Category Archives: Research work

Jacobi algebroids and quasi Q-manifolds

In “Jacobi algebroids and quasi Q-manifolds”  (arXiv:1111.4044v1 [math-ph]) I reformulate the notion of a Jacobi algebroid (aka generalised Lie algebroid or Lie algebroid in the presence of a 1-cocycle) in terms of an odd Jacobi structure of weight minus one  on the total space of the “anti-dual bundle” \(\Pi E^{*}\). This mimics the weight minus one Schouten structure associated with a Lie algebroid. The weight is assigned as zero to the base coordinates ans one to the (anti-)fibre coordinates.

Recall that a Lie algebroid can be understood as a weight one homological vector field  on the “anti-bundle” \(\Pi E\). What is the corresponding situation for Jacobi algebroids?

Well, this leads to a new notion, what I call a quasi Q-manifold…

A quasi Q-manifold is a supermanifold equipped with an odd vector field \(D\) and an odd function \(q\) that satisfy the following

\(D^{2}= \frac{1}{2}[D,D] = q \: D\)



The extreme examples here are

  1. Q-manifolds, that is set \(q=0\). Then \(D^{2}=0\).
  2. Supermanifolds with a distinguished (non-zero) odd function, that is set \(D=0\).  (This includes the cotangent bundle of  Schouten and higher Schouten  manifolds)
  3. The entire category of supermanifolds if we set \(D=0\) and \(q =0\).


The theorem here is that a Jacobi algebroid,  understood as a weight minus one Jacobi structure on \(\Pi E^{*}\) is equivalent to  \(\Pi E\) being a weight one  quasi Q-manifold.  I direct the interested reader to the preprint for details.

A nice example is \(M:= \Pi T^{*}N \otimes \mathbb{R}^{0|1}\), where \(N\) is a pure even classical manifold.  The supermanifold \(M\) is in fact an odd contact manifold or equivalently an odd Jacobi manifold of weight minus one, see arXiv:1101.1844v3 [math-ph]. Then  it turns out that \(M^{*} := \Pi TN\otimes \mathbb{R}^{0|1}\)  is a weight one quasi Q-manifold. It is worth recalling that \(\Pi T^{*}N\) has a canonical Schouten structure (in fact odd symplectic) and that \(\Pi TN\) is a Q-manifold where the homological vector field is identified with the de Rham differential on \(N\).  Including the “extra odd direction” deforms these structures.

As far as I can tell quasi Q-manifolds are a new class of supermanifold that generalises Q-manifolds and Schouten manifolds.  It is not know if other examples of such structures outside the theory of Lie and Jacobi algebroids are interesting. Only time will tell…

A simple QS and odd Jacobi manifold

Let us quickly recall what I mean by a QS and an odd Jacobi manifold.

Definition A supermanifold equipped with a Schouten structure S and a homological vector field Q such that

\(\{ S, \mathcal{Q} \} =0 \),

where \(\mathcal{Q}\) is the symbol of the homological vector field is said to be a QS-manifold.

This definition allows us to write everything in terms of an odd function quadratic in momenta and an odd function linear in momenta, ie. functions on the total space of the cotangent bundle of our supermanifold. The bracket in the above is the canonical Poisson bracket.  (The example I will give will make this clearer.)

Definition A supermanifold equipped with an almost Schouten structure  S and a homological vector field Q such that

\(\{ S, S \} ={-} 2 \mathcal{Q} S\),

\(\{ S,\mathcal{Q} \} =0\),

where \(\mathcal{Q}\) is the symbol of the homological vector field is said to be an  odd Jacobi manifold.

Both these species of supermanifold are very similar.  QS-manifolds have a genuine Schouten structure, that is an odd function quadratic in momenta such that it Poisson self-commutes and Poisson commutes with the symbol of the homological vector field.  An  odd Jacobi manifold consists of an almost Schouten structure that has a very specific Poisson self-commutator and Poisson commutes with the symbol of the homological vector field.

On to our example…

Consider the supermanifold \(\mathbb{R}^{1|1}\), which we equip with local coordinates \((t, \xi)\). Here \(t\)  is the commuting coordinate and  \(\xi \) is the anticommuting coordinate. This supermanifold comes equipped with a canonical Schouten structure

\(S = {-}\pi p\),

where we employ fibre coordinates \((p, \pi)\) on the cotangent bundle.  As the above structure does not contain conjugate variables is it cleat that

\(\{S,S \}=0\).

We can go a little further than this as we also have a canonical homological vector field, which indeed gives rise to a symbol that Poisson commutes with the Schouten structure:

\(\mathcal{Q} = {-}\pi\).

So \(\mathbb{R}^{1|1}\) is a QS-manifold, canonically.  The associated Schouten bracket is given by

\([f,g]_{S} = ({-}1)^{\widetilde{f}}\frac{\partial f}{\partial \xi} \frac{\partial g}{\partial t} {-} \frac{\partial f}{\partial t}\frac{\partial g}{\partial \xi}\),

for all \(f,g \in C^{\infty}(\mathbb{R}^{1|1})\).

Interestingly, we can also consider these structures as being odd Jacobi. Explicitly one can calculate the Poisson self-commutator of the Schouten structure and arrive at

\(\{ S, S\} = {-} 2 \left(  {-} \pi\right)\left(  {-}\pi p\right)\),

which is of course zero as \(\pi^{2}=0\). But also notice that this defines an odd Jacobi structure! We then can assign an odd Jacobi bracket as

\([f,g]_{J} = ({-}1)^{\widetilde{f}}\frac{\partial f}{\partial \xi} \frac{\partial g}{\partial t} {-} \frac{\partial f}{\partial t}\frac{\partial g}{\partial \xi}{-}({-}1)^{\widetilde{f}}\left(  \frac{\partial f}{\partial \xi}\right)g {-}f\left(  \frac{\partial g}{\partial \xi}\right) \).


The Schouten bracket satisfies a strict Leibniz rule as where the odd Jacobi bracket does not, we have an “anomaly” term in the derivation property. Both satisfy the appropriate graded version of the Jacobi identity.


Interestingly, the Schouten structure on \(\mathbb{R}^{1|1}\) is in fact non-degenerate so we have an odd symplectic supermanifold. One can also consider \(\mathbb{R}^{1|1}\) as an even contact manifold, but I will delay talking about that for now.

One could of course “compactify” \(\mathbb{R}\) and consider the supercircle \(\mathbb{S}^{1|1}\), and this naturally also can be considered as QS and odd Jacobi. Again we have a natural contact structure here and this has been studied in relation to super versions of the Schwarzian derivative. This is really another story…

More details can be found in an older post of mine here. A preprint about odd Jacobi structures can be found on the arXiv here.



Contact structures and supersymmetric mechanics

Contact structures and supersymmetric mechanics

Andrew James Bruce

We establish a relation between contact structures on supermanifolds and supersymmetric mechanics in the superspace formulation. This allows one to use the language of contact geometry when dealing with supersymmetric mechanics.

arXiv:1108.5291v1 [math-ph]

In the preprint above I show that aspects of  d=1, N=2 supersymmetric quasi-classical mechanics in the superspace formulation can be understood in terms of  a contact structure on the supermanifold \(R^{1|2}\).


In particular if we pick local coordinates \((t, \theta, \bar{\theta})\) then the super contact structure is given by


\(\alpha = dt + i \left(  d \bar{\theta}\theta + \bar{\theta} d \theta  \right)\),
which is a Grassmann odd one form. One could motivate the study of such a one form as a “superisation” of the contact form on \(R^{3}\).


Associated with any odd one form that is nowhere vanishing is a hyperplane distribution of codimension (1|0). That is we have a subspace of the tangent bundle that contains one less even vector field in its (local) basis as compared to the  tangent bundle.  This is why we should refer to the above structure as an even (pre-)contact structure.


The hyperplane distribution associated with the super contact structure is spanned by two odd vector fields. These odd vector fields are exactly the SUSY covariant derivatives. More over we do have a genuine contact structure as the exterior derivative of the super contact form is non-degenerate on the hyperplane distribution. For more details see the preprint.


Generalising contact structures  on manifolds to  supermanifolds appears fairly straight forward. We have the non-classical case of odd contact structures to also handle, here the hyperplane distribution is of corank (0|1), i.e. one less odd vector field. There is also a subtly when defining kernels and contactomorphisms as we will have to take care with nilpotent objects.

Comments on the preprint will be very much appreciated.



Update A third revised version has now been submitted. 08/02/2012

Higher Lie-Schouten brackets

I thought it would be interesting to point out a geometric construction related to  \(L_{\infty}\)-algebras.  (See earlier post here) Recall that given a Lie algebra \((\mathfrak{g}, [,] )\) one can associate on the dual vector space a linear Poisson structure known as the Lie-Poisson bracket.  So, as a  manifold \((\mathfrak{g}^{*}, \{, \}) \) is a Poisson manifold.  It is convenient to  replace the “classical” language of linear and replace this with a graded condition. That is, if we associate weight one to the coordinates on \(\mathfrak{g}^{*} \) then the Lie-Poisson bracket is of weight minus one.

The Lie Poisson bracket is very important in deformation quantisation (both formal and C*-algebraic). There are some nice theorems and results that I should point to at some later date.

Now, it is also known that one has an odd version of this known as the Lie-Schouten brackets on \(\Pi \mathfrak{g}^{*}\). The key difference is the shift in the Grassmann parity of the “linear” coordinates.  Note that this all carries over to Lie super algebras with no problem.  I will drop the prefix super from now on…


So, let us look at the situation for \(L_{\infty}\)-algebras. We understand these either as a series of higher order brackets on a vector space  \(U\) that satisfies a higher order generalsiation of the Jacobi identities or more conveniently we can understand all this in terms of a homological vector field on the formal manifold \(\Pi U\).

Definition An \(L_{\infty}\)-algebra is a vector space \(V = \Pi U\) together with a homological vector field \(Q = (Q^{\delta} + \xi^{\alpha} Q_{\alpha}^{\delta} + \frac{1}{2!} \xi^{\alpha} \xi^{\beta} Q_{\beta \alpha}^{\delta} + \frac{1}{3!} \xi^{\alpha} \xi^{\beta} \xi^{\gamma} Q_{\gamma \beta \alpha}^{\delta} + \cdots) \frac{\partial}{\partial \xi^{\delta}}\),

where we have picked coordinates on \(\Pi U\)  \(\{  \xi^{\alpha}\}\). Note that these coordinates are odd as compared to the coordinates on \(U\). Thus we assign the Grassmann parity \(\widetilde{\xi^{\alpha}} = \widetilde{\alpha} + 1\)  Note that \(Q\) is odd and that if we restrict to the quadratic part then we are back to Lie algebras.

I will simply state the result, rather than derive it.

Proposition Let \((\Pi U, Q)\) be an \(L_{\infty}\)-algebra. Then the formal manifold \(\Pi U^{*}\) has a homotopy Schouten algebra structure.

Let us pick local coordinates \(\{ \eta_{\alpha}\}\) on \(\Pi U^{*}\). Furthermore, we consider this as a graded manifold and attach a weight of one to each coordinate.  A general function,  a  “multivector” has the form

\(X = \stackrel{0}{X} + X^{\alpha} \eta_{\alpha} + \frac{1}{2!}X^{\alpha \beta}\eta_{\beta} \eta_{\alpha} + \cdots \)

The higher Lie-Schouten brackets are given by

\((X_{1}, X_{2}, \cdots, X_{r}) = \pm Q_{\alpha_{r}\cdots  \alpha_{1} }^{\beta}\eta_{\beta}\frac{\partial X_{1}}{\partial \eta_{\alpha_{1}}} \cdots   \frac{\partial X_{1}}{\partial \eta_{\alpha_{r}}}\),

being slack with an overall sign.  Note that with respect to the natural weight the n-bracket has weight (1-n). Thus not unexpectedly, restricting to n=2 gives an odd bracket of weight minus one: up to conventions this is the Lie-Schouten bracket of a Lie algebra.

The above collection of brackets forms an \(L_{\infty}\)-algebra in the “odd super” conventions that satisfies a derivation rule of the product of “multivectors”. Thus the nomenclature homotopy Schouten algebra and higher Lie-Schouten bracket.

A similar statement holds in terms of a  homotopy Poisson algebra on \(U^{*}\). Here the brackets as skewsymmetric and of  even/odd Grassmann parity for even/odd number of arguments.  (I rather the odd conventions overall).

Now this is quite a new construction and the technical exploration of this nice geometric construction awaits to be explored. How much of the geometric theory associated with Lie algebras and Lie groups carries over to \(L_{\infty}\)-algebras and \(\infty\)-groups is an open question.

Details can be found in Andrew James Bruce ” From \(L_{\infty}\)-algebroids to higher Schouten/Poisson structures”, Reports on Mathematical Physics Vol. 67, (2011), No. 2  (also on the arXiv).


Also see earlier post here on Lie infinity algebroids.

29th North British Mathematical Physics Seminar

I have been invited to give a talk at the 29th North British Mathematical Physics Seminar (NBMPS) in Edinburgh on the 16th February 2011.

The NBMPS has been running since 2001 and is a forum for mathematical physicists in North Britain to meet up. They organise four one-day meetings that are held in rotation every year in Durham, Edinburgh, York and Nottingham.

The 29th meeting is in Edinburgh and I am listed a the first speaker!

The topic of my talk will be my preprint on Odd Jacobi manifolds and classical BV-gauge systems. This paper is also discussed on my blog here.

I will place a link to the slides in due course.

I will also place an update of the event at some later date.


The talk I felt went well. I had a few questions and comments, but nothing off putting or massively critical.


Update: The slides for the talk can be found here.

Odd Jacobi structures and BV-gauge systems

In this paper we define Grassmann odd analogues of Jacobi structures on supermanifolds. We then examine their potential use in the Batalin-Vilkovisky formalism of classical gauge theories.

arXiv:1101.1844v1 [math-ph]

In my latest preprint I construct a Grassmann odd analogue of Jacobi structures on supermanifolds.

Without any details (being slack with signs) an odd Jacobi structure on a supermanifold is an ” almost Schouten structure”, \(S\) that is an odd function on the total space of the cotangent bundle of the supermanifold quadratic in fibre coordinates and a homological vector field \(Q\) on the supermanifold together with the compatibility conditions

\(L_{Q}S = \{\mathcal{Q}, S \} = 0\),
\( \{S, S \} = 2 S \mathcal{Q}\),

where \(\mathcal{Q} \in C^{\infty}(T^{*}M)\) is the “Hamiltonian” or principle symbol of the Homological vector field. The brackets here are the canonical Poisson brackets on the cotangent bundle.

An odd Lie bracket can then be constructed on \(C^{\infty}(M)\)

\([f,g] = \pm \{ \{ S,f \},g\} \pm \{ \mathcal{Q},fg \} \).

So, this odd bracket satisfies all the properties of a Schouten bracket i.e. symmetry and the Jacobi identity, but the Leibniz rule is not identically satisfied. There is an “anomaly” to the Leibniz rule of the form

\( [f,gh] = \pm [f,g]h \pm g [f,h] \pm [f,1] gh\)

In the preprint I examine the basic properties of odd Jacobi manifolds. The definition and study of odd Jacobi manifolds appears to be missing from the previous literature despite the wide interest in Schouten manifolds and Q-manifolds in mathematical physics.

One should note that for classical or even Jacobi structures (if you know what these are) the Reeb vector field has no constrain on it like being homological. For odd structures the homological condition is essential.

I also consider if the classical BV-antifield formalism can be generalised to odd Jacobi manifolds. In short, does one require the antibracket to be a Schouten bracket or can one weaken the Leibniz rule? I show that it looks possible to extend the BV formalism, classically anyway to odd Jacobi manifolds with the extra condition that the extended classical action not just be a Maurer-Cartan element,

\([s,s] = 0\),

but in addition should be Q-closed,

\(Qs =0\).

Much work needs to be done to generalise the BV formalism to odd Jacobi manifolds including adding the required gradings of ghost number, antifield number etc as well as understanding the quantum aspects.

UPDATE: 22 March 2011. I have found a mistake in one of the examples I suggest. This is corrected and an updated version of the preprint will appear in due course. The mistake does not really effect the rest of the preprint.

Lie-Infinity Algebroids? II

This post should be considered as part two of the earlier post Lie-Infinity Algebroids?

The term \(L_{\infty} \) -algebroid seems not to be very well established in the literature. A nice discussion of this can be found at the nLab.

To quickly recall, the definition I use is that the Q-manifold \((\Pi E, Q)\) is an \(L_{\infty} \) -algebroid, where \(E \rightarrow M \) is a vector bundle and \(Q \) is an arbitrary weight homological vector field. The weight is provided by the assignment of zero to the base coordinates and one to the fibre coordinates. If the homological vector field is of weight one, then we have a Lie algebroid.

It is by now quite well established that a Lie algebroid, as above is equivalently described by

i) A weight minus one Schouten structure on the total space \(\Pi E^{*}\).
ii) A weight minus one Poisson structure on the total space of \(E^{*}\).

In other words, Lie algebroids are equivalent to certain graded Schouten or Poisson algebras. Recall, a Schouten algebra is an odd version of a Poisson algebra. The point is ignoring all gradings and parity, we have a Lie algebra such that the Lie bracket satisfies a Leibniz rule over the product of elements of the Lie algebra. We need a notion of multiplication, in this case it is just the “point-wise” product of functions.

Thus, there is a close relation between Poisson/Schouten algebras (or manifolds) and Lie algebroids.

The natural question now is “does something similar happen for \(L_{\infty}\)-algebroids?”

The answer is “yes”, but we now have to consider homotopy versions of Schouten and Poisson algebras.

Definition: A homotopy Schouten/Poisson algebra is a suitably “superised” \(L_{\infty}\)-algebra (see here) such that the n-linear operations (“brackets”) satisfy a Leibniz over the supercommutative product of elements.

This definition requires that we don’t have just an underlying vector space structure, but that of a supercommutative algebra. I will assume we also have a unit. Though, I think that noncommutative and non-unital algebras are no problem. The point is, I have in mind (at least for now) algebras of functions over (graded) supermanifolds.

Theorem: Given an \(L_{\infty}\)-algebroid \((\Pi E, Q)\) one can canonically construct
i) A total weight one higher Schouten structure on the total space of \(\Pi E^{*}\).
ii) A total weight one higher Poisson structure on the total space of \(E^{*}\).

Proof and details of the assignment of weights can be found in [1].

So, the point is that there is a close relation between homotopy versions of Poisson/Schouten algebras \(L_{\infty}\)-algebroids. To my knowledge, this has not appeared in the literature before. The specific case of \(L_{\infty}\)-algebras (algebroids over a “point”) also seems not to be discussed in the literature before.

The way we interpret this is interesting. We think of a Lie algebroid as a generalisation of the tangent bundle and a Lie algabra. The homological vector field \(Q \) “mixes” the de Rham differential over a manifold and the Chevalley-Eilenberg differential of a Lie algebra \(\mathfrak{g}\). Furthermore, we have a Poisson bracket on \(C^{\infty}( E^{*})\) which “mixes” the canonical Poisson on \(T^{*}M\) with the Lie-Poisson bracket on \(\mathfrak{g}^{*}\). Similar statements hold for the Schouten bracket.

For \(L_{\infty}\)-algebroids the homological vector field again generalises the de Rham and Chevalley-Eilenberg differentials, but it is now inhomogeneous. It resembles a “mix or higher order BRST-like” operator [3]. A homotopy version of the Maurer-Cartan equation naturally appears here. It is clear that we can consider the homotopy Schouten/Poisson algabras associated with an \(L_{\infty}\)-algebra as playing the role of the Lie-Poisson algebras, however there is no obvious higher brackets to consider on the cotangent bundle. It is not clear to me what should replace the tangent bundle here, if anything.

Exactly what technical use the theorem above is awaits to be explored. There are some interesting related notion in Mehta [2], I have yet to fully assimilate them. Maybe more on that another time.

[1] From \(L_{\infty}\)-algebroids to higher Schouten/Poisson structures. Andrew James Bruce, arXiv:1007.1389 [math-ph]

[2]On homotopy Poisson actions and reduction of symplectic Q-manifolds. Rajan Amit Mehta, arXiv:1009.1280v1 [math.SG]

[3] Higher order BRST and anti-BRST operators and cohomology for compact Lie algebras. C. Chryssomalakos, J. A. de Azcarraga, A. J. Macfarlane, J. C. Perez Bueno, arXiv:hep-th/9810212v2