kuopion pataljoonassa, ennen kuin v. 1884 liittyi
Aspegrenin Kansanteatteriin; Aspegrenin seurueesta hän keväällä
1892 siirtyi Suomalaiseen teatteriin, jossa ensimmäisen kerran
esiintyi syyskuun 14 p:nä 1892 "Mustalaisen" Petinä — arvostelu
merkitsi heti hänen näyttelijälahjansa "erinomaisiksi" —, saman
vuoden joulukuussa hän näytteli "Villisorsan" Gregers Werlen, jota
sitten seurasivat m.m. mielipuoli Amandus (Halben) "Nuoruudessa",
Onervo-noita "Kullervossa", Mitritsh "Pimeyden vallassa", Vaari
(Sjöströmin) "Marissa", Teemu "Nummisuutareissa", Störmer
(Dreyerin) "Koeopettajassa", Geronte "Scapinin vehkeilyissä", Reita
"Panussa." — Kun syksyllä 1903 näytäntökausi Kansallisteatterissa
alkoi, ei Otto Närhi enää kyennyt työtään jatkamaan; ankara
hermotauti oli pakottanut hänet sairaalaan, jossa hänen elämänsä
sammui toukokuun 7 p:nä 1904.]

Otto Närhi on suomalaisen näyttämötaiteen Aleksis Kivi, ei


samanarvoinen "suuruus" vaan sukulaisilmiö.

Muistoni hänen taiteestaan perustuvat pääasiallisesti vain kahden


luoman varaan: "Hanneleen" juopon isän ja "Myrskyn" Calibanin.
Mutta niiden kuvat ovatkin sieluuni lähtemättömiksi syöpyneet;
intensiivisemmin elettyjä rooleja en ole milloinkaan enkä missään
teatterissa nähnyt.

— — — Selailen vanhoja muistiinpanojani Hauptmannin näytelmän


esityksestä. Näen, että mielestäni Olga Leino on ollut erinomainen
Hannele — ja kun ajattelen hänen hentoa ja herkkää olemustaan ja
hänen taiteensa laatua, niin uskon yhä vieläkin, että hän oli aivan
erinomainen! — sekä Katri Rautio ylevä ja rakkautta säteilevä äiti ja
Iisakki Lattu mainio räätäli. Mutta kuitenkin on muurari Matternin
kuva ainoa, joka vielä tällä hetkellä ihan ilmielävänä eroittuu silmiini,
kun Hauptmannin unennäkyä vanhassa Arkadiassa ajattelen. Jos
osaisin käyttää piirrintä, tahi jos sanat riittäisivät, niin voisin milloin
tahansa loihtia sen pienimpiä yksityispiirteitä myöten toistenkin
nähtäväksi. Muistan sen ulkonaisen hahmon: paheen paaduttamat ja
pöhöttämät kasvot lippalakin ja nahkaesiliinan väliltä; kuulen vieläkin
raa'an äänensävyn, sammaltelevan puheen ja — ennen kaikkea —
näen katseen ja lyöntiin kohoavan käden… Se oli kappale
tärisyttävintä taidetta.

*****
Mimiikki oli ehkä kaikkein tehoisin myöskin Närhin taiteen
ilmaisukeinoista. Hän sai kasvojen eleisiin mahtumaan kokonaisia
tunteitten maailmoja. Hänen miimillinen asteikkonsa oli erinomaisen
laaja ja hän hallitsi sen salamannopeasti vaihtuvia vivahteita aivan
suvereenisesti. Symboolisilla eleillä hän osasi tulkita mielentilat,
maalailevilla havainnollistaa lauselmien sisällyksen ja alleviivaavilla
huomauttaa sanojen ja lauseitten merkityksellisyyttä ja tärkeyttä. —
Mutta myöskin äänikielensä hän oli kehittänyt taipuisan tottelevaksi
aseeksi. Hänen lausunnassaan oli iskevää nasevuutta, mutta
"liukuvan puheen" taidon hän myöskin oli omistanut. Hän puhui
pitämättä puheita. Tahi — kuten Coquelin vanhemmalla oli tapana
sanoa —: Hän ei näyttämöllä puhunut, vaan hän "sanoi"; puhuminen
on nimittäin vain sanojen kuuluviksi tekemistä äänen avulla,
"sanominen" tähtää määrättyyn sisällykseen. Saattaa puhua hyvinkin
paljon ja kuitenkin sanoa mitättömän vähän… Närhi ymmärsi
korkeus- ja mataluussuhteitten, valojen ja varjojen jakoperusteiden
tärkeyden ja osasi niiden varaan lausuntansa rakentaa.

*****

"Myrskyn" Caliban paljasti erään puolen näyttelijän taiteessa, joka


sille antoi niin yksilöllisesti erikoisen omintakeisuuden, että Otto
Närhiä olemme oikeutetut sanomaan näyttämötaiteen Aleksis
Kiveksi. Eino Leino on (kirjasessaan "Suomalainen näyttämötaide")
osunut tämän asian ytimeen: "Hra Närhi on luonnonvoima. Hänellä
on aivan oma tekotapansa, rohkea ja raju, terve-realistinen ja niin
sanoaksemme taiteen metsä-ilmassa viihtyväinen. Hän tuo
mukanaan näyttämölle sellaisen määrän tuoreutta ja alkuperäistä
voimaa, että yleisöä melkein pelottaa sen ylenpalttisuus." Calibanissa
purkautui tuo luonnonvoima väkevänä ilmoille. Shakespearen
valtaavan mielikuvituksen muovailemasta, puoleksi
vertauskuvallisesta ja ikään kuin kehitysopin teorioja ennustavasti
havainnollistavasta epäsikiöstä Närhi loi hämmästyttävän todellisen
olion: kammottavan ja kauhean, mutta irvokkaisuudessaan ja
avuttomuudessaan syvästi traagillisen satujen hirviön. Ja juuri tuo
viimeksi mainittu ominaisuus oli sen jalostavana aateluuden
merkkinä.

Närhin Calibanissa oli jotakin, mikä sitä muistellessa panee


ajattelemaan niitä kuvia, joita Shakespearen taruolennot ovat
Reinhardtin mieleen herättäneet. Niinpä esim. "Kesäyön unelman"
Puck, Reinhardtin käsityksen mukaan on oleva luonnon
alkuperäisinten ilmaisujen kuva — itse Urlachen der Natur. Sen tulee
olla lähellä eläimellisiä alkuvaistoja; Panin, paimenten pukinjalkaisen
suojelijan, makean mahlan juojan karhea ja harjaksinen, huhuileva
poika; Böcklinin pieni Pan, jonka ympärillä kesäyön satuhämärä
kiiltää, kuiskii ja tuoksuu… Mutta Reinhardtin näyttelijät eivät ole
kyenneet näitä olentoja havainnollistamaan — ei Eysoldtkaan — niin
selkeäpiirteisiksi ja samalla niin sielukkaan tosiksi kuin miksi tämä
suomalainen näyttelijä elävöitti väkevän mielikuvituksensa
hahmotteleman Caliban-kuvan.

Vähääkään liioittelematta saattaa sanoa "Myrskyn" esityksen juuri


tämän Caliban-tulkinnan takia muodostuneen merkkitapaukseksi
meidän teatterihistoriassamme.

*****

En tunne Otto Närhin sukupuuta. Mutta minusta tuntuu aivan


todennäköiseltä, että hänen suonissaan saattoi olla mustalaisverta.
Ei vain sen vuoksi, että hän voitti ensimmäiset taiteelliset voittonsa
puolivilleillä mustalaistyypeillään, joihin hän valoi taiteilijaluonteensa
vielä jalostamatonta, mutta sitä hehkuvampaa tulta. Vaan myöskin
— ja ennen kaikkea — siksi, että hänen luonnonvoimansa pysyi yhtä
alkuperäisen väkevänä ja hänen hehkunsa yhtä tummahohtoisena
vielä sittenkin, kun hänestä oli kypsynyt itsetietoinen taiteilija. Nuo
ominaisuudet olivat hänessä suku- ja heimoperintöä.

*****

Otto Närhin taiteessa oli nerouden merkki. Se eroittautui väkevästi


persoonallisen sävynsä takia ympäristöstä, mutta se taipui samalla
alistuvaisena kokonaisuutta palvelemaan. Sen komiikka — ja sitä
siinä oli paljon — oli perustuksiltaan jäyheää ja karkeasyistä;
samanlaista oli sen tragiikkakin. Sen luomat olivat kuin graniittiin
hakattuja. Mutta taiteilija oli oppinut vuosien varsilla raskaan
moukarin ohella käyttelemään myöskin keveämpää vasaraa ja
ohkaisempiteräistä talttaa; ja niin syntyi teoksia, joissa arkaistista
ankaruutta lievensi — mutta ei silti laimentanut — pehmeämpi
pintakäsittely. Mutta graniittiaineen vaatimaa erikoisluonnettaan ne
eivät milloinkaan kadottaneet, sillä ne olivat aina ja johdonmukaisesti
graniittiin ajateltuja. Siksi ne säilyivät omaperäisinä. Eikä Närhi
kuitenkaan koskaan väkipakosta tunkenut omaa persoonallisuuttaan
osiinsa aikaansaadakseen omintakeisen vaikutelman; hänen ei
tarvinnut sitä tehdä, tuo vaikutelma syntyi itsestään.

Sellaista oli tämän miehen taide, miehen, joka seisoo aivan yksin
suomalaisen teatterin nuoressa historiassa. Moni suomalainen
näyttelijä — ajattelen esim. parhaassa miehuudessaan manalle
mennyttä Kaarlo Braxénia — on astellut samansuuntaisia teitä kuin
Otto Närhi, mutta ei yksikään niitä polkuja yhtä pitkälle kuin hän.
5.

Mimmi Lähteenoja.

[Mimmi Lähteenoja (ent. nimi Puls) syntyi Helsingissä elokuun 27


p:nä 1865 pohjalaisista vanhemmista. Hän on — omien sanojensa
mukaan — käynyt kolmea koulua: kansakoulun, rippikoulun ja
maailman koulua. Esiinnyttyään seuranäyttelijänä hän monien
kehoituksesta liittyi August Aspegrenin seurueeseen elokuun 30 p:nä
1888 esiintyen ensimmäisen kerran tullinhoitajan rouvana "Pukkisen
pidoissa" (Hämeenlinnassa lokak. 2 p:nä). Helmikuun 5 p:nä 1892
hän sai "tohtorilta" kirjeen, jossa ilmoitettiin, että hänet oli
hyväksytty Suomalaiseen teatteriin; "hauskempaa Runebergin päivää
en milloinkaan ole viettänyt", sanoo neiti Lähteenoja. Ollut sekä
valtion että teatterin myöntämillä matkarahoilla ulkomailla.]

Mimmi Lähteenoja tuli Aspegrenin seurueesta Suomalaiseen


teatteriin.
Hän oli silloin liikkunut näyttämöllä neljä vuotta ja oli 26-vuotias.

Hänen aikaisempia näyttämötehtäviään pääkaupungissa olivat


Fekluska Ostrowskin "Ukkosilmassa", Gina Ekdal Ibsenin
"Villisorsassa", Katri "Daniel Hjortissa", Rotta-Ulla Ibsenin
näytelmässä "Pikku Eyolf", Armi J.H. Erkon "Kullervossa", Martta
Sjöströmin "Marissa", Noita Kasimir Leinon näytelmässä "Jaakko
Ilkka ja Klaus Fleming", Louhi Erkon "Pohjolan häissä" ja
kalastajanleski Heyermansin näytelmässä "Siunauksen toivo."
Myöhempiä on turha luetella — ja luettelo venyisikin liian pitkäksi —;
ne ovat vielä elävinä osin muistissamme, osin silmiemme edessä.
Kenenpä mieleen eivät olisikaan syöpyneet nuo melkein
lukemattomat akkatyypit, jotka Mimmi Lähteenoja merkillisen
varmalla luonnekuvauskyvyllään ja hienotunteisen taiteellisella
tekotavallaan on elävöittänyt!

*****

Mimmi Lähteenoja on älykäs taiteilija. Hänen älykkyytensä ilmenee


jo siinäkin, että hän alunpitäen on ollut selvillä omasta alastaan.
Luulen hänen aina olleen vapaan näyttelijämaailman ylen tavallisesta
ja ylen narrimaisesta turhamaisuudesta, joka saa niin näyttämön
herrat kuin sen naisetkin halajamaan itselleen "nuoria" ja "kauniita"
osia. Hän tuntee lahjojensa rajat ja tietää, että rajoituksessa
myöskin piilee voima. Mutta hänen älykkyytensä ilmenee myöskin
hänen luomissaan, joiden takana aina on mitä tarkin ja huolellisin
harkinta ja tutkimustyö. Tässä suhteessa on hänen ja Adolf
Lindforsin taiteella selviä kosketuskohtia.

Sukulaispiirre on myöskin detaljirikkaus, jota nti Lähteenojakin,


kuten kaikki realistiseen taidekäsitykseen pohjautuvat taiteilijat,
viljelee. Hänenkin luomissaan pistävät silmiin nuo erinomaiset ja
todellisuustuntuiset pikkupiirteet, usein aivan vähäiset ja ikäänkuin
ohimennen, jopa aivan kuin sattumaltakin vetäistyt, mutta aina
luonnekuvaan sopivat, sitä selventävät ja jotakin ominaisuutta
alleviivaavat. Nti Lähteenoja ei kuitenkaan perusta tulkintaansa niin
suuressa määrin yksityiskohtien varaan kuin Lindfors, joka terävän
analyysinsä eroittelemista alkutekijöistä ja irralleen purkamista
rakennuspuista kokoonpanee kuvansa alusta loppuun.
Näyttelijättäremme menetelmä on alunpitäen enemmän syntetisoiva
— se tietysti voisi johtua siitäkin, ettei hänen tekniikkansa
riittäisikään Lindforsin tuhatvivahteiseen taituruuteen taipumaan,
mutta varmaankin se pohjimmiltaan sittenkin riippuu
taiteilijaluonteen laadun erilaisuudesta.
Sillä aivan olennaisia eroavaisuuksiakin on olemassa.
Silmiinpistävin niistä on ehkä nti Lähteenojan tunneherkkyys, joka
ilmenee milloin herttaisena sydämellisyytenä, milloin jonakin
toisenlaisena välittömänä tunteen läikähtelynä, jopa joskus kohoaa
voimakkaaksi paatokseksikin — esim. Katrissa. Jonkun kerran
korostuu tämä tunteellisuus pieneksi senttimenttaalisuudenkin
pisaraksi. Se tapahtuu kuitenkin harvoin, sillä yleensä on Mimmi
Lähteenojan taide hyvin tervettä.

Todellinen ja vakuuttava ja elävä, ne lienevät muuten arvostelun


useimmiten käyttämät laatusanat näyttelijättären luomista. Noilla,
tosin verrattain vähän sanovilla, mutta kuitenkin kaikessa
ylimalkaisuudessaan oikeaan osuvilla adjektiiveillä voikin määritellä
useimpia hänen tulkintansa iskevimpiä ominaisuuksia. Hänen
esityksensä ovat todellisia siksi, että ne ovat erehtymättömän
varmalla vaistolla todellisuushavaintojen pohjalle tyylitellyt; ne ovat
vakuuttavia siksi, että eleitten ja äänenpainojen takaa aina erottaa
sydämen sykkeen; ne ovat eläviä siksi, että näyttelijättärellä on
hartaan antaumuksen ja välittömän myötäelämisen ihana kyky ja
taito.

Mimmi Lähteenojan ohjelmisto on vuosien varsilla ollut jokseenkin


samanlaatuinen. Siitä huolimatta hän kykenee miltei jokaisesta
uudesta tehtävästään saamaan irti uuden muunnelman. Se
merkitsee, että hänen asteikkonsa, ohjelmiston ahtaasta
liikkumisalasta huolimatta, on harvinaisen laaja. Ja edelleen se
merkitsee, että myöskin hänen tekniikkansa on joustava ja varma,
kun se ulottuu aina uusiin ilmaisumuotoihin, vaikka sisällyksen — jos
niin saa sanoa — eroavaisuus usein on kutakuinkin hiuksenhieno.
Mehän tiedämmekin nti Lähteenojalla olevan tukenaan kaksi oivaa
avua: taipuisa mimiikki (siihen liittyy myöskin varma
naamioimistaito) ja vivahdusrikas ääni, jota näyttelijätär
erinomaisella taidolla käyttelee.

*****

— Lopuksi pieni persoonallinen muiston sirpale, ehkä itsessään


aivan merkityksetön, mutta kenties kuitenkin omiaan osaltaan vähän
valaisemaan Mimmi Lähteenojan taiteellista työskentelymetoodia!

Istuin aamupäivää näyttelijättären aurinkoisen kesäkodin


parvekkeella Mankalassa. Oli kuunneltu linnunlaulut pihan vihreässä
koivikossa, oli käyty näköalapaikat joen töyräällä ja tarkasteltu kukat
ja kylvömaat. Ja miten olikaan, niin työntyi kahvikuppien ääressä
tarina teatteriin — kuten aina tapahtuu siellä, missä kaksi tahi kolme
teatteri-ihmistä yhteen tulee. Mimmi Lähteenoja pakisi puheliaalla
tavallaan muistoistaan ja entisistä ajoista, "tohtorin" päivistä ja Ida
Aalbergin vierailuista. Ja vähitellen myöskin muusta ja muista.
Ehkäpä hän ei itse huomannutkaan, että hänen siinä juttuja
kertoessaan liikkeet, eleet ja ääni taipuivat matkimaan niiden
henkilöiden ilmaisukieltä, joiden vuorosanoja hän kulloinkin kertaili.
Imitatsionia oli ilo katsella ja kuunnella; siinä välähtelivät "mallin"
erikoisuudet, tietysti karrikoidun alleviivattuina, mutta sitä
selkeämmin havaittavina:

— Muistatteko, kun hän tuli näin ja teki tämän liikkeen ja sanoi —


—?

Miten terävän huomiokyvyn ja miten tarkan havaintojen


muistamisen taidon tuo matkiminen paljastikaan! Muistiini juolahti
erään kuuluisan ruotsalaisen näyttelijän kerran linkoama paradoksi:
"Eihän hänestä voi tulla näyttelijää; hän matkii ihmisiä niin perin
huonosti." Ja kun joku pyysi selitystä, jatkoi hän tähän tapaan:
"Näyttelijän täytyy usein käyttää elävää mallia, jolta hän lainaa
peruspiirteet näyttämön hahmoon. Hän ei tietenkään kopioi
todellisen elämän ihmistä sellaisenaan, vaan hän ottaa mallista
muutamat piirteet, joihin hän liittää toisia, muualta kokoomiaan
huomioita ja mielikuvituksensa luomia lisiä."

Jotenkin tuohon tapaan arvailin siinä parvekkeella istuessani


monen Mimmi Lähteenojankin luoman syntyneen. Niissä on niin
väkevänä ja välittömänä elämän väri, että katsoja ehdottomasti tulee
ajatelleeksi niiden olevan kirkastettujen todellisuushavaintojen
pohjalle rakennettuja.

6.

Kirsti Suonio.

[Kirsti Suonio, o.s. Kjäll, syntyi Helsingissä heinäkuun 24 p:nä


1872 (vanhemmat: työmies Erik Johan Kjäll ja Erika Wass); käytyään
kansakoulun nautti yksityisopetusta; hyväksyttiin oppilaaksi
Suomalaiseen teatteriin joulukuun 17 p:nä 1890 ja on siitä lähtien
yhtämittaa palvellut kansallista päänäyttämöämme; esiintyi
ensimmäisen kerran näyttämöllä tammikuun 1 p:nä 1891 Kiven
näytelmässä "Yö ja päivä" Annan (Liisan äidin) osassa; ollut valtion
matka-apurahalla ulkomailla.]

Kukapa teatterinharrastaja ei tuntisi Kirsti Suonion Suorasuuta!


Sehän on niin erinomaisen elävä ja veikeä näyttämöluoma. Siinä on
mehevän huumorin ohella myöskin sydäntä, joten hyväntuulisuus on
mitä herttaisinta. Kuinka verrattomasti tämä Catherine
sinkahuttaakaan sutkauksiaan, kuinka vakuuttavan naivia onkaan
hänen kursailemattomuutensa! Danzigin herttuatar herättää meidän
vilpittömimmän myötätuntomme; Kirsti Suonio tempaa meidät
vastustamattomasti mukaansa. Tuskinpa muistamme panna merkille
sitäkään, että tästä Catherinestä tykkänään puuttuu ranskalaisuus.
Vasta perästäpäin tulemme ajatelleeksi, että niin pesijätär kuin
urhean Lefèbvren puolisokin olivat aivan — supisuomalaisia. Kirsti
Suonion Suorasuu on reipas ja rivakka, välittömän sydämellinen ja
aurinkoisen avomielinen ilmestys; ja nämä avut kai siinä
korvaavatkin gallialaisen vilkkauden ja synnynnäisen keveyden
puutteen. Joka tapauksessa: suomalainen teatteriyleisö laskee tämän
esityksen näyttämönsä klassillisiin luomiin, aivan kuten Lindforsin
Napoleoninkin.

Eikä Suorasuu kuitenkaan ole Kirsti Suonion, puhtaasti


taiteellisessa suhteessa, kaikkein täysipitoisimpia saavutuksia.
Nähdäkseni hän monetkin kerrat on näyttelijänä kohonnut
korkeammalle kuin Sardoun historiallisessa komediassa; osunut
paremmin ja täydellisemmin elävöitettävänään olevan ihmisen
ytimeen ja hermoihin. Mutta siitä huolimatta Catherine on yleisön
silmissä ikään kuin keskeisenä merkkiluomana Kirsti Suonion
"tuotannossa." Se johtunee ensiksikin itse roolista ja toiseksi siitä
seikasta, että Suorasuu oli näyttelijättären ensimmäinen loistava
"lupaus" hänen uransa alkutaipaleella.

Yllä on jo viitattu niihin Kirsti Suonion päivänpaisteisen


taiteilijaluonteen ominaisuuksiin, jotka välittömimmin katsomoon
tehoavat. Mutta ne eivät vielä — niin merkityksellisiä kuin ovatkin —
riitä hänen taiteensa määritelmäksi. Nuo luontaiset ominaisuudet
ovat näyttämön taiteilijalle arvokkaita avuja, mutta yksin niiden
varaan ei voi syvällisempää taidetta perustaa.
Kirsti Suonion taide on rehellistä ja tervettä. Siinä ei ole mitään
teennäistä, ei pakotuksen jälkeä. Taiteilijatar kulkee halki elämän
avoimin silmin, ja hänen silmänsä verkkokalvo on herkistynyt
vaikutelmia vastaanottamaan. Mutta hänellä on myöskin varma
valinnan kyky: hän osaa huomioistaan seuloa olennaisimmat detaljit,
ne, jotka asian ytimen ilmaisevat. Ja niitä hän käyttää luomissaan.
Hän ei rasita henkilökuviaan kaikilla siihen mahdollisesti sopivilla
piirteillä; se pelastaa hänet pikku realismin arkisesta ahtaudesta.

Hänen taiteensa on realistista alkujuurta, mutta siinä on myöskin


pieni romantiikan pisara; se on kai kotoisin siitä näyttelijättären
ominaisuudesta, jota saattaisimme sanoa tunnekulttuuriksi —
luontaista herkkää tunnettaan hän näet kernaasti runontuoksuiseksi
hienostaa. Kirsti Suonio on kauniimmat voitonhetkensä viettänyt
liikkuessaan niillä aloilla, joilla hän on tuntenut todellisuuden vankan
maaperän jalkainsa alla. Mutta hän ei ole yksipuolinen. Myöskin
mielikuvituksen ilmavissa maailmoissa hän on valloitusretkillä
pistäytynyt. Niinpä on — vain yhden esimerkin mainitakseni —
muistiini jäänyt hänen mehevällä huumorilla ja levein ottein
tulkitsemansa (Larin Kyöstin) "Taikapeilin" hovirouva; se oli koko
esityksen aidoin ja ehein satukirjaolento.

Kysyin kerran Kirsti Suoniolta, mistä osistaan hän itse enimmin


pitää. "Oh, niitä on niin paljon sellaisia, joista pidän, että on vähän
vaikea luetella. Olenhan näytellyt ihan kaikenlaisia — ingenueitä en
kuitenkaan koskaan, en edes vihreimmässä nuoruudessani. Mutta
jos nyt joitakin erityisesti rakastamiani ajattelen, niin johtuvat ehkä
ensimmäisinä mieleeni Anna Mellilä (Tavaststjernan) 'Uramon
torpassa', Anna Liisa ja Maija 'Tukkijoella.' Niin, ja sitä paitsi
venäläiset akat, joista pidän aivan hirveästi."
Eikö tämä tunnustus tue juuri lausumaamme ajatusta, että Kirsti
Suonion taide parhaiten soveltuu todellisuuden maaperästä
välittömästi versonneita henkilökuvia elävöittämään?

Ja mitä noihin "venäläisiin akkoihin" tulee, niin on Kirsti Suoniolla


juuri niissä ollut tilaisuus käytellä hyväntuulisen huumorinsa
höystämää, luistavaa luonnehtimistaitoaan — jota usein on tukenut
viisaasti ja taidolla käytetty eloqventia corporis — ja edellä mainittua
romanttista vivahdettaan.

Kirsti Suonion huumori ansaitsisi oman erikoisen lukunsa. Siinä


ovat pääominaisuuksina, kuten sanottu, herttaisuus ja
suorasukaisuus. Se merkitsee, että se on vapaata katkeruudesta ja
vihlovasta ivasta; siinä on hyvin vähän ilkeyttä, mutta sitä enemmän
suvaitsevaa ymmärtäväisyyttä ja aurinkoista anteeksiantavaisuutta.
Siinä on — niin sanoakseni — aivan itsetiedoton, tahi ainakin siltä
näyttävä, filosoofinen piirre, joka on sukua Diogeneen
ylevämmyydelle.

*****

Kirsti Suonio ei ole niitä taiteilijoita, jotka aina osuvat naulan


päähän. Toisinaan sattuvat hänen vasaraniskunsa koko paljonkin
syrjään hairahtumaan — esim. silloin kuin hänen on liikuttava
ympäristössä, joka on koko hänen rehevälle olemukselleen vieras.
Kirsti Suonio on liiaksi tuores ilmestys mahtuakseen ahtaisiin
budoaareihin ja liikkuakseen aivan kompastumatta liukkailla
parkettipermannoilla. Salonkinäytelmässä häneen jää aina ja
auttamattomasti sirunen Suorasuuta ja konversatsionikappaleissa
hiven — Maijaa. Mutta mitä se todistaa? Vain sitä, että Kirsti Suonio
aina ja empimättä uskaltaa ja tahtoo olla rehellinen. Ja loppujen
lopuksi pitää taiteessakin paikkansa esi-isiemme sananlaskuun
keskittämä elämänviisaus: rehellisyys maan perii! — Satutteko
muistamaan Kirsti Suonion näyttämönaurua? Se on harvinaisen
tarttuvaa. Se tulee sydämen pohjasta. Sellainen nauru ei ole aivan
jokapäiväistä näyttämön maailmassa. Omasta puolestani luulen, että
se on harvinaisempaa kuin oikeat kyyneleet…
MUUTAMIA
NÄYTTÄMÖESITYKSIÄ.

1.

"Venetsian kauppias."

(Kansallisteatteri 1/III 1917).

On jo monta vuotta vierähtänyt siitä, kun Shylockin vuorosanat


viimeksi kielellämme kajahtivat. Jäähyväisnäytännössä vanhassa
Arkadiassa — viisitoista vuotta takaperin — esitettiin tosin neljäs
näytös, mutta koko näytelmää ei suomalainen yleisö ole nähnyt
sitten vuoden 1898, jolloin se — huhtikuun 13 p:nä — oli Benjamin
Leino-vainajan 25-vuotisjuhlan ohjelmana.

Tämä Shakespearen valtaava näytelmä liitettiin Suomalaisen


teatterin ohjelmistoon jo vuonna 1882; sen ensi-ilta oli Viipurissa
joulukuun 1 p:nä mainittuna vuonna. Helsingissä se näyteltiin
ensimmäisen kerran huhtikuun 13 p:nä 1883.

"Venetsian kauppias" oli järjestyksessä toinen suuren brittiläisen


mestariteoksista, johon suomalainen yleisö näyttämönsä palkeilta sai
tutustua; ensimmäinen oli kaksi vuotta aikaisemmin esitetty "Romeo
ja Julia."

Teatterihistorioitsijamme sanoo, että "Venetsian kauppias" herätti


tuskin vähempää juhlatunnelmaa kuin "Romeo ja Julia."
Menestyksenhän väittää riippuneen "paremmin oivallisesta
yhteisnäyttelemisestä ja näyttämölle panosta kuin yksityisten
näyttelijäin taiteesta." Näyttämölle-asetus lieneekin ollut oloihin
nähden aivan erinomainen. Niinpä kerrotaan karnevaaliyön
gondooleineen, serenaadeineen ja riehuvine naamiojoukkoineen,
samoin kuin kuutamokohtauksen Belmontin puistossa, Portian
juhlasalin j.n.e. "antaneen enemmän kuin oli osattu odottaa."

Yksityisistä osista näyteltiin Shylock ja Portia etevimmin.


Edellisessä tehtävässä oli Benjamin Leino "voimakas ja
luonteenomainen ja hänen mahtavassa ulkonaisessa olennossaan oli
jotakin, joka sai ajattelemaan langennutta patriarkkaa." Esityksen
huipuksi kohosivat kohtaukset Tubalin kanssa ja oikeuden edessä,
joissa koronkiskurin intohimo nousee korkeimmilleen. "Niissä", sanoo
Aspelin-Haapkylä, "Leino merkillisellä kyvyllä käytti lahjojaan,
saavuttamatta kuitenkaan korkeinta kantaa, jossa yksityispiirteiden
tarkoitus kerrassaan katoaa kokonaisuuteen."

Sekin, jolle ei ole suotu nähdä Benjamin Leino-vainajaa


Shylockina, voi uskoa ja ymmärtää, että tämä etevä luonnenäyttelijä,
jonka suurpiirteinen ja tärisyttävän vaikuttava kuningas Lear kuuluu
muistista hälvenemättömiin luomiin, Venetsian kauppiaana oli kuin
"luku Mooseksen laista."

Portiana oli ensiesityksessä Kaarola Avellan. Sanotaan, että hänen


luomansa oli "yhtä huolekas kuin taidokas käsityksen ja esityksen
puolesta ja nousi korkeimmilleen tuomiosalissa, missä hänen,
koettaessaan lepyttää julmaa juutalaista, onnistui saada sydämen
ääni kuuluville." Väitetään Portian olleen nti Avellanin kaikkein
parhaita osia.

Muista esiintyjistä mainittakoon Inez Borg Jessikana, Böök


suruttomana ja ritarillisena Bassaniona, Lindfors hullunkurisena
Lancelot Gobbona ja Ahlberg Antoniona.

Kun "Venetsian kauppias" sitten v. 1898 uusintona esitettiin, oli


Portiana Katri Rautio ja Jessikana nti Hertzberg.

*****

"Venetsian kauppiaan" uusinnossa näimme ensi kerran Suomessa


käytännössä n.s. kiertonäyttämön ("Drehbühne"). Muutamat tiedot
sen rakenteesta ja tarkoituksesta puolustanevat sen vuoksi tässä
yhteydessä paikkaansa.

Shakespearen draamojen pahimpana näyttämöteknillisenä


kompastuskivenä ovat niiden lukuisat näyttämömuutokset. Mestarin
oman aikakauden teattereissa ne eivät sanottavia häiriöitä
tuottaneet. Yleisesti tunnettuahan on, miten hänen primitivisessä
teatterissaan aivan yksinkertaisin ja katsojien välittömään ja
herkkään mielikuvittelukykyyn perustuvin keinoin kulissivaihdosten
pulista pelastauduttiin.
Shakespearen teatterin näyttämönä oli aivan yksinkertainen lava,
joka eteni katsomon keskelle, niin että näyttelijöitä voitiin katsella
useammilta puolilta. Teatterin keskiosa, johon saakka näyttämö
ulottui, oli katoton, ja näytännöt tapahtuivat päivänvalossa.
Maalattuja kulisseja ei lainkaan ollut. Näyttämön taustana kohosi
puusta rakennettu luolan tapainen laite, jonka aukon peitti verho ja
jota käytettiin mitä erilaisimpiin tarkoituksiin: se oli esim. Julian
hautana ja Desdemonan vuodekammiona. Tämän rakenteen yläosa
taasen esitti Julian parveketta ja Shylockin talon ikkunaa, josta
Jessika heittää ulos lippaan, j.n.e.

Ei ollut siis suinkaan monimutkainen sen näyttämön rakenne, jota


varten mestari suurteoksiaan loi. Sen ajan ihmisille olivat kuitenkin
nuo yksinkertaiset puitteet aivan riittävät runon tapahtumien
taustaksi. Mutta mitäpä sanoisikaan meidän päiviemme yleisö, jos
sille tarjottaisiin näytelmä yhtä karuissa kehyksissä? Tokkopahan se
jaksaisi uskoa meren olevan jollakin näyttämön kohdalla, jos pelkkä
tienviittataulu sen vakuuttaisi siinä olevan? Varmaankaan ei! Se on jo
aikoja sitten menettänyt herkästi uskovaisen lapsenmielen, sen
mielikuvitus on tylsistynyt. Sen täytyy saada nähdä meren sineä ja
välkkyviä laineita. Sen täytyy kuulla ukkosen jyrinää ja tuulen
tohinaa uskoakseen rajuilman raivoavan.

Miten tulee siis nykyaikaisen ohjaajan menetellä tyydyttääkseen


yleisönsä illusionin tarvetta, silti ylenmäärin esitystä pitkittämättä ja
kovin pahoin draaman eheyttä särkemättä — siinä problemi, jonka
ratkaisua monin keinoin on yritetty. Kokeiluista mainittakoon tässä
muutamia. Otto Dewrientin yritys uudelleen rakentaa keskiajan
kolmikerroksinen mysterionäyttämö (alkujaan ja etupäässä "Faust"-
esityksiä varten) voidaan tässä yhteydessä pelkällä maininnalla
syrjäyttää. Merkityksellisempi on sen sijaan Lautenschlägerin
suunnittelema v. 1889 ensi kerran käytäntöön ottama Shakespeare-
näyttämö. Siinä oli muuttumaton etunäyttämö, jonka verho erotti
taka-alasta, mitä muutettiin sill’aikaa kun toiminta etunäyttämöllä
häiriytymättä jatkui.

Lautenschlägerin periaatetta, joka tavallaan on illusioninäyttämön


ja tyyliteltyjen puitteiden välimuoto, on edelleen kehittänyt m.m.
Hagemann Mannheimissa, rakentamalla kolmiosaisena kohoavan
näyttämön pilareineen ja erivärisine jakajaverhoineen. Myöskin
Kansallisteatterimme on osaksi tätä esimerkkiä seurannut "Hamlet"-
esityksessään.

Tällaisella näyttämöjärjestelyllä on aivan ilmeiset etunsa: se sallii


draamallisen runon vaikuttaa etupäässä omalla elävällä voimallaan ja
sisäisellä tehollaan. Mutta se vaatii ehyttä ja etevää tulkintaa, joka ei
hetkeksikään saa herpautua eikä kalveta. Toisin sanoen: se paljastaa
näyttelijäluomien ja yhteisesityksen heikkoudet paljon
säälimättömämmin kuin illusioniperiaatetta täydelleen noudattava
menettely, joka suo ohjaajalle tilaisuuden levittää nähtäväksemme
nykyaikaisen teatterin koko kulissikomeuden ja sillä ainakin osaksi
peittää mahdollisesti esiintyviä näyttelijätaiteellisia heikkouksia.

Illusioniperiaatetta ovat Shakespeare-esityksissään noudattaneet


esim.
Reinhardt ja Beerbohm Tree. Se on käynyt mahdolliseksi
kiertonäyttämön
avulla, joka sekin on Lautenschlägerin keksimä ja ensi kerran
Miinchenissä koettelema ("Don Juan"-esityksessä).

Lautenschläger rakennutti näyttämön lattialle toisen näyttämön,


ympyränmuotoisen ja akselinsa ympäri pyörivän tason (jonka
lävistäjä oli 16 m). Tälle kiertonäyttämölle hän rakensi eri
kuvaelmien kehykset siten, että kuvaelman loputtua seuraava
dekoratsioni heti voitiin heilauttaa näyttämöaukon eteen. Kulloinkin
käyttämättä oleville ympyrän osille voitiin siis uusia puitteita
rakennella näytöksien aikana mielin määrin.

Kiertonäyttämön käytännölliset edut huomattiin heti, ja


mannermaan suurimpiin teattereihin on useimpiin pyörivät lattiat
rakennettu. Tälläkin keksinnöllä on kuitenkin vakavat vikansa ja
voittamattomat varjopuolensa. Pahimpana puolena siinä on se
seikka, että kulloinkin käytettävänä oleva ala on liiaksi rajoitettu eikä
niin muodoin salli tarpeeksi pitkien perspektiivien ja syvyyksien
käyttämistä. Mutta siitä huolimatta on erittäin huomattaviin
koristesaavutuksiin kiertonäyttämöllä päästy. Ovathan esim. juuri
Reinhardtin Shakespeare-asettelut — nimenomaan "Kesäyön
unelma", mutta myöskin "Venetsian kauppias" — saaneet osakseen
hyvinkin suurta tunnustusta.

*****

Useimmat "Venetsian kauppiaan" näyttämötulkitsijat ovat


kuninkaallisen kauppiaan tarinan ympärille rakentuvasta
huvinäytelmästä tehneet Shylockin kovia kohtaloita esittävän
murhenäytelmän, jossa venetsialainen ilonpito muodostaa vain
valoisan taustan, jota vastaan jo itsessään synkän juutalaisen hahmo
vieläkin synkemmäksi piirtyy. Se onkin aivan luonnollista, sillä
läheneehän — kuten Valter Raleigh huomauttaa — koko näytelmän
toiminta tragedian rajoja ja pelastuu töin tuskin suistumasta
synkeyden kuiluun. Eikähän kukaan voinekaan kieltää sitä tosiasiaa,
että juutalainen ottaa valtaansa miltei koko näytelmän, niin että
hänen murheellinen tarinansa muodostaa sen varsinaisen ja
olennaisimman ytimen. Yksinpä loppunäytöksen lemmenleikittelyä ja
kuutamoista romantiikkaa uhkuvaan keveäpoljentoiseen
huvinäytelmäänkin Shylockin tumma muisto vaimentavia varjojaan
heittää.

Tällainen huvinäytelmän murhenäytelmäksi muuttaminen perustuu


vanhaan ja vakiintuneeseen perintätapaan. Johan aikoinaan Heine,
muistellessaan Shylockin takia vuotaneita kauniita kyyneleitä, jotka
hän Drurylanen aitiossa oli nähnyt nousevan englantilaisen
kaunottaren suuriin silmiin, luki "Venetsian kauppiaan"
murhenäytelmiin kuuluvaksi, "joskin kappaleen puitteita iloiset
naamiot, satyyrikuvat ja Amorit koristavat ja vaikkapa runoilijakin
alkujaan on tahtonut meille huvinäytelmän antaa." Mutta vetoaa
Heine sentään muihinkin syihin eikä vain jonkun kauniin
englannittaren huudahdukseen: "The poor man is wronged!" (mies-
paralle tehdään vääryyttä!). Hän sanoo oiettavansa Shakespearen
tahtoneen suuren yleisön ratoksi kuvata kiusatun ihmissuden,
vihatun taruhirviön, joka verta janotessaan menettää tyttärensä ja
tukaattinsa ja joutuu vielä päälle päätteeksi ivan alaiseksi. Mutta —
jatkaa hän — runottaren tahto on ollut runoilijan yksityistä tahtoa
voimakkaampi, ja niin onkin tapahtunut, että Shakespearen Shylock
on, räikeästä irmastelustaan huolimatta, muodostunut onnettoman
lahkokunnan syyttömyyden todistajaksi, heimon, jonka kannettavaksi
sallimuksen tutkimaton tahto on määrännyt alhaisen ja ylhäisen
roskaväen sammumattoman vihan, mitä se ei aina suostu
rakkaudella kostamaan. Toisin sanoen: Shylock on saavuttanut siihen
määrin Shakespearen myötätunnon, että hän on kuvannut hänet
aivan toisenlaiseksi kuin. miksi juutalainen esitetään keskiaikaisen
mielikuvituksen luomassa kantatarinassa ja miksi näytelmän
tapahtumain kulun yleinen suunta hänet vaatisi. Hän on onneton
ihminen, jota vastaan rikotaan enemmän kuin hän itse rikkoo ja joka

You might also like