Cho 1975
Cho 1975
Y. M. Cho
Related Articles
Angular momentum at null infinity in five dimensions
J. Math. Phys. 52, 032501 (2011)
The Weitzenböck connection and time reparameterization in nonholonomic mechanics
J. Math. Phys. 52, 012901 (2011)
Singular sources in gravity and homotopy in the space of connections
J. Math. Phys. 50, 122505 (2009)
Scalar field theory in κ-Minkowski spacetime from twist
J. Math. Phys. 50, 102304 (2009)
Collinear central configurations in the n-body problem with general homogeneous potential
J. Math. Phys. 50, 102901 (2009)
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
Higher-dimensional unifications of gravitation and gauge
theories
Y. M. Cho"
The Enrico Fermi Institute and the Department of Physics, The University of Chicago, Chicago, Illinois 60637
(Received 9 April 1975)
2029 Journal of Mathematical PhysiCS, Vol. 16, No. 10, October 1975 Copyright © 1975 American Institute of Physics 2029
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
phism of the ~i's and ~i'S. The second set is nothing
more than the right invariance of the L's in the fiber
for every a E G, where the second equality comes from space. To understand the third set, one has only to
the left invariance of ~ /. realize that the horizontal component of the commutator
of any two horizontal lift vector fields [h, E] should be
The basis ~j (i=4 +1···4 +n) of G can be mapped
the same as the horizontal lift of [1'/, 1;] because
in a natural way into the fiber space of the bundle since
R 4J /(t) •• can also be viewed as a one-parameter group IT(H[~, EJ)=IT([~, E])= [1'/, 1;J.
of transformations acting on the bundle.
Therefore, a. . , L]
has to be vertical which is just the
We will write the vector fields in the bundle induced third equation (1). We would like to emphasize that the
by R 4J /(t).e(i=4 +1···4 +n) as ~r (i=5···4 +n), and as commutator of two horizontal vector fields has a
usual call them the fundamental vector fields. 4 These vertical component in general. F~v correspond to
fundamental vector fields are clearly tangent to the Yang-MillS fields, which we will soon justify.
fiber space and form a subspace, called the vertical
There is one more observation to make with the third
subspace, of the tangent space T/P) to P at each pEP.
commutation relations. Since the ~,,'s are defined to be
Now, we want to give one more crucial structure, a right invariant, the commutator -F~v~: should also be
connection r to our bundle. A connection r in P is a right invariant. This immediately dictates the fiber
choice of a tangent subspace called the horizontal sub- space dependence of F~v' In fact, from one of the Jacobi
space Hp at each PEP which satisfies 4 : identities
2030 J. Math. Phys., Vol. 16, No. 10, October 1975 Y.M. Cho 2030
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
The condition (a) is trivial by the definition. To :::::R a1xl ' (a· ~) + (a'(x)' a-1(x» . (a 0 ~.J
prove (b), it is enough to verify it in the following two
:::::Ra(xl (a·~) +a'(x) (L.-l(xl a
0 0 0 ~u),
cases since w is linear in t:
where u, when acting on a vector a ~" in T(G), is 0
satisfies D-2', one can always define a connection r =0 ~ iwi[Ralxl . (u 0 ~u) +a'(x) L a-1lxP ~"l
0 0
We can decompose w using the basis ~i of q: where we have used the fact that w is linear and have
i
taken into account the property (b) of D-2'.
W =w ~i> (3)
We now notice that w reverses the mapping a'. Indeed
where the wi are now real valued 1-forms, i. e., if I: is the left invariant vector field on G which is equal
covariant vectors in the usual sense, which satisfy to a ~" at a(x) E G, then the value of /; at the origin
0
wi(H)::::: w~ ~r::::: oL e EGis La-l1xl a ~" and a' . (L a-11xla ~,,) is the value of
0 0
One can express a connection form w on P by a family Now, by observing that in the adjoint representation
of forms each defined in an open subset U of M in the ~j:::::(~j)\=f/~
following way. 7
and
Let a cross section a(x) be a submanifold in P which
a(x)::::: etJ9jlxl,
is diffeomorphic to U with II(a(x» ::::: x and let a' ~<! be a
vector in the tangent space Ta1X)(P) to P at a(x) induced where eJ(x) (j::::: 4 + 1· .04 + n) are the canonical coordi-
by the mapping a:XE U-a(x)EW1(U). nates of a(x)E G, Eq. (5') can be written as
Given a connection r with the connection form w in t Ala')j-e-tJ9J(X)t etJ9jlX)A(ali+e-'J8J(xla (etj9jlX»
Si ,,- Si " " •
P, for every cross section u(x) of the bundle we can
q
define a -valued connection form Ala) on U by (5")
2031 J. Math. Phys .• Vol. 16, No. 10, October 1975 Y.M. Cho 2031
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
the bundle P compatible with the metrics g"v and gik [~IL' ~J= [a,., avJ=o.
on M and G if Y.b~:~e =g"v and Yabn·nb=gik' If, in
addition to the compatibility with the metrics g "v and Now from the definition D-3 we can attach gauge poten-
gik, we require the metric ')lab to make the horizontal
tials A (aa)'; to the family of trivial cross sections aa(x),
aEG:
and vertical subspaces orthogonal to each other, then
the metric Yab is unique. This metric is given by (8)
These trivial cross sections ("gauges") a.(x) form a
D-4: (1) ')Iab~~~e=g,,", family parametrized by the group element a, i. e" to
(2) Y.b~~~:b=O, every a E G there corresponds such a cross section.
In other words, for every a, aa(x) determines a gauge.
(3) Y.b~ra~:b =gik'
For simplicity we introduce the notation
Notice that this definition is independent of the choice B~ (x, a) =A (Oa)'; = W j(~,,(x, a»). (8')
of a basis. Such a metric has been previously con-
structed by Kaluza 1 for the Abelian (with g55 = 1) and As a consequence of Eq. (4), the horizontal lift €,.
by deWitt and Kerner 2 for the non-Abelian caseS can be written in this local direct product basiS as
respectively. (see Fig. 2)
We still did not prove that F!v can be viewed as From the group transformation properties of the con-
Yang-Mills fields. To show this and to calculate the nection form w specified by property (b) of D-2', we
curvature of the bundle space, we now introduce a basis have
for the bundle. aiB~(x,a)=atB~=-f:jB~(x,a). (10)
Remember that from (c) of D-1, the bundle is locally From Eqs. (9) and (10) we can reconfirm the commu-
trivial and we can always parametrize locally P E WI(U) tation relations (1):
by (II(p), rp(p») =(x,a) E UXG.
In this parametrization WI( U) is diffeomorphiC to the * * ] = [-~i'~j
[ ~i'~j k - k *
- J=fij~k=fjj~k'
direct product space UXG, and we will write p=(x,a) [~t, ~,,] = [~I> L - ~j B~] = [~it D" J= 0,
whenever convenient.
[~'"
U= [~ll - ~iB~, ~v tB~l = [D", DJ -
Given any bases of U and G, one can always introduce
a basis in WI( U), the local direct product baSis of the = - F:v ~k = - F!v~:, (I')
two bases of U and G. In particular, one can take the
local direct product basis E:" (IJ. = 1 .. ·4) and t where now F:v are determined in terms of the potentials
(i = 5···4 +n) of the coordinate baSis ~IJ. (IJ. = 1· ··4) of B~ by
U and the left invarient basis ~ i (i = 4 + 1 .. ·4 + n) of G F~" == iJ"B~ - iJvB~ +fi~B~B~,
for our basis of n -I( U).
the usual Yang-Mills form.
We clearly have "(I = ~i (i = 5· . ·4 +n). But notice that
~,,"* L (the horizontal lift). This is so because for every One can also derive from Eq. (10) the gauge covari-
a E G the ~u (IJ. = 1···4) form a basis of the tangent ance of F!v:
space of the trivial cross section aa(x) defined as the set (2')
of pOints p = (x, a) with fixed a, which is by itself a
smooth submanifold. Consequently, tIL (1J.=1 .. ·4) Thus we have proved our claim that F~v do corre-
should form a closed Lie algebra within themselves. spond to Yang-Mills fields.
On the other hand L
in general do not by themselves From one of the Jacobi identies
form a closed Lie algebra: The commutator [L, t], as
we have seen in Eq. (1), does acquire a vertical
component. we have
A connection which gives vanishing vertical compo- DpF!,. +DjJ,Fv~ +DvFpk" =0 (11)
nent to all [g", tJ
is called the flat connection, in which
case F:v vanish identically although the cross section
dependent vector potentials A (a)'; need not vanish.
2032 J. Math. Phys., Vol. 16, No. 10, October 1975 Y.M. Cho 2032
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
for every /J., I), p, and k, which are the well-known where r;k and r~v are the torsion free Christoffel
gauge Bianchi identies. In particular, they guarantee coefficients of G and M with the metrics glk and g"v,
the absence of non-Abelian magnetic monopoles. and
Since the bundle space P has been organized into a R"k=Rk" =tgkIB! +tgklgjjg''''gMB~F~BF;6
(4 +n)-dimensional Riemann space, we can calculate + igkl g"'BVocF!s,
the curvature of the bundle P. In the local direct prod-
uct basis the metric defined by D -4 can be written R"v=R"v +tgikB~B~ +tgiJgklg"'YgBO B!B~F~BF:o
explicitly as _tg"BgikF~"'~B +tgik~B(B~ V,,.F~8 +B~V",F:B)'
Yab= (g "+gk~j~
B
B~ ) (12) (16)
gik " gik
where Rjk = tg ik and R"v are the Ricci tensors of G and
and M, and Voc is the gauge and generally covariant
derivative, e. g.,
r~jL = rL=Hfj~B~ +gklg"~B~F~jL)' Clearly these ~. form a basis for the tangent spaces
of !r l( U) and their commutation relations are
~k=r:.. =ig"8gkzF~~, (15)
r~,,=i(a .. Bi +o"B~)+ig"~gikB!(F~"B~ +F~"B:) [~j, ~Jl=fi~ ~k'
-r~"B~, [~j,~.. J=O, (20)
r~"=r~" +tg"'B gik(B~F:,B +B~F';,B)' [~'" U=O.
2033 J. Math. Phys., Vol. 16, No. 10, October 1975 Y.M.Cho 2033
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
It is straightforward to retrace every previous step We now want to comment on the cosmological term 12
in this basis and to obtain the same result. We will not RG..J-g" in the action Eq. (19). The only dimensional
duplicate the arguments here. constants in the theory are the universal gravitational
It is worthwhile mentioning that there exists yet
constant G and the typical length in group space (i. e. ,
another basis where the calculations become particular- essentially the magnitude of glj)' Using units 1i=c=l,
ly simple. Notice that L
(f.1. = 1 .. ·4) and (/ =' ~;
one can therefore consider the Planck length Lp = rc
(i = 5 ... 4 + n) can also be used as a basis for the bundle
and the length in group space L G. If LGis left arbitrary,
then so is the cosmological term. It is not clear what
with the commutation relations (1). We will call this
dynamical considerations are to determine it. If LG
basis the horizontal lift basis.
were to be of the order of the Planck length, so that G
In this horizontal lift basis Yab can be written as be the only dimensional parameter, one would obtain
a by far (-10 '20 ) too large cosmological term. In that
Yab=(~\ case, one might try a metric for G which gives a
\ 0 'gik) vanishing RG and remove the cosmological term al-
and (21) together. Otherwise, one can add a countercosmological
term "by hand. "
We also emphasize at this point that the universal
coupling constant g of the Yang-Mills field is also
arbitrary in this theory. For Simplicity we have set
Again assuming the torsion-free Christoffel coeffi- g = 1 in this paper. The theory for arbitrary g is then
cients for simplicity, we find in this horizontal lift basis obtained by systematically replacing A~(J)I and B~ by
A
gA~)i and gB~.
r j"=2fj~=rjk'
ill I
r"k=r~,,=o, fields. 14
(22) Beyond these technicalities, however, we would like
to emphasize more the general structure of the theory:
the combined geometry of the space-time with the in-
ternal symmetry space. In this geometry, once an
internal symmetry is given, one is led to have gauge
and
fields without referring to the existence of any "matter"
Rlk = Ri~ + tg lj gkl g" 6g YO F~y F1o, fields.
R"k=R k " =~gklg"'6'VOlF~B' (23) Of course, one can always introduce matter fields in
our geometry by hand by introducing another bundle
-R uv-2gik5
R j.l.V-
A !. n<>BFI Fk
j"HI! vB, space: the fiber bundle associated to the principal fiber
and bundle. 15 It would be very interesting if one could obtain
the Fermi fields as a part of the gauge fields of some
(24)
supersymmetry.16
Again we get the same result we had before. Notice It is well known that the usual theory of gravitation
that, although the components of tensor quantities do itself can be viewed as a theory of connections of the
depend upon the choice of a basis, the curvature R bundle P with the space-time as the base manifold and
being a scalar does not depend upon the choice of a G = 0(1,3) as the structural group. 17 In this picture the
basis. Christoffel coefficients r:v
appear as the cross section
dependent connection forms on Uwith G=0(1,3). This
In this horizontal lift basis ~" are not the basis vec- explains the well-known nontensorial transformation
tors of the tangent space to any cross section a(x), be- properties of r:v' One is led to Einstein's theory from
cause they do not commute under commutation. the torsion-free connection, Cartan's theory 'S from the
Potentials A~(Jli = wi(a' ~..) can be introduced as before, torsional connection. 19 Utiyama's and Kibble's work 20
since they are defined independent of choice of a basis can be argued more elegantly and conSistently in this
in the bundle space. The horizontal lift basis is very picture.
convenient for calculations.
ACKNOWLEDGMENTS
9. DISCUSSION It is a great pleasure to thank Professor P. G. O.
Freund for suggesting this problem and for the continu-
We thus have combined gravitation with gauge theory ous advice which was essential in completing this paper.
in the context of a unified geometric theory in the bundle I am grateful to Professor Y. Nambu for careful reading
space with the metric D-4. of the manuscript and Professor R. Lashof in the
In the process of the argumant we have assumed the Department of Mathematics of the University of Chicago
torsion-free Christoffel connection for our bundle P and for helpful discussions. I am also indebted to R. M.
the space-time independent metric Yab~ja~:b =glk for Wald, J. Friedman, and P. S. Jang for many discus-
the fiber space. In general one can relax these assump- sions. Finally, I would like to thank S.O. Ahn for
tions and obtain different variants of the theory. encouragement.
2034 J. Math. Phys., Vol. 16, No. 10, October 1975 Y.M. Cho 2034
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions
*Work supported in part by the National Science Foundation, SOf course, g ..~ and gu are in terms of the previously chosen
Contract No. MPS75-08833. coordinate basis of M and the left invariant basis of G.
IT. Kaluza, Sitzber. Preuss. Akad. Wiss., 966 (1921). lOsee, for example, C. W. Misner, K. S. Thorne, and J.A.
20. Klein,'. Z.Physik 37, 895 (1926). Wheeler, Ref. 6.
3B. DeWitt, Dynamical Theories of GrouPs and Fields (Gordon HOur argument in this section is similar to that of R. Kerner,
and Breach, New York, London, Paris, 1965), p. 139; R. R,ef. 3. However, our results [Eq. (14), (15), and (16)] differ
Kerner, Ann. !nst. H. Poincare 9, 143 (1968); A. Trautman, from his.
Rep. Math. Phys. 1, 29 (1970). 12 As for cosmological models with a cosmological constant,
(For general mathematical background we refer to S. Stern- !me, for example, H. Bondi, Cosmology (Cambridge U. P.,
berg, Lectures on Differential Geometry (Prentice-Hall, Cambridge, 1960), 2nd ed., Chap. IX. See also S. Weinberg,
Englewood Cliff, N.J., 1964); S. Kobayashi and K. Nomizu, Gravitation and Cosmology (Wiley, New York, 1972), pp.
Foundations of Differential Geometry (Interscience, New 1!>5, 613-16.
York, 1963); K. Nomizu, Lie GrouPs and Differential Geom- 13p. Jordan, Z. Phys. 157, 112 (1959); Y.R. Thirry, C. R.
etry, Publ. Math. Soc. Japan #2, (Math. Soc. Japan, Tokyo, Acad. Sci. Paris 226, 216 (1948); see also, C. Brans and
1956); R. L. Bishop and R. J. Crittenden, Geometry of Mani- R.H. Dicke, Phys. Rev. 124, 925 (1961).
folds (Academic, New York and London, 1965). 14y. M. Cho and P. G. O. Freund, to be published.
'For Simplicity we will use in this paper the same notation for 15s.~e, for example, A. Trautman, Ref. 3; K. Nomizu, Ref.
a mapping of a manifold to another and for the corresponding 4. or S. Kobayashi and K. Nomizu, Ref. 4.
mapping of a vector (I. e., an element of the tangent space) I GFor a similar approach but in somewhat different context,
to another induced by the mapping of the manifolds. For ex- Sl3e D. V. Volkov and V. P. Akulov, Zh. Eksp. Teor. Fiz.
ample, if L. hits an element of the group G, it should be in- Fis'ma Red. 16, 621 (1972) [JETP Lett. 16, 438 (1972)];
terpreted as a mapping of G into itself, and if L. hits a vec- D. V. Volkov and V. A. Soroka, ibid. 18, 529 (1973) [ibid. 18,
tor, it should be interpreted as a mapping of the vector to 312 (1973)J.
another vector induced by the mapping L. of G. I7For a precise mathematical justification, see S. Kobayashi
6For a noncoordinate basis, see for example, C. W. Misner, and K. Nomizu, R~f. 4, p. 158.
K.S. Thorne, and J.A. Wheeler, Gravitation (Freeman, tBE. Cartan, Ann. Ecole Normale 40, 325 (1923); 41, 1 (1924).
San FranciSCO, 1973). 1'1:. W. B. Kibble, J. Math. Phys. 2, 212 (1960).
7Se e , for example, S. Kobayashi and K. Nomizu, Ref. 4, p. 2DJl. Utiyama, Phys. Rev. 101, 1592 (1956); T. W.B. Kibble,
66. Ref. 19.
BA. Trautman, Ref. 3.
2035 J. Math. Phys., Vol. 16, No. 10, October 1975 Y.M. Cho 2035
Downloaded 12 Dec 2012 to 139.184.30.132. Redistribution subject to AIP license or copyright; see http://jmp.aip.org/about/rights_and_permissions