447

Progress of Theoretical Physics, Vol. 90, No.2, August 1993

Each Other's Complementary Massless Theory

for Totally Antisymmetric Tensor Gauge Fields

Masanori KOBAYASHI

Physics Department, Giju University, Giju 501-11

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015
(Received April 13, 1993)

Each other's complementary massless theory for totally antisymmetric tensor gauge fields is given in
terms of differential forms. There exist two sequential gauge transformations by which the
Faddeev-Popov ghosts are induced sequentially in a covariant quantization using the Becchi-Rouet-
Stora transformations.

§ 1. Introduction
Totally anti symmetric tensor gauge fields have received a fair amount of atten-
tion in the last decade. They have appeared in studying the problems of QCD and in
quantum gravity as well as various extended theories of gravity and supergravity.l)
The theory for totally antisymmetric tensor gauge fields is called the generalized
Maxwell theory.1)-3) The dual theory for totally antisymmetric tensor gauge fields is
obtained by the Hodge star transformation. This theory is incorporated with the
duality-transformed fields. The other theory for totally anti symmetric tensor gauge
fields follows from the replacement of the duality-transformed fields in the dual theory
by the original fields.
The aim of this paper is to propose each other's complementary massless theory
for totally antisymmetric tensor gauge fields in differential forms. 'Differential
geometry is indispensable in a neat and concise formulation.

§ 2. The massive theory in differential forms

For a massive totally antisymmetric tensor field of rank r (r=O, 1, ... , n-1) we
have arrived at the pithy expressions by using differential forms: 4 )

(2 ·1)

and

(2·2)

Here rP represents a differential form of degree r (r=O, 1, ... , n-1) in an

n-dimensional Minkowski space (a potential r-form) and m stands for a mass of the
fields, where the Minkowski metric is 7]=diag (-1,1, ... , 1) and we use the convention
cO I2 ... (n-I)=1 for the Levi-Civita tensor in n-dimensional spacetime.
We may assign an r-form smoothly at each point on a Lorentz manifold M. The
space of smooth r-forms on M is denoted by QT(M).
The Hodge star operator (the duality transformation) is a linear map * : QT(M)
448 M. Kobayashi

---> Qn-T(M), where n=dimM. The adjoint exterior derivative operator d+ : QT(M)
---> QT-l(M) is defined by

(2·3)
where d: QT(M) ---> Qr+l(M) is the exterior derivative operator.
. The Lorentz condition (d+ ¢=O) and the Bianchi identity (d¢=O) immediately
follow from the operation d+ and d on Eqs. (2·1) and (2·2) from the left-hand side,

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015
respectively.4),5)
Again operating -(dd+ + m2 ) and (d+ d + m2 ) from the left-hand side of Eqs. (2·1)
and (2·2), respectively, we obtain
-(dd+ + m2 )(d+ d + m2 )¢=(d+ d + m2 )( -l)(dd+ + m2 )¢

= -m 2 (L1+m 2 )¢, (2·4)

where L1 is the Laplace-Beltrami operator (the generalized Laplacian), which is a

linear map L1: QT(M)--->QT(M), defined by
(2·5)
Notice that the generalized Laplacian reduces to the d'Alembertian in the Minkowski
spacetime. Thus Eq. (2·4) yields that the operators (d+d+m 2 ) and -(dd++m2 ) are
each other's Klein-Gordon divisor. 6 ) This reciprocal relationship is also reflected in
the commutation relations. 7l
We have demonstrated in Ref. 4) the following scheme:

r.r. I r.r. I
*
<-------->

where * represents the Hodge star operator and r.r. stands for the existence of the
reciprocal relationship as well as the complementary relationship. The complemen-
tary relationship is used for clarifying that the two operators being each other's
Klein-Gordon divisor have to be associated with the same field. Here use has been
made of the relationship,
**¢=( _l)I+T(n-r)¢, (2·6)
where ¢EQT(M).

§ 3. Each other's complementary massless theory

The scheme in § 2 tells us that in the massless limit no reciprocal relationship

exists due to the fact m=O and the complementary relationship still survives. The
key to a formulation of massless totally antisymmetric tensor gauge fields appears to
be this complementarity.
 Each Other's Complementary Massless Theory 449

3.1. Equations of motion

For a massless totally antisymmetric gauge field the equations of motion turn out
to be
(3·1)
and

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015
(3·2)
The potential r -form (r = 0, 1, ... , n - 1) has a large number of degrees of freedom. A
¢ can always be chosen in order to yield the Lorentz condition,
(3·3)
or else the Bianchi identity,
d¢=O. (3·4)
Imposing the subsidiary conditions (3·3) and (3·4) on Eqs. (3·1) and (3·2), respective-
ly, we can obtain the d'Alembert equations
LJ¢=o. (3·5)
Equations (3 ·1) and (3·2) are rewritten in terms of field strengths as
d+F=O (3·6)
and
-dF=O, (3·7)
respectively, where F is defined by
F=d¢ (3·8)
for Eq. (3·6) and
F=d+¢ (3·9)
for Eq. (3·7). From the combined use of Eqs. (3·6) and (3·8) or Eqs. (3·7) and (3·9)
we can easily obtain
LJF=O .. (3·10)
3.2. The gauge transformations
Let us consider the gauge transformations for the equations of motion (3 ·1) and
(3· 2).
The equation of motion (3.1) and the field strength (3·8) are manifestly invariant
under the gauge transformation,
(3 ·11)
where A (r-l) is an arbitrary (r-1)-form. Since d+ ¢=O, we can impose the subsidiary
condition on A (r-l)
450 M. Kobayashi

(3 '12)
Thus we can specify A (r-l) in order to satisfy the differential forms,
d+ A(r-l)=O (3·13)
and
L/A (r-l)=O . (3·14)

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015
Next we consider the gauge transformation,
¢---7¢+d+A(r+l) , (3'15)
where A(r+l) is an arbitrary (r+ I)-form. The gauge invariance of both the equation
of motion (3·2) and the field strength (3'9) is manifest. We can impose the subsidiary
condition on A (r+l)
dd+ A(r+l)=O, (3·16)
since we have the condition d¢=O. Here A (r+l) is associated with the differential
forms,
(3 ·17)
and
(3 ·18)
3.3. The duality transformations
The space [Y(M) is isomorphic to Qn-r(M) on an n·dimensional Lorentz mani-
fold M. By operating the Hodge star we can transfer to the massless theories on
Qn-r(M). This is easily done by using the relationships (2·3) and (2·6) and the
results turn out to be tabulated as follows:
(a) the generalized Maxwell theory
eq. of motion d+d¢=O ---7 ddH¢=O
field strength F=d¢ ---7 *F=( -ly+ld H ¢
subsidiary condo d+¢=O ---7 (-IYd*¢=O
d' Alembert eqs. L/¢=O ---7 L/*¢=O
L/F=O ---7 L/*F=O

gauge transformations;
potential ¢'=¢+dA(r-l) ---7 * ¢ ' = * ¢ + ( -ly~r* A(r-l)

field F'=F ---7 *F'=*F

subsidiary condo d+dA(r-l)=O ---7 dd H A (r-l)=O
associated eqs. d+A(r-l)=O ---7 (-ly+ld* A(r-l)=O
L/A(r-l)=O ---7 L/*A(r-l)=O

(b) the other generalized Maxwell theory

eq. of motion
 Each Other's Complementary Massless Theory 451

field strength F=d+¢ ---+ *F=(-lYd*¢

subsidiary condo d¢=O ---+ ( -1)r+ 1d H ¢=0
d' Alembert eqs. LJ¢=O' ---+ LJ*¢=O
LJF=O ---+ LJ*F=O

gauge transformations;

potential ¢'=¢+d+ A(T+l) ---+ *¢'=*¢+( -lY+ld* A(T+I)

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015
field F'=F ---+ *F'=*F
subsidiary condo dd+ A(T+l)=O ---+ d+d*A(T+I)=O
associated eqs. dA(T+l)=O ---+ (-lY d H A (r+1)=0
LJA(r+I)=O ---+ LJ* A (T+I)=O .

3.4. Conclusions
In summary we have demonstrated the scheme:

c. I c. I
where c. stands for the existence of the complementary relationship. Thus we have
established two massless theories which are complementary to each other's.

§ 4. Sequential gauge transformations

We show in this section the existence of sequential gauge transformations.

Let ;l(T) be ;«T)=dA(T-I). Then we have

(4·1)
due to the subsidiary condition (3 ·12). Since the definition of ;«T) yields the condition,

(4 ·2)
we obtain

(4·3)
Consider the gauge transformation,

(4 ·4)
Then ;l(T) is obviously invariant under the transformation (4 ·4). Let ;l(T-I) be ;l(T-I)
=dA(T-2). The similar discussion is applied to this case by imposing the subsidiary
condition d+ dA (T-2) =0 and d+ A (T-2) =0. Thus we have obtained the sequential gauge
transformations.
Similarly let ;l(rl be d+ A (r+l). Then the subsidiary condition (3 ·16) yields

(4 ·5)
452 M. Kobayashi

Combining the condition obtained by the definition of AIr),

d+ ,,(r)=O (4·6)
with Eq. (4·5) yields
Ll,,(r)=o. (4 ·7)
Again let us consider the gauge transformation,

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015
A(r+l)->A(r+l)+d+A(r+Z) . (4·8)
Then the gauge invariance of ,,(r) is manifest. And let ,,(r+l) be d+ A (r+z). The similar
results are obtained by applying the discussion before. Here we have imposed the
subsidiary condition dd+ A (r+z)=o and dA (r+Z)=O. Thus we have obtained the other
sequential gauge transformations.
These state that there exist the two kinds of sequential gauge transformations:
(4 ·9)
and
(4 ·10)
where n=dimM. The fact that the sequential gauge transformations (4·9) and (4·10)
exist has appeared to be a covariant quantization by introducing the sequential
Faddeev-Popov ghosts. Quantization has been carried out for the sequence (4·9)
using the Becchi-Rouet-Stora transformations. 3 )
The sequential gauge transformations on Qn-r(M) turn out to be
*,,(r)~* ,,(r-I)~ ••• ~* AID) (4·11)

and
(4 ·12)

§ 5. Discussion

Following Tokuoka Z) we easily count the degrees of freedom for massless totally
antisymmetric gauge fields. The correct numbers of degrees of freedom for the
(r + I)-rank field strength and the (r -I)-rank field strength are denoted by I L( r) and
her), respectively. The results turn out to be /L(r)=n-zCr and h(r)=n-zCr-z for
the Lorentz condition and the Bianchi identity satisfied by the field, respectively.
Here n=dimM and r is the degree of r-form potential. Note that either of /L and
I B becomes zero for a given r. This means that the relevant systems carry no
physical modes.
The Maxwell theory appears in the case n=4 and r=l for /L(r). An interesting
theory comes up in the case n=4 and r=3 for IB(r). If we replace ¢I"VA by
(l/l!)cl"vAa¢a, the theory becomes the usual dual Maxwell theory. There is no compel-
ling reason for this choice.
 Each Other's Complementary Massless Theory 453

References
1) A. Aurilia and Y. Takahashi, Prog. Theor. Phys. 66 (1981), 693.
A good list of references can be found in this article.
2) Z. Tokuoka, Phys. Lett. 87A (1982), 215.
3) T. Kimura, Prog. Theor. Phys. 65 (1981), 338.
4) M. Kobayashi, Prog. Theor. Phys. 88 (1992), 1231.
5) A. Z. Capri and M. Kobayashi, Prog. Theor. Phys. 74 (1985), 368.
6) A. Z. Capri and M. Kobayashi, J. of Phys. A18 (1985), 2561.
7) A. Z. Capri and M. Kobayashi, Nuovo Cim. 97B (1987), 141.

Downloaded from http://ptp.oxfordjournals.org/ at UCSF Library and Center for Knowledge Management on March 13, 2015