Reduction to integration over single Lagrangian submanifold
Here we will explain that integration over a family of Lagrangian submanifolds can be reduced to integration over a single Lagrangian submanifold in a larger phase space.
A family of Lagrangian submanifolds 
 defines a single
Lagrangian submanifold in 
. We will
describe the construction of 
 in two steps: first locally in a vicinity of some point
of 
, and then globally.
1 Local construction
Let us pick some fixed Lagrangian submanifold 
 from our family 
:
Let us define a “first try” Lagrangial submanifold 
 as follows:
this is a direct product of two Lagrangian submanifolds,
and the zero section
Locally, in the vicinity of 
, we can present 
 as a family of gauge fermions 
(for each 
, sufficiently close to 
, the 
 is obtained from 
 as a deformation
corresponding to the gauge fermion 
). Let us  consider the following function:
We consider it as a gauge fermion deforming 
 to some new Lagrangian submanifold.
This new Lagrangian submanifold is what we need; we call it 
.
2 Global definition of  
 
As a first step, let us consider a submanifold 
 which is defined as follows:
This can be promoted to a subspace 
 quite trivially:
Finally, we will construct 
 as some section of the vector bundle
restricted to 
. Which section?
The simplest guess would be the zero section. However
that would not be a Lagrangian submanifold, so zero section is a wrong guess.
There is, however, a natural nonzero section. It is constructed as follows:
where 
 computes for every tangent vector to 
 the value of its corresponding generating function on 
.
This section defines our big Lagrangian submanifold:
3 Integration
There is a natural BV Hamiltonian 
 on 
. It descirbes
the lift to 
 of the natural nilpotent vector field 
 on 
.
We have:
  | 
In the case of Yang-Mills theory, the commonly accepted notations for fields are:
4 Proof that 
 is Lagrangian
First of all, we have to eliminate the ambiguity in the definition of 
. We do this by lifting
 to a subset of 
. Now to every point 
 corresponds 
, and:
The way we defined the moment map,is a function on
. But since we lifted
to
, it defines for us a function on
, which is the same as a section of
. On the other hand, from any section:
 we can naturally construct a sectionof
in the following way: first construct
where
and
 and then compose it with the natural embedding. This is what Eq. (50) does.
We have to prove that the restriction of the symplectic form of 
 to 
 is zero.
We notice that it is equal to the pullback under the natural projection 
(see Eq. (48)) of the following 2-form on 
:
Therefore we need to prove that 
 is zero.
Let 
 be a vector field on 
. By our definition of the moment map,
it generates the vector field 
 on 
.
Let 
 be a vector in 
 tangent to 
. Then, using general formulas:
Therefore 
 vanishes on any pair of vectors
when one of them is tangent to 
.
Since  is a Lagrangian submanifold, this completes the proof that 
 is zero.