Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
1 to 3 of 3
The BV-BRST complex of some local Lagrangian L over some manifold Σ famously comes equipped with a canonical Poisson bracket of grade +1: the antibracket.
The whole point of this construction is that it serves as a tool for computing another bracket that one is interested in: the ordinary grade-0 bracket induced from the symplectic form that is obtained by descending the canonical presymplectic form on covariant phase space (as discussed there) to the quotient by the symmetries.
Recently I was beginning to entertain myself with the idea that there should be a more immediate relation between these two brackets than I see considered in the usual literature. I was beginning to think that it should be possible to regard a Poisson n-algebra as a “homotopy” between two Poisson (n−1)-algebras. (Does that sound reasonable to anyone?)
So I started toying around with some formulas to see if I can substantiate my guess. I am not as sophisticated yet with these computations as Frédéric, who is reading here, and likely I am making some silly mistakes. But taken together this is are two good reasons to chat about my thoughts here!
So for the purposes of this post here, I’ll be a bit naive, following the physics literature, and say things like:
Let, locally, {Φa(x)} be a collection of generators for the fields and ghosts…
as well as
… and let {ˉΦa(x)} be the corresponding collection of generators for the antifields and antighosts.
The point is that I want to write down the following suggestive symbols:
In this basis the symplectic form of which the antibracket is the Poisson bracket is locally
ωBV:=∫ΣdˉΦa(x)∧dΦa(x).Here this expression is meant to be an object in the (Kähler-)differential forms on our BV-complex. If we think of the BV-complex as the Chevalley-Eilenberg algebra of a derived L-infinity algebroid, then this expression lives in the corresponding Weil algebra. As such, the situation is very similar to that of symplectic Lie n-algebroids.
Contrary to that case, however, the 2-form here is not closed under the given differential – here: the BV-differential – and what I am after is precisely the question:
What is the image of this 2-form under the BV-differential?
Notice that ωBV is of “grade” -1 and that the BV-differential will send it therefore to a 2-form of grade 0, hence an ordinary 2-form. I am about to claim that this ordinary 2-form is
ωin−ωout,where ωin is the canonical presymplectic form on covariant phase space for the incoming boundary of Σ, and ωout that for the outgoing boundary. So I am about to be claiming that
ωout=ωin+dBVωBV.If true, this would formalize the idea that I mentioned at the beginning: ωBV would be exhibited as a homotopy between the two choices for the canonical presymplectic form.
In order to check this, I now appeal to the notation and results of the old but seminal article
Gregg Zuckerman, Action principles and global geometry (pdf).
For the remainder of this message here I will assume that you either are familiar with this, or that you now spend five minutes with glancing over the first 10 pages, which is all I’ll need.
I think to make sense of my ωBV I want to identify the symbol d there with the variational derivative that Zuckerman writes ∂ on p. 4
d=∂For the action of the BV-differential I observe that by definition
dBV∫ΣˉΦa(x)∧dΦa(x)=∫Σ(dBVˉΦa(x))∧dΦa(x)=δ∫ΣLδΦa(x)EL∧dΦa(x)=∫ΣEa∧dΦa=∫ΣE,where E is Zuckerman’s Euler-Lagrange form from page 9. With the equations there this is equivalently (still writing “d” for Zuckerman’s “∂”)
⋯=∫ΣdL−∫ΣDM,where now on the right we have the boundary contribution M that induces the canonical presymplectic form that I am after: this is defined as
ωin:=∫∂inΣdMand similarly for ωout.
Collecting all this together I get finally
dBVωBV=dBVdˉΦa∧dΦa=−d(dBVˉΦa∧dΦa)=−d(∫ΣdL−∫ΣDM)=−dd∫ΣL−∫ΣDdM.Here the first term vanishes, due to d2=0. To the second term the Stokes theorem applies, as on p. 10 of Zuckerman’s article. Therefore we are left with
⋯=−∫∂ΣdM=∫∂outΣdM−∫∂inΣdM=ωout−ωin.QED.
…
Anyone who made it so far, please give me a sanity check.
Igor Khavkine kindly points out by private email that all this boils down to the following simple statement:
let in the variational bicomplex, following Zuckerman’s notation:
L∈ΩdimX,0(ConfigSpace) be the local Lagrangian;
M∈ΩdimX−1,1(ConfigSpace) be the presymplectic potential density;
U∈ΩdimX,2(ConfigSpace) be the presymplectic density;
and finally
Then the statement is simply that.
dBVA=DU.Simple as it is, has this been considered anywhere before?
I have added a discussion along the above lines to phase space in a new section via BV.
Since this still needs attention, maybe I should better put this on my personal web. For the moment I just left a cautionary remark. I think I’ll move it tomorrow to a more suitable place.
1 to 3 of 3