Downloaded from https://academic.oup.

com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
866

Progress of Theoretical Physics, Vol. 64, No. 3, September 1980

Gravity from Poincare Gauge Theory

of the Fundamental Particles. I
--General Formulation--

Kenji HAYASHI and Takeshi SHIRAFUJI

Institute of Physics, University of Tokyo, Komaba, Tohyo 153

Physics Department, Saitama University, Saitama 338

(Received March 27, 1980)

We study Poincare gauge theory with linear and quadratic Lagrangians, where there
are the translation and Lorentz gauge fields whose sources are the energy-momentum tensor
and the spin tensor, respectively. Ten parameters, a, a, (3, r, a,, ···, a,, are contained in the
gravity action. Most general field equations covariant under the Poincare gauge group are
derived, and also their alternative forms are obtained. For geometrical analysis the present
space-time is that of Riemann-Cartan, endowed with curvature and torsion. General Relativity
is recovered when a spin source vanishes and the parameters satisfy the conditions, 3a, + 2a,
=O=as + 12a,.

§ l. Introduction

Gravity is usually treated by Einstein's gravity theory called General Relati-

vity, which explains all known experiments in gravity physics. It is based on the
geometry of the Riemann space-time manifold which is characterized by the curva-
ture tensor alone and the absence of the torsion tensor. Recently, however, it is
shown that New General Relativity is nearly the same as General Relativity on
the macroscopic scale and agrees with all gravity experiments :11 New General
Relativity is based on the W eitzen bock space-time manifold which is characterized
by the torsion tensor alone and the absence of the curvature tensor. Thus one
is not obliged to stay within the conventional theory.
Then what is the principle by which we can go beyond General Relativity
and even New General Relativity? It is well known that General Relativity is
plagued by nonrenormalizability,"1 which prevents its microscopic version from being
a correct quantum field theory. So microscopic gravity theory may have to be a
renormalizable field theory.
Everything is made of the fundamental particles of intrinsic spin 1/2, which
are called quarks and leptons, whereas the fields of spin 1, such as the electro-
magnetic field, W± mesons, Z meson, gl uons, etc., are regarded as the gauge par-
ticles. This gauge picture is provided by particle physics with the success of the
W einberg-Salam theory for the electroweak interactions 31 and with the invention
of quantum chromodynamics of the strong interaction. We shall also use the
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity from Poincare Gauge Theory of the Fundamental Particles. I 867

gauge principle in describing gravity at the fundamental level of the microscopic

scale. The gauge group this time is the Poincare gauge group, which 1s the
extension of the Poincare group. In this case the gravitational variable is the
tetrad field, which is more fundamental than the metric tensor.
Here are several problems in applying the Poincare gauge approach to gravity.
In this approach there are two sets of the gauge fields. One is associated with
the translation gauge group and called the translation gauge field, and the other
is connected with the internal Lorentz gauge group and called the Lorentz gauge
field. As for the internal group, Yang and Mills 4' were the first to study the
SU(2) isospin gauge group. Then Utiyama investigated the possibility of treating
the Poincare group, 5' which was later inherited by Kibble. 6' However, Poincare
gauge theory is much more complicated than that of the internal group: Accord-
ing to Yang-Mills gauge approach there must be a gauge-invariant Lagrangian
which is quadratic in the gauge field strength. The well-known example is the
electromagnetic Lagrangian which is quadratic in the electromagnetic field strength.
But Einstein's Lagrangian of General Relativity is linear in the Lorentz gauge
field strength. This is the first difficulty in treating gravity by the Yang-Mills
gauge approach. In 1968 Hayashi 7' tried to resolve this dilemma by extending
to Lagrangians quadratic in the Lorentz gauge field strength, and Hayashi and
Bregman8' further pursued this problem. There is a huge amount of the literature
since then. 9'
As for quadratic Lagrangians, there are two methods in handling the Lorentz
gauge field whose source is an intrinsic spin tensor. One is conventional in which
the Lorentz gauge field A is decomposed into the Ricci rotation coefficients .d, which
is given by first derivatives of the tetrad field, and contorsion K,

A=.d+K. (1·1)

When only the linear Lagrangian is used, this conventional method yields Einstein-
like equations, 6'' 1 ' but the contorsion field does not propagate in vacuum but is
frozen at the place of matter. Quantum field theory for such a frozen field is
almost impossible and not appealing. When quadratic Lagrangians come in, field
equations, which are derived by taking the translation and Lorentz gauge fields
as independent variables, contain third derivatives of the translation gauge field
when (1·1) is inserted in field equations.7l.sl Thus we would be led to higher-
derivative gravity theory. However, in fact field equations in the weak field ap-
proximation will be of second order differential equations and higher derivatives
will be shifted to source terms, as will be shown in this series. Then the contor-
sion field K is capable of propagating in vacuum when its kinetic energy is sup-
plied by quadratic Lagrangians.
The other is unconventional in which the Lorentz gauge field has its own
degree of freedom, not decomposable into any other quantities. This case is very
much alike the Yang-Mills gauge approach and makes possible for the contorsion
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
868 K. Hayashi and T. Shirafuji

field to propagate 1n vacuum.

We make a few remarks on quadratic Lagrangians. The present approach
will differ from the "Quadratic Lagrangian Method in General Relativity",lOl whose
quantum theory was investigated by Stelle.11l Neville studied quadratic Lagrangi-
ans of Poincare gauge theory, however, neglecting the translation gauge field
altogether.12l This is obviously incomplete and differs from our results shown in
this series.
In this series we shall explore the problem of what is essential in gravity
and various problems concerning gravity of the fundamental particles in order;
here we deal with general formulation of the Poincare gauge invariant gravity
theory. Various limits of the theory and the weak field approximation on the
field equations will be discussed in the forthcoming papers.
This paper is arranged as follows: In § 2 our arguments are based on the
fundamental particles of spin 1/2 by introducing the invariance under the Poincare
gauge group. In § 3 we contemplate two sets of the gauge field strengths, called
the translation and Lorentz gauge field strengths, and then build up three pieces
quadratic in the translation gauge field strength, six pieces quadratic in the Lorentz
gauge field strength, and finally one invariant linear in the Lorentz gauge field
strength. (There is no other invariant linear and quadratic in the gauge field
strengths.) Then we derive most general gravitational field equations, which con-
sist of two sets of field equations, associated with two variations of the translation
and Lorentz gauge fields. In § 4 space-time manifold is analyzed, and the most
general one is the Riemann-Cartan space-time with the curvature tensor and the
torsion tensor. In § 5 we follow the conventional method mentioned above and
supply alternative forms of most general gravitational field equations, thus compar-
ing with other approaches. Final section will be devoted to conclusion.

§ 2. Fundamental particles and Poincare gauge group

We describe the fundamental particles of intrinsic spin 1/2, which are called
quarks and leptons; color and flavor indices of quarks are here suppressed. The
action integral in special relativity is given by

(2 ·1a)

with

where q denotes collectively. the quarks and leptons (q=q*{ 0) . Here the electro-
magnetic field, 1V± mesons, Z meson, and gl uons can easily be incorporated in
by the Yang-Mills gauge procedure: See, for example, the W einberg-Salam
theory. 3l So they are skipped here.
We now demand that the action should satisfy Poincare gauge invariance.
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
s. I 869
Gravity from Poincar e Gauge Theory of the Fundam ental Particle

y coordin ate
To meet this requirem ent, two steps are needed. First, for arbitrar
131

spinor field q (x) should behave

transfor mations , that is, extende d translat ions, the
like scalar,
q' (x') =q(x) (2·2)

for x' = f(x), where f is an arbitrar y differen tiable function . Then the derivati ve

of the spinor field should change as follows :

a,q (x) ~ (r)/ + c/ (x)) 8/,q (x), (2. 3)

coordin ate patch.

where Greek index runs from 0 to 3, referrin g to arbitrar y
transfor ms inhomo-
Here ck~ (x) is called the transla tion gauge field, because it
geneous ly under the extende d translat ion,
(2· 4)

for the invarian t form of the new derivati ve,

(2. 5)

ate patch by ap-

In fact it can be transfor med away for an infinites imal coordin
propriat e transfor mation.
changes ac-
Second, 61 for the coordin ate x to be invarian t, the spinor field
cording to the spin transfor mation,

q' (x) =L (x) q (x), (2· 6)

followin g pro-
where L is a 4 X 4 matrix dependi ng on the coordin ate x. The
ry. The previou s derivati ve
cedure is quite typical for the case of internal symmet
(2 · 3) should be modified as
(2 ·7)

group,
Here Sij is the infinites imal generat or of the internal Lorentz
(2· 8)

and Aii~ (x) =- Aji 1, (x) is called the Lorentz gauge field, which transfor ms under
(2 · 6) as
(2. 9)

with Ylmn the Minkow ski metric tensor, ( YJmn) = cliag ( -1, +1, +1, +1), for the

vector form of the modified derivati ve,

(ok~ + c' k~ (x)) D' ~q' (x) = L (x) Akj (x) (o/ + c/ (x)) D~q (x) (2 ·10)

with A7/ (x) a Lorentz transfor mation dependi ng on x.

 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
870 K. Hayashi and T. Shirafuji

Now it 1s very convenient to introduce e/ (x) as

(2 ·11)
where ekll is called the tetrad field which is not the gauge field, because it is the
Lorentz vector with respect to Latin index and the world vector with respect
to Greek index,

(2 ·12)

Thus the mverse field ek 11 is defined by

(2 ·13)
through which the gauge field a\ is introduced by

(2·14)

Then we are able to define the determinant m terms of this mverse field,

(2·15)

In summary the modified derivative now takes the form,

(2 ·16)

where the translation gauge field is contained in e/.

To investigate the gauge field structure in the new derivative (2 ·16), the
following points are essential: The Poincare group is the semi-direct product of
the translation group T and the Lorentz group L, T&JL, where L consists of
the orbital part Lorb and of the intrinsic spin part Lint· The gauge fields
Partially reflect this group structure, and indeed they are incorporated as if the
group structure were the direct product of the translation and internal Lorentz
groups, T&JLint·
Now we consider an infinitesimal form of the translation and internal Lorentz
transformations. First, for arbitrary change of the coordinate x,

(2 ·17)

where ell (x) is an arbitrary infinitesimal, differentiable function, the translation

gauge field transforms as (2·4) dictates; here c'k 11 (x') =ck''(x) +ockli(x) and
a'\(x') =a\(x) +oa\(x) with
oc/'(x) =8~vck"(x) +e;~ok",
(2 ·18)
oakp(x) = -C:~,ua\(x) -e~jJ\.

Hence the tetrad field changes as follows:

oek"(x) =C:~,ek"(x),

oek"(x) = -e~"ek,(x). } (2 ·19)

 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity from Poincare Gauge Theory of the Fundamental Particles. I 871

Here the partial derivative with respect to the coordinate is abbreviated by the
comma. This time the spinor field does not change at all,

0q (X) = 0 = Oq (X) . (2. 20)

Second, the spm transformation of the spinor field, denoted by (2 · 6), now takes
the form

oq(x) = (i/2)wij(x)Sijq (x), (2·21)

and the modified derivative changes as follows:

where wij (x) = - w j i (x) is a set of six arbitrary infinitesimal, differentiable func-
tions. Then the Lorentz gauge field changes as

(2. 23)
and the tetrad field transforms as

(2. 24)
Finally, the action integral for the fundamental particles of intrinsic spin 1/2,
which satisfies Poincare gauge invariance, reads

(2 -25)

where

(2. 26)
with

(2. 27)
The modified Dirac equation is derived from this action integral as follows:

[ir#(Dj!+ (1/2)vj!) -m]q=O, (2. 28)

q [i (D#+ (1/2) vj!) rj! + m] = 0 (2. 29)
or more concisely,

[irpaj!+ (i/2e) (aj!erj!)- (3/2)A,,r 5rj!-m]q=O, (2. 30)

q[i8j!rj!+ (i/2e) (a,,er!!) + (3/2) Aj!trj!+m] =0, (2·31)
where

(2· 32)
(2. 33)
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
R72 K. Hayashi and T. Shirafuji

(2. 34)

Here ckijm is the usual totally antisymmetric tensor m the Minkovvski space-time
with c 0123 = +1 = - co123·

§ 3. Most general gravitational field equations

We shall derive the most general gravitational action which is invariant under
the Poincare gauge group. It is given by

(3 ·1)

where La denotes a gravitational Lagrangian density and Lx stands for the matter
Lagrangian density of the fundamental particles of spin 1/2, given by (2 · 26).
There are two classes of the gauge field strengths; one is the translation gauge
field strength, defined through the derivative of the translation gauge field,
(3. 2)

where Cuk and Au, are of the form

cij k = e / e k' (IJ l'e ;, -f) ,e ;,,) ' (3. 3)

Aijk = ek" Aif".

The other is the Lorentz gauge field strength which 1s defined through the
derivative of the Lorentz gauge field,

(3. 4)

As yet we do not know the behavior of the translation gauge field and
the Lorentz gauge field Aijw So the translation and Lorentz gauge fields are
treated independently of each other until field equations for them are to be solved.
Notice again that e/ is not the gauge field, because it transforms like vectors,
but we shall use it in place of ckP since it is very convenient to write various
formulae.
It is necessary that the gravitational Lagrangian density is of a linear term
in the Lorentz gauge field strength and of quadratic terms in the translation
and Lorentz gauge field strengths. To derive them all, the irreducible decom-
position is best fitted for this purpose. First, let us apply the Young table
method 14J to the translation gauge field strength. The first is the tensor, obtained
from the Young table [21] minus all possible traces,

(3. 5)

The second is the vector, obtained from trace,

(3. 6)
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity from Poincare Gauge Theory of the Fundamental Particles. I 873

and the last is the axial vector, constructed from the Young table [111],

(3 ·7)

Using the symbol (jh }2) for the irreducible non-unitary representation of the proper
orthochronous Lorentz group L!, we find that the tensor transforms according
to (3/2, 1/2) EB (1/2, 3/2) of 16 dimensions, the vector according to (1/2, 1/2),
and the axial vector according to (1/2, 1/2). Furthermore, the tensor satisfies
the cyclic identity,

(3. 8)

The original translation gauge field strength is represented from these irreducible
building blocks as follows:

(3. 9)

Next, we carry out the irreducible decomposition of the Lorentz gauge field
strength by means of the Young table method:
(i) The tensor (in fact, the scalar) corresponding to the Young table [1111],

(3 ·10)

(ii) The tensor corresponding to the Young table [22] mmus all possible
traces,

(3 ·11)

where the tensor 1Viimn is the same tensor as Fijmn except that it 1s traceless in
all subscripts, its being called the Weyl tensor of Fijmn,

(3 ·12)

(iii) The tensor corresponding to the Young table [211 J mmus all possible
traces,

(3 ·13)

but 'half' of c~jmn is indeed irreducible (upon investigation)'

(3 ·14)

(iv) The tensor corresponding to the Young table [2] mlnUS trace,

Iij = (1/2) (Fij + Fji) - (1/ 4) 'llijF. (3 ·15)

(v) The tensor corresponding to the Young table [11],

E;j = (1/2) (F;j- Fji). (3 ·16)

 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
874 K. Hayashi and T. Shirafuji

(vi) The scalar,

(3 ·17)

Here we have used the abbreviation

(3 ·18)

With respect to the proper orthochronous Lorentz group Ll, the tensor F;Jmn
of 36 components splits into the following irreducible tensors; the first is (2, 0)
EB (0, 2) for the tensor BiJmn of 10 components, the second is (1, 1) for the tensor
C,1mn of 9 components, the third is (0, 0) for the scalar A, 1mn of one component,
the fourth is (1, 1) for the tensor Iii of 9 components, the fifth is (1, 0) EB (0, 1)
for the tensor E,1 of 6 components, and the last is (0, 0) for the scalar F of
one component. Finally, the tensor FiJmn is represented in terms of the irredu-
cible building blocks as follows:

F,Jmn = W;1mn + (1/2) {rhm (E1,. + ~,.) +'l/1,. (E;m + I,m)

-'l/}m (E;,. + 1;,.) -r;;,.(E}m + ~m)}- (1/12) ('l/;m'l/jn-1'/in'l/}m) F, (3·19)
where
(3· 20)

Now we demand that the gravitational Lagrangian density be linear in the

Lorentz gauge field strength and quadratic in the translation and Lorentz gauge
field strengths, and that it be invariant under the operation of space inversion.
The required Lagrangian consists of three parts,

(3. 21)

where F is the linear invariant with an arbitrary parameter a, and Lr and LF

are quadratic in the translation gauge field strength and the Lorentz gauge field
strength, respectively,

(3 ·22)

(3. 23)

where a, /3, r
and a~> · · ·, a 6 are arbitrary parameters. In the past the last term
LF has not been considered. 6J,sJ In this paper we shall take account of all possible
quadratic terms in the translation and Lorentz gauge field strengths.
Now we shall derive the most general gravitational field equations:*) As for

*l The most general gravitational field equations with respect to the Pauli metric were first
derived by Hayashi in Ref. 7), but they are rederived here with respect to the Minkowski metric
(71, 1) =diag (-1, +1, +1, +1).
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity from Poincare Gauge Theory of the Fundamental Particles. I 875

the translation gauge field, we get the field equation*)

(3. 24)

where

(3. 25)
(3. 26)

(3. 27)

(3. 28)

Here H;j is symmetric upon introducing the irreducible decomposition of the trans-
lation gauge field. The energy-moment um tensor is as usual defined by

(3. 29)

Next, for the Lorentz gauge field we obtain the field equation

(3. 30)
where

H;jk=- (a+ (2a/3)) (tk;j-tkj;)- (/3- (2a/3))

X (rh;Vj-r;kjv;)- (2/3) Cr+ (3a/2)) ckijma"' (3. 32)

and the spin tensor 1s defined by

(3. 33)

§ 4. Space-time manifold with curvature and torsion

We have so far not considered the geometry of space-time manifold but only
paid attention mainly to the Poincare gauge theory with the translation and Lo-
rentz gauge fields and corresponding strengths. Here we shall discuss the geome-
trical structure of space-time manifold, thus giving a connection between the Poin-
care gauge approach and the purely geometrical approach.
The Riemann-Carta n space-time U1 is a paracompact, Hausdorff, connected c=
four-dimensiona l manifold with a locally Lorentzian metric tz and a linear affine
connection r which obeys the metric condition,

*l vVe denote symmetrization and antisymmetrizatio n of tensor indices by a round

bracket
) and a square bracket [ ], respectively: For example, J[kmJCnn= (1/4) (Jkmnj-Jmkn;-Jkm;n +Jmk;n).
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
876 K. Hayashi and T. Shirafuji

(4·1)

From this equation we get

fJ.V-
J. } + x1.
r" - {f1V "I'' '
(4·2)

where the first term denotes the Levi-Civit a connection ,

(4·3)

and the second stands for the con torsion tensor,

(4·4)

with the torsion tensor

(4 ·5)

The curvature tensor 1s defined in terms of the affine connection ,

(4·6)

We are concerned with the fundament al particles of spin 1/2, which are the
sum of the two-compo nent spinors transformi ng under the two-valued representa tion
of the connected, proper, orthochron ous Lorentz group. So we must demand that
the Riemann-C artan space-time should permit the sPinor structure, ) which means
15

that the spinor can be defined all over the manifold, or mathemati cally speaking,
the tetrad field is smoothly spanned all over the manifold. The metric tensor
is expressed in terms of the tetrad field as
(4· 7)

The spin or structure thus yields the spin or connection for the two-compo nent spin or,
from which the Lorentz connectio n is obtained for the sum of the two-compo nent
spinor. Therefore , it is this Lorentz connection that the Poincare gauge approach
has introduced , by demanding that the theory remains invariant under the opera-
tion of the internal Lorentz group.
The Lorentz connection A, derived from the spinor connection , has the follow-
ing expression : 16 l
(4·8)

with .d the Ricci rotation coefficient s,

(4·9)

where cijk is defined in (3. 3); it is constructe d from the derivatives of the tetrad
field. The second term K is called the con torsion ( 4 · 4). Greek or Latin index
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity from Poincare Gauge Theory of the Fundamental Particles. I 877

is converted to Latin or Greek index by the help of the tetrad field, for example,

(4·10)

It follows from the spinor analysis on the Riemann-Cartan space-time that

the affine connections with respect to the coordinate basis, r;v,
and with respect to
the Lorentz basis, A-il~· are related with each other by 16J

(4·11)

Therefore, the Lorentz gauge field A is directly connected with -the affine con-
nection with respect to the coordinate basis, r,

(4·12)

Finally, the connection between the Poincare gauge approach and the geome-
trical approach becomes very clear. The translation gauge field strength Til<
of (3 · 2) is nothing but the torsion tensor T~vl of ( 4 · 5) ,
(4 ·13)

or

(4·14)

and the Lorentz gauge field strength FiJmn of (3 · 4) IS the curvature tensor R~"""
of (4· 6),

(4·15)
or

(4·16)

§ 5. Alternative form of the gravitational field equations

Following the conventional method mentioned in the Introduction, we shall

rewrite the gravitational field equations, denoted by (3 · 24) and (3 · 30), by using
the following expression for the Lorentz gauge field A;1 ~:

(5 ·1)

(See Eq. (4·8).) Using (5·1) in (3·4), we find that the Lorentz gauge field
strength Fili•v is decomposed into two parts,

(5·2)
with

(5·3)
(5. 4)
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
878 K. Hayashi and T. Shirafuji

where f7"K;1• is the covariant derivative with respect to the Ricci rotation coeffi-
cients when the index is Latin, and with respect to the Christoffel symbol when
the index is Greek. As is well known, R;1"" ( { } ) is related to the Riemann-
Christoffel curvature tensor by

Using (5 · 2) in the formulae (3 ·10) '""'-/ (3 ·18), we find that each irreducible part
of FiJmn is split into two parts: In particular, JiJmn of (3 · 28) can be expressed
as

Jijmn = Jijmn ( { } ) + Jijmn (K) , (5·6)

where J;Jmn ( { } ) and JiJmn (K) are formed of the irreducible parts of RiJmn ( { } )
and FiJmn (K), respectively: In particular, the antisymmetric part of JiJmn ( { } ) is
given by

+ 7J jnRim ( { } ) -7J;nRjm ( { } ) -7] jmRin ( { }) ]

1 (5·7)
+4(a2-as+4aa) (7J;m71Jn-7lin71Jm)R({ }),

where R;1 ( { } ) and R ( { } ) are the Ricci tensor and the Riemann-Christoffel scalar
curvature, respectively,

(5·8)

The gravitational Lagrangian density La of (3 · 21) with (3 · 22) and (3 · 23) can
then be rewritten as

La= aR ( { } ) + ~a2Rijmn ( { } ) RiJmn ( { } ) - __!_ (3a2- 2ds) RiJ ( { } ) RiJ ( { } )

4 2

+__!_(a 2-a 5 +4a 6)R({ }) 2+L'T+L'F+2a8"(ev")/e, (5·9)

4
where

(5 ·10)

and L' F is obtained from LF by subtracting all the quadratic terms of the Riemann-
Christoffel curvature tensor,
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity from Poincare Gauge Theory of the Fundame ntal Particles. I 879

L'F=LF -[! a2R;jmn({ })Rijmn({ })

_l_(3a2- 2a5)Rii( { })Rii({ })+l_(a2 -as+4a 6)R({ }) 2] .

2 4
(5·11)
We are now ready to rewrite the gravitation al field equations. Using the
above formulae in (3 · 24), we get the alternative form of the field equation for
the translation gauge field,

2aGij({ }) + (3a2+2a5) [Rim({ })Rjm({ }) +Rmn({ }) (Rimjn({ })

- ~ 'lJiiRmn ( { }) ) J- (2a2 + a5- 4a6) R ( { }) ( Rii ( { } ) - ! 'lJiiR ( { }) )

+ 2R~rr:~i ( { } ) J[km][nj] (K) + 2F~rr:~i (K) (J[km][nj] ( { } ) + J[km][njJ (K))

(5 ·12)
where Gii ( { } ) is the Einstein tensor,

(5 ·13)

and we have used the identity/ 7l

R~rr:~i ( { }) Rkmnj ( { } )

=2Rimjn( { })Rmn({ }) +2Rim({ })Rjm({ }) -R({ })Rij({ })

+! 'lJij[Rklmn({ })Rklmn({ }) -4Rmn({ })Rmn({ }) +R({ }) 2].

(5 ·14)
Here F' iik and H' ii are given by

(5·15)

H 'ij == T mni ptmn

• · j --
1T ;"mn
pt·mn
i · (5 ·16)
2

Comparing (5 ·15) with (3 · 32), we find that

 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
880 K. Hayashi and T. Shirafuji

F'iJk=! (HiJk-H ikJ-HJkJ , (5·17a)

or conversely

IIiJk = F' ijlc- F' Jik . (5 ·17b)

From (3 · 30) we obtain the field equation for the Lorentz gauge field,

(3a2 + 2a5) PciG j]k ( { } ) + (az + a5 + 4aa) 'hciO nG ( { })

-2 (Dm- pm) J[ij][km] ( { } ) - 2Dm J[ij][km] (K)

+ (T;,mn- 2iJ k mvn) (J[ij][mn] ( { } ) + J[ij][mn] (K)) - Hijk

1 (5 ·18)
=- 2 siJk,
where G ( { } ) = r;iJGiJ ( { } ) , and we have used the identity/ J
8

(5 ·19)

These alternative forms of the gravitation al field equations, (5 ·12) and

(5 ·18), lead to the following results:
(i) Vacuum solutions of the Einstein equation also satisfy the gravitation al
field equations in vacuum. Namely, if the metric tensor, g~, = e\ek, satisfies the
Einstein equation in vacuum,

RiJ ( { }) =0, (5. 20)

then the translation gauge field ck'' and the Lorentz gauge field A. ;J 1, = tli.i" (i. e.,
torsion= 0) satisfy the gravitation al field equations without sources, Tij = 0 = Sijk·
(ii) Equation (5 ·18) contains third derivative s of the metric tensor, unless
the parameter s satisfy the conditions , *J
(5. 21)

(iii) Suppose that the condition (5 · 21) is satisfied, then the left.. hand side of
the gravitation al field equations, denoted by (5 ·12) and (5 ·18), are considerab ly
simplified. In Eq. (5 ·12), the quadratic terms of the Riemann·C hristoffel curvature
tensor are all vanishing. In (5 ·18), the first two terms disappear, and all the re-
maining terms are linear or quadratic in the torsion tensor. According ly, if the int-
rinsic spin of the source can be ignored, i.e., if Sijk = 0, then Eq. (5 ·18) is trivially
satisfied by the vanishing torsion, arid therefore, Eq. (5 ·12) reduces to the Einstein
equation,
GiJ ( { } ) = tcT1J with a= 1/2tc, (5. 22)

*l Equation (5·21) is equivalent to 3a,-i-2a,=O =az+a,+4ao .

 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
Gravity ji·om Poincare Gauge Theory of the Fundamental Particles. I 881

where Tu is symmetric because of Sijk = 0 by the assumption. Here tc denotes

Einstein's gravitational constant, /'C 8nG, with G Newton's gravitational constant.
Thus the case of 3a 2 +2a 5 =0=a,+12a 6 reduces to General Relativity, if the
intrinsic sPin tensor of the source is vanishing.

§ 6. Conclusion

vVe have started general formulation of Poincare gauge theory with linear and
quadratic Lagrangians, applied to the fundamental particles of spin 1/2, that is,
quarks and leptons. According to the group structure of the Poincare gauge
group, there were two sets of the gauge field, namely, the translation and Lorentz
gauge fields, from which we constructed the translation and Lorentz gauge field
strengths. We fixed the gravity Lagrangian which consists of the three pieces
quadratic in the translation gauge field strength, the six pieces quadratic in the
Lorentz gauge field strength, and one invariant linear in the Lorentz gauge field
strength, with ten free parameters, a, a, {3, r, a~o · .. , a 6 • Our main conclusion
here is the gravitational field equations invariant under the Poincare gauge group;
see (3 · 24) and (:3 · 30) for the independent variations of the translation and Lo·
rentz gauge fields, respectively. Further, using the conventional method men-
tioned in the Introduction, we rewrite the gravitational field equations more trans-
parent in the familiar language; see (5 ·12) and (5 ·18). These are the fundamen-
tal equations \Ve shall follow in forthcoming articles in this series. From these
alternative forms we have obtained the following results: (i) Vacuum solutions
of the Einstein equation also satisfy the gravitational field equations in vacuum.
(ii) Equation (5 ·18) contains third derivatives of the metric tensor, unless the
parameters obey the conditions, 3a 2 + 2a, = 0 = a 5 + 12a 6 • (iii) The case of 3a 2
+ 2a, = 0 = a 5 + 12a6 reduces to General Relativity if the intrinsic spin tensor of
the source is vanishing.

References
1) K. Hayashi and T. Shirafuji, Phys. Rev. D19 (1979), 3524.
2) G. 't Hooft and M. Veltman, Ann. l'Inst. Henri Poincare 20 (197 4), 69.
S. Deser and P. van Nieuwenhuizen, Phys. Rev. D10 (197 4), 401, 411.
S. Deser, P. van Nieuwenhuizen and I-I. S. Tsao, Phys. Rev. 10 (197 4), 3337.
:-J) S. Weinberg, Phys. Rev. Letters 19 (1967), 1264.
A. Salam, in Proceedings of the Eighth Nobel Symposium, edited by N. Svartholm
(Almquist and Wiksells, Stockholm, 1968).
4) C. N. Yang and R. L. Mills, Phys. Rev. 96 (195,1), 191.
5) R. Utiyarna, Phys. Rev. 101 (1956), 1597.
6) T. W. B. Kibble, J. Math. Phys. 2 (1961), 212.
See also, D. W. Sciarna, in Recent Developments in General Relativity (Pergamon, Oxford,
1962)' p. 415.
7) K. Hayashi, Pro g. Theor. Phys. 39 (1968), 494.
8) K. Hayashi and A. Bregman, Ann. of Phys. 75 (1973), 562.
9) For the ,linear Lagrangian, sec, for example, F. W. Hehl, P. von cler Heyde,
 Downloaded from https://academic.oup.com/ptp/article-abstract/64/3/866/1912611 by Centro de Investigacion y de Estudios Avanzados del Instituto Politecnico Nacional user on 31 July 2019
882 K. Hayashi and T. Shirafuji

G. D. Kerlick and J. M. Nester, Rev. Mod. Phys. 48 (1976), 393.

10) For the classical gravitational field, see R. Utiyama and B. S. DeWitt, J. Math. Phys. 3
(1962), 608. For the quantized gravitational field, see S. Deser, in Proceedings of the
Conference on Gauge Theories and Modern Field Theory, edited by R. Arnowitt and P.
Nath (MIT Press, Cambridge, Mass., 1976); S. Weinberg, in Proceedings of the XVII
International Conference on High Energy Physics, edited by J. R. Smith (Rutherford
Laboratory, Chilton, 1974).
11) K. S. Stelle, Phys. Rev. D16 (1977), 953.
12) D. N. Neville, Phys. Rev. Dl8 (1978), 3535.
13) K. Hayashi and T. Nakano, Prog. Theor. Phys. 38 (1967), 491.
14) See, for example, H. Weyl, :classical Groups (2nd edition, Princeton University Press,
Princeton, 1946); M. Hamermesh, Group Theory (Addison-\'Vesley, Reading, Mass., 1962),
chap. 7.
15) R. Penrose, in Battelle Rencontres, edited by C. M. De Witt and J. A. Wheeler (Benjamin,
New York, 1968), p. 121.
See also, K. Hayashi and T. Shirafuji, Frog. Theor. Phys. 57 (1977), 302.
16) K. Hayashi and T. Shirafuji, Frog. Theor. Phys. 57 (1977), 302,
in particular see §4 of this reference.
17) P. Bach, Math. Z. 9 (1921), 110.
C. Lanczos, Ann. of Math. 39 (1938), 842.
18) See, for example, L. P. Eisenhart, Riemannian Geometry (Princeton University Press,
Princeton, 1949), p. 82.