Appendix: the 2-form on a space of extremals

P. Deligne

In his lectures, Bernstein considered lagrangian densities L( ) on V , for in the space M of maps V ! M , or more generally for a section of a ber bundle M over V . The case of maps to M correspond to M = M V . He considered the case where the density L( ) at v 2 V depends only on and d at v , and explained how to construct a closed 2-form ! on the space E of extremals. The construction depended on a cohomology class of hypersurfaces with an oriented normal bundle: has a \past" side and a \future" side. In practice, is a \space-like" hypersurface. The construction uses integration on , hence requires some control at the in nity of , and a similar control on homologies among 's. We will ignore this by pretending that is compact and that cohomology is taken with compact support. In this appendix, our aim is to explain a small modi cation of the construction which works as well when L( ) depends on higher derivatives of . One can also take to be for instance a bundle with connection, rather than a map to some M . If V is a supermanifold, the new construction makes clear that the 2-form obtained is the same whether one works in term of super elds : V ! M and of superdensities L( ) on V , or in term of \components" of and of the Lagrangian density L( ) on Vred deduced from L( ) by \integrating out" the odd variables, for some projection V ! Vred. What we don't get are 2-forms ! , depending on , on the space M of all , and inducing ! on the space E of extremals. The basic construction is local around a space-like hypersurface and, to explain it, we will assume that V = R, the projection to R being called \time". For a 2-parameter family a1; a2 ] of extremals, we now de ne the value at (a1 ; a2 ) = (0; 0) of the pull back of ! by (a1 ; a2 ) 7! a1 ; a2]. We choose a family 1 a a2 ] of maps to
V V ;


M , deforming a1 ; 0], agreeing with a1; a2 ] in the future, and agreeing with a1; 0] in
the past:

a1 ; 0]; 1 a1 ; a2 ] = a1 ; a2 ] 1 a1 ; a2 ] = a1 ; 0]

1 a1 ; 0] =

in the (far) future; in the (far) past:

We similarly choose 2, having the same properties with the roles a1 and a2 permuted. De nition:

!(@1 ; @2 ) =


(@1 @2 L( 1) @2 @1L( 2)) at (a1 ; a2 ) = (0; 0):

The integrand vanishes in the future, where 1 = 2, as well as in the past, where both terms vanish. This makes the integral de ned. Using , with equation t = 0, we now express ! as an exterior derivative d and show independence of the choices of 1 and 2. Let Y be the function on R which is 1 for x < 0 and 0 for x > 0: Y = (x). We will have to integrate on V up to t = 0 some smooth density vanishing in the past. In the supervariety case, this means integrating Y (t) (integration on a supervariety with boundary ). This depends on , not just on red . The 1-form is de ned only at extremals, but makes sense on tangent vectors to the space M of all . Fix an extremal 0] and consider a family a] deforming it. We now de ne at the corresponding tangent vector. Let 1 be a deformation of 0] agreeing with a] for t > ", and with 0] in the past. De nition: Z (@ ) = Y (t)@ L( 1 a]) at a = 0
d dx a a

The integrand vanishes in the future, because Y (t) does, and in the past, where 1 a] is constant. This makes the integral de ned. Because 0] is an extremal, it does not depend on the choice of 1: the density @ L( 1 a]) is the variation of the density L for a variation 1 of 0], and if we consider another deformation 2 , 1 2 has support in a time


interval A; "]. It is the tangent vector for a deformation


with a support in t < 0 and

Y (t) @ L( 1 a]) @ L( 2 a])] (at a = 0)




Y (t)@ L( 3 a]) (at a = 0) = @ L( 3 a]) (at a = 0) = 0
a a


To see that ! = d , one uses the formula that for commuting vector elds X and Y , one has (d )(X; Y ) = X: (Y ) Y: (X ) : For the pull back of by (a1 ; a2) 7! a1; a2 ], this gives, provided that a1a2 ] = a1a2 ] for t > ":
1 a1 ; a2 ]


d (@1 ; @2 ) = @1 Y (t)@2 L( 1) @2 Y (t)@1 L( 2)
= =



Y (t)(@1 @2 L( ) @2 @1 L( 2))

@1 @2 L( 1) @2 @1 L( 2)

at (a1 ; a2 ) = (0; 0): the de nition of !. In the case treated by Bernstein, integration by parts shows that the form is the restriction to E of the one he considers, on M. The modi cation we presented does not require the tangent space of M at to be the space of sections of a vector bundle on V . Basically, it requires only that it be the space of sections of a soft sheave of R-vector spaces on V . One has to be able to deform in a prescribed way in di erent regions of V . For V a supermanifold, this does not preclude imposing suitable constraints on .