Coisotropic submanifolds in $b$-symplectic geometry
Stephane Geudens, Marco Zambon

TL;DR
This paper extends symplectic geometry concepts to $b$-symplectic manifolds, proving normal form theorems for coisotropic submanifolds and introducing strong $b$-coisotropic submanifolds with reduced $b$-symplectic structures.
Contribution
It establishes a normal form theorem for $b$-coisotropic submanifolds and introduces strong $b$-coisotropic submanifolds with inherited reduced structures.
Findings
$b$-coisotropic submanifolds determine local $b$-symplectic structure.
Normal form theorem extends Gotay's theorem to $b$-symplectic geometry.
Reduced $b$-symplectic structures exist on quotients of strong $b$-coisotropic submanifolds.
Abstract
We study coisotropic submanifolds of -symplectic manifolds. We prove that -coisotropic submanifolds (those transverse to the degeneracy locus) determine the -symplectic structure in a neighborhood, and provide a normal form theorem. This extends Gotay's theorem in symplectic geometry. Further, we introduce strong -coisotropic submanifolds and show that their coisotropic quotient, which locally is always smooth, inherits a reduced -symplectic structure.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Coisotropic submanifolds in -symplectic geometry
Stephane Geudens
KU Leuven, Department of Mathematics, Celestijnenlaan 200B box 2400, BE-3001 Leuven, Belgium
and
Marco Zambon
Abstract.
We study coisotropic submanifolds of -symplectic manifolds. We prove that -coisotropic submanifolds (those transverse to the degeneracy locus) determine the -symplectic structure in a neighborhood, and provide a normal form theorem. This extends Gotay’s theorem in symplectic geometry. Further, we introduce strong -coisotropic submanifolds and show that their coisotropic quotient, which locally is always smooth, inherits a reduced -symplectic structure.
Contents
- 1 Background on -geometry
- 2 -coisotropic submanifolds and the -Gotay theorem
- 3 Strong -coisotropic submanifolds and -symplectic reduction
Introduction
In symplectic geometry, an important and interesting class of submanifolds are the coisotropic ones. They are the submanifolds satisfying , where denotes the symplectic orthogonal of the tangent bundle . They arise for instance as zero level sets of moment maps, and in mechanics as those submanifolds that are given by first class constraints (see Dirac’s theory of constraints). The notion of coisotropic submanifolds extends to the wider realm of Poisson geometry, and it plays an important role there too: for instance, a map is a Poisson morphism if and only if its graph is coisotropic, and coisotropic submanifolds admit canonical quotients which inherit a Poisson structure.
The Poisson structures which are non-degenerate at every point are exactly the symplectic ones. Relaxing slightly the non-degeneracy condition, one obtains Poisson structures for which the top power is transverse to the zero section of the line bundle (here ): they are called log-symplectic structures. They are symplectic outside the vanishing set of , a hypersurface which inherits a codimension-one symplectic foliation. Log-symplectic structures are studied systematically by Guillemin-Miranda-Pires in [11], and turn out to be equivalent to -symplectic structures. The latter are defined on manifolds with a choice of codimension-one submanifold , as follows: they are non-degenerate sections of which are closed w.r.t. the de Rham differential, where is the -tangent bundle (a Lie algebroid over which encodes ). In other words, they are the analogue of symplectic forms if one replaces the tangent bundle with the -tangent bundle. Because of this, various phenomena in symplectic geometry have counterparts for log-symplectic manifolds.
This paper is devoted to coisotropic submanifolds of log-symplectic manifolds. We single out two classes, which we call -coisotropic and strong -coisotropic. We prove that certain properties of coisotropic submanifolds in symplectic geometry – properties which certainly do not carry over to arbitrary coisotropic submanifolds of log-symplectic manifolds – do carry over to the above classes. Moreover, we show that these classes of submanifolds enjoy some properties that are -geometric enhancements of well-known facts about coisotropic submanifolds in Poisson geometry. We now elaborate on this.
Main results. Let be a -symplectic manifold, and denote by the corresponding Poisson tensor on . We consider two classes of submanifolds which are coisotropic (in the sense of Poisson geometry) with respect to .
A submanifold of is called -coisotropic if it is coisotropic and a -submanifold (i.e. transverse to ). An equivalent characterization is the following: a -submanifold such that . The latter formulation makes apparent that this notion is very natural in -symplectic geometry. Section 2 is devoted to the class of -coisotropic submanifolds.
We show that the -conormal bundle of a -coisotropic submanifold is a Lie subalgebroid. We also show that for Poisson maps between log-symplectic manifolds compatible with the corresponding hypersurfaces, the graphs are -coisotropic submanifolds, once “lifted” to a suitable blow-up [9]. Both of these statements are -geometric analogs of well-known facts about coisotropic submanifolds in Poisson geometry. Next, in Theorem 2.13 we show that Gotay’s theorem in symplectic geometry [8] extends to -coisotropic submanifolds in -symplectic geometry. The main consequence is a normal form theorem for the -symplectic structure around such submanifolds:
Theorem**.**
A neighborhood of a -coisotropic submanifold is -symplectomorphic to the following model:
[TABLE]
where the vector bundle denotes the kernel of the pullback of to , and is a -symplectic form which is constructed out of the pullback and is canonical up to neighborhood equivalence (see equation (15) for the precise formula).
Such a normal form allows to study effectively the deformation theory of as a coisotropic submanifold [7]. Another possible application is the construction of -symplectic manifolds using surgeries, as done for instance in [6, Theorem 6.1]. We point out that in the special case of Lagrangian submanifolds, the above result is a version of Weinstein’s tubular neighborhood theorem, and was already obtained by Kirchhoff-Lukat [13, Theorem 5.18].
In Section 3 we consider the following subclass of the -coisotropic submanifolds. A submanifold is called strong -coisotropic if it is coisotropic and transverse to all the symplectic leaves of it meets. We remark that Lagrangian submanifolds intersecting the degeneracy hypersurface never satisfy this definition.
The main feature of strong -coisotropic submanifolds is that the characteristic distribution
[TABLE]
is regular, with rank equal to . Recall the following fact in Poisson geometry: when the quotient of a coisotropic submanifold by its characteristic distribution is a smooth manifold, then it inherits a Poisson structure, called the reduced Poisson structure. We show (see Proposition 3.6 for the full statement):
Proposition**.**
Let be a strong -coisotropic submanifold of a -symplectic manifold. If the quotient by the characteristic distribution is smooth, then the reduced Poisson structure is again -symplectic.
Instances of the above proposition arise when a connected Lie group acts on a -symplectic manifold with equivariant moment map, in the sense of Poisson geometry, and is the zero level set of the latter, see Corollary 3.10. At the end of the paper we provide examples of -symplectic quotients, and – by reversing the procedure – in Corollary 3.16 we realize any -symplectic structure on the 2-dimensional sphere as such a quotient.
In order to state and prove these results, in Section 1 we collect some facts about -geometry. A few of them are new, to the best of our knowledge, and are of independent interest. More specifically, in Lemma 1.10 we show that, while the anchor map of the -tangent bundle does not admit a canonical splitting, distributions tangent to do have a canonical lift to the -tangent bundle. In Proposition 1.19 we provide a version of the -Moser theorem relative to a -submanifold, which we could not find elsewhere in the literature.
Acknowledgements. We acknowledge partial support by the long term structural funding – Methusalem grant of the Flemish Government, the FWO under EOS project G0H4518N, the FWO research project G083118N (Belgium).
1. Background on -geometry
In this section, we address the formalism of -geometry, which originated from work of Melrose [18] in the context of manifolds with boundary. We review some of the main concepts, including -symplectic structures, and we prove some preliminary results that will be used in the body of this paper.
1.1. -manifolds and -maps
We first introduce the objects and morphisms of the -category, following [11].
Definition 1.1**.**
A -manifold is a pair consisting of a manifold and a codimension-one submanifold .
Given a -manifold , we denote by the set of vector fields on that are tangent to . Note that is a locally free -module, with generators
[TABLE]
in a coordinate chart adapted to . Thanks to the Serre-Swan theorem, these -vector fields give rise to a vector bundle .
Definition 1.2**.**
Let be a -manifold. The -tangent bundle is the vector bundle over satisfying .
The natural inclusion induces a vector bundle map , which is an isomorphism away from . Restricting to , we get a bundle epimorphism , which gives rise to a trivial line bundle . Indeed, is canonically trivialized by the normal -vector field , which is locally given by where is any local defining function for . So at any point , we have a short exact sequence
[TABLE]
but this sequence does not split canonically.
Since is a Lie subalgebra of , it inherits a natural Lie bracket . The data endow with a Lie algebroid structure. The map is called the anchor of .
Definition 1.3**.**
Let be a -manifold. The -cotangent bundle is the dual bundle of .
In coordinates adapted to , the -cotangent bundle has local frame
[TABLE]
We will denote the set of Lie algebroid -forms by , and we refer to them as --forms. The space is endowed with the Lie algebroid differential , which is determined by the fact that the restriction is a chain map. Note that the anchor induces an injective map , which allows us to view honest de Rham forms as -forms.
Definition 1.4**.**
Given -manifolds and , a -map is a smooth map such that is transverse to and .
Given a -map , the usual pullback extends to an algebra morphism , see [14, Proof of Proposition 3.5.2]. That is, we have a commutative diagram
[TABLE]
This -pullback has the expected properties; for instance, the assignment is functorial, and the -pullback commutes with the -differential .
We can now define the Lie derivative of a -form in the direction of a -vector field by the usual formula
[TABLE]
where the -pullback is well-defined since the flow of consists of -diffeomorphisms. Cartan’s formula is still valid
[TABLE]
Dual to the -pullback , a -map induces a -derivative , which is the unique morphism of vector bundles that makes the following diagram commute [14, Proposition 3.5.2]:
[TABLE]
At each point , the derivative and the -derivative have the same rank, by the next result proved in [5].
Lemma 1.5**.**
Let be a -map. The anchor of restricts to an isomorphism for all .
We finish this subsection by observing that, if a -vector field can be pushed forward by the derivative of a -map , then its lift to a section of the -tangent bundle can be pushed forward by the -derivative .
Lemma 1.6**.**
Let be a surjective -map, and let be such that pushes forward to some element . Then is a well-defined section of , and it equals the unique element satisfying .
Proof.
Since is a -map, we have that is tangent to , so indeed for unique . Now, first consider . Commutativity of the diagram (2) implies that
[TABLE]
But we also have , so that injectivity of at implies . Next, we choose . Since is a -map, we can take a (one-dimensional) slice through transverse to , such that the restriction is a diffeomorphism. Since is a vector bundle map covering the diffeomorphism , the expression is well-defined and smooth. Moreover, it is equal to on the dense subset , as we just proved. By continuity, the equality holds on all of , so that in particular . This concludes the proof.∎
1.2. -submanifolds
Given a -manifold , a submanifold transverse to inherits a -manifold structure with distinguished hypersurface . Such submanifolds are therefore the natural subobjects in the -category.
Definition 1.7**.**
A -submanifold of a -manifold is a submanifold that is transverse to .
Let be a -submanifold. The inclusion of -manifolds induces a canonical map that is injective by Lemma 1.5. This allows us to view as a Lie subalgebroid of . In particular, we have the following fact.
Lemma 1.8**.**
If is a -submanifold, then for all .
Proof.
Fixing some notation, we have anchor maps and , and we put and as before. If denotes the inclusion, then we get a commutative diagram with exact rows, for points :
[TABLE]
We obtain : the inclusion “” holds by the above diagram, and the equality follows by dimension reasons since is injective. In particular, is contained in the image of , as we wanted to show. ∎
The notions of -map and -submanifold are compatible, as the next lemma shows.
Lemma 1.9**.**
Let be a -map, and assume that we have -submanifolds and such that .
- a)
Restricting gives a -map
[TABLE] 2. b)
Further, .
Proof.
- a)
We first note that
[TABLE]
since is a -map and . Next, choosing , we have to show that
[TABLE]
We clearly have the inclusion “”. For the reverse, we choose . By transversality , we know that . So we have for some and . Next, since , we have so that for some and . So we have
[TABLE]
The term in square brackets clearly lies in , and being equal to it also lies in . So it lies in , using the transversality . Hence the decomposition (5) is as required in (4). 2. b)
Denoting the inclusions and , we have . Hence by functoriality, , which implies the claim.
∎
1.3. Distributions on -manifolds
We saw that the short exact sequence (1) does not split canonically. However, its restriction to suitable distributions does split.
Lemma 1.10**.**
Let be a -manifold with anchor map .
- a)
Given a distribution on that is tangent to , there exists a canonical splitting of the anchor . 2. b)
Let denote the set of distributions on tangent to , and let consist of the subbundles of intersecting trivially . Then there is a bijection
[TABLE]
where the splitting is as in . The inverse map reads .
Proof.
a) One checks that the inclusion induces a well-defined vector bundle map
[TABLE]
where is any extension of . This map satisfies , so in particular .
b) We only have to show that if is a subbundle of intersecting trivially , then . Denote , a distribution on tangent to . The canonical splitting is injective, and and have the same rank, hence it suffices to show that . If is a section of , then for unique . We get
[TABLE]
and since the anchor is injective on sections, this implies that . ∎
Corollary 1.11**.**
Let be a -map of constant rank. Notice that is a distribution on that is tangent to . It satisfies
[TABLE]
where denotes the canonical splitting of the anchor .
Proof.
Under the bijection of Lemma 1.10 b), corresponds to , as a consequence of Lemma 1.5. ∎
1.4. Vector bundles in the -category
If is a -manifold and a vector bundle, then is naturally a -manifold and the projection is a -map. Along the zero section , the -tangent bundle splits canonically as follows.
Lemma 1.12**.**
Let be a -manifold and a vector bundle. Then at points we have a canonical decomposition
[TABLE]
Proof.
Denote by the vertical bundle. By Corollary 1.11 there is a canonical lift such that . So we get a short exact sequence of vector bundles over
[TABLE]
Here
[TABLE]
is the pullback of the vector bundle by , and the surjective vector bundle map
[TABLE]
is induced by the -map .
Restricting (6) to the zero section gives a short exact sequence of vector bundles over :
[TABLE]
This sequence splits canonically through the map induced by the inclusion . ∎
The following result makes use of the decomposition introduced in Lemma 1.12.
Lemma 1.13**.**
- a)
Let be a vector bundle over the -manifold . Denote by and the anchor maps of and respectively. Under the decomposition of Lemma 1.12, we have that the map
[TABLE]
equals . 2. b)
Consider a morphism of vector bundles over -manifolds covering a -map :
[TABLE]
Then is a -map, and its -derivative along the zero section
[TABLE]
equals .
Proof.
- a)
Since is a -submanifold of , we have that is a Lie subalgebroid of . In particular, and agree on . Next, we know that takes isomorphically to , thanks to Lemma 1.5 applied to . To see that , we choose and extend it to . Denote by the canonical splitting of , as in the proof of Lemma 1.12. Then . 2. b)
It is routine to check that is a -map, so we only prove the second statement. Taking the -derivative of both sides of the equality at a point , we know that , since . Hence by the proof of Lemma 1.12. Using a) and the diagram (2), we have a commutative diagram
[TABLE]
It implies that
[TABLE]
Finally, holds by Lemma 1.9 b). ∎
1.5. Log-symplectic and -symplectic structures
The -geometry formalism can be used to describe a certain class of Poisson structures, called log-symplectic structures. These can indeed be regarded as symplectic structures on the -tangent bundle.
Definition 1.14**.**
A Poisson structure on a manifold is a bivector field such that the bracket is a Lie bracket on . Equivalently, the bivector field must satisfy , where is the Schouten-Nijenhuis bracket of multivector fields. A smooth map is a Poisson map if the pullback is a Lie algebra homomorphism.
The bivector induces a bundle map by
[TABLE]
and the rank of at is defined to be the rank of the linear map . Poisson structures of full rank correspond with symplectic structures via .
For every , the operator is a derivation of . The corresponding vector field is the Hamiltonian vector field of . Any Poisson manifold comes with a (singular) distribution , generated by the Hamiltonian vector fields. This distribution is integrable (in the sense of Stefan-Sussman) and each leaf of the associated foliation has an induced symplectic structure .
Definition 1.15**.**
A Poisson structure on a manifold is called log-symplectic if is transverse to the zero section of the line bundle .
Note that a log-symplectic structure is of full rank everywhere, except at points lying in the set , called the singular locus of . If is nonempty, then it is a smooth hypersurface by the transversality condition, and we call bona fide log-symplectic. In that case, is a Poisson submanifold of with an induced Poisson structure that is regular of corank-one. If is empty, then defines a symplectic structure on .
Since log-symplectic structures come with a specified hypersurface, it seems plausible that they have a -geometric interpretation. As it turns out, log-symplectic structures are exactly the symplectic structures of the -category.
Definition 1.16**.**
A -symplectic form on a -manifold is a -closed and non-degenerate -two-form .
Here, non-degeneracy means that the bundle map is an isomorphism, or equivalently that is a nowhere vanishing element of .
Example 1.17*.*
[11, Example 9] In analogy with the symplectic case, the -cotangent bundle of a -manifold is -symplectic in a canonical way. Note that is naturally a -manifold, and that the bundle projection is a -map. The tautological -one-form is defined by
[TABLE]
where and . Its differential is a -symplectic form on . To see this, choose coordinates on adapted to , and let denote the fiber coordinates on with respect to the local frame . The tautological -one form is then given by
[TABLE]
with exterior derivative
[TABLE]
A log-symplectic structure on with singular locus is nothing else but a -symplectic structure on the -manifold , see [11, Proposition 20]. Indeed, given a -symplectic form on , its negative inverse defines a -bivector field , and applying the anchor map to it yields a bivector field that is log-symplectic with singular locus . Conversely, a log-symplectic structure on with singular locus lifts uniquely under to a non-degenerate -bivector field , whose negative inverse is a -symplectic form on . These processes are summarized in the following diagram:
[TABLE]
We will switch between the -symplectic and the log-symplectic (i.e. Poisson) viewpoint, depending on which one is the most convenient.
1.6. A relative -Moser theorem
We will need a relative Moser theorem in the -symplectic setting. First, we prove the following -geometric version of the relative Poincaré lemma [3, Proposition 6.8].
Lemma 1.18**.**
Let be a -manifold and a -submanifold. Denote by the inclusion. If is -closed and , then there exist a neighborhood of and such that
[TABLE]
Proof.
We adapt the proof of [3, Proposition 6.8]. We first choose a suitable tubular neighborhood of that is compatible with the hypersurface . Due to transversality , we can pick a complement to in such that for all . Fix a Riemannian metric for which is totally geodesic (e.g. [19, Lemma 6.8]). The associated exponential map then establishes a -diffeomorphism between a neighborhood of in and a neighborhood of in .
So we may work instead on the total space of . Consider the retraction of onto given by , and notice that the are -maps. The associated time-dependent vector field is given by , which is a -vector field that vanishes along . It follows that we get a well-defined -de Rham homotopy operator
[TABLE]
which satisfies
[TABLE]
Since and , the formula (10) gives . Now set . ∎
Proposition 1.19** (Relative -Moser theorem).**
Let be a -manifold and a -submanifold. If and are -symplectic forms on such that , then there exists a -diffeomorphism between neighborhoods of such that and .
Proof.
Consider the convex combination for . There exists a neighborhood of such that is non-degenerate on for all . Shrinking if necessary, Lemma 1.18 yields such that and . As in the usual Moser trick, it now suffices to solve the equation
[TABLE]
for , which is possible by non-degeneracy of . The -vector fields thus obtained vanish along since . Further shrinking if necessary, we can integrate the to an isotopy defined on . Note that the are -diffeomorphisms that restrict to the identity on . By the usual Moser argument, we have , so setting finishes the proof. ∎
Remark 1.20*.*
We learnt from Ralph Klaasse that the work in progress [15] contains a version of Proposition 1.19 that holds in the more general setting of symplectic Lie algebroids.
2. -coisotropic submanifolds and the -Gotay theorem
This section is devoted to coisotropic submanifolds of -symplectic manifolds that are transverse to the degeneracy hypersurface. The main result is Theorem 2.13, a -symplectic version of Gotay’s theorem, which implies a normal form statement around such submanifolds. This can be used, for instance, to study the deformation theory of -coisotropic submanifolds [7].
2.1. -coisotropic submanifolds
In this subsection we introduce -coisotropic submanifolds and we discuss some of their main features. First recall the definition of a coisotropic submanifold in Poisson geometry.
Definition 2.1**.**
Let be a Poisson manifold with associated Poisson bracket . A submanifold is coisotropic if the following equivalent conditions hold:
- a)
, where denotes the annihilator of . 2. b)
, where denotes the vanishing ideal of . 3. c)
is a coisotropic subspace of the symplectic vector space for all , where denotes the symplectic leaf through .
The singular distribution on appearing above is called the characteristic distribution. If is symplectic, the coisotropicity condition becomes .
Definition 2.2**.**
Let be a -symplectic manifold, and denote by the corresponding Poisson bivector field on . A submanifold of is called -coisotropic if it is coisotropic with respect to and a -submanifold (i.e. transverse to ).
Remark 2.3*.*
A -coisotropic submanifold of middle dimension is necessarily Lagrangian, i.e. is a Lagrangian subspace of the symplectic vector space for all , where denotes the symplectic leaf through . Indeed, at points away from there is nothing to prove. At points , we have
[TABLE]
where the last equality follows from transversality . On the other hand, is at least -dimensional, being a coisotropic subspace of the -dimensional symplectic vector space . Hence , which proves the claim.
Definition 2.2 can be rephrased in terms of the -symplectic form : a -coisotropic submanifold is precisely a -submanifold such that .
Proposition 2.4**.**
Let be a -submanifold of a -symplectic manifold . Then is coisotropic if and only if .
Notice that the latter condition states that is a coisotropic subbundle of the symplectic vector bundle .
Proof.
If is coisotropic, then at points of we have that , i.e. . By continuity, this inclusion of subbundles holds at all points of . Conversely, if this inclusion holds on , it follows that is coisotropic in , and using characterization b) in Definition 2.1 we see that is coisotropic in . ∎
We give an alternative description of the characteristic distribution of a -coisotropic submanifold.
Lemma 2.5**.**
Let be any -submanifold of a -symplectic manifold , and let denote the anchor of so that is the Poisson bivector corresponding with . Then
[TABLE]
Proof.
At points , the equality (11) holds by symplectic linear algebra. So let . Denote by the lift of as a -bivector field. Note that
[TABLE]
where the annihilator is taken in . We now assert:
Claim 0:
To prove the claim, we first note that the dimensions of both sides agree since
[TABLE]
where the last equality holds by transversality . Now it is enough to show that the inclusion “” holds, which is clearly the case since .
We thus obtain
[TABLE]
where in the first equality we used (12) and the claim just proved, and in the second we used the diagram (9). ∎
A general fact in Poisson geometry is that the conormal bundle of any coisotropic submanifold is a Lie subalgebroid of the cotangent Lie algebroid. We now show that the -geometry version of this fact holds for -coisotropic submanifolds.
Proposition 2.6**.**
Let be a -symplectic manifold with corresponding Poisson bivector field . Recall that is a Lie algebroid (endowed with the Lie bracket induced by ), fitting in the diagram of Lie algebroids (9). Let be a -coisotropic submanifold.
- a)
* is a Lie subalgebroid of .*
- b)
* fits in the following diagram of Lie subalgebroids of the diagram (9):*
[TABLE]
Proof.
Diagram (13) is a diagram of vector subbundles of diagram (9), by the claim in the proof of Lemma 2.5 and by equation (12).
For a), since the morphism in diagram (9) is an isomorphism of Lie algebroids, it suffices to show that is a Lie subalgebroid of . Since is the kernel of the closed -2-form , a standard Cartan calculus computation shows that this is indeed the case. It is well-known that and are also Lie subalgebroids, proving b). ∎
2.2. Examples of -coisotropic submanifolds
We now exhibit some examples of -coisotropic submanifolds. The main result of this subsection is Proposition 2.8, which shows that graphs of suitable Poisson maps between log-symplectic manifolds give rise to -coisotropic submanifolds, once lifted to a certain blow-up.
Examples 2.7*.*
- a)
Given a log-symplectic manifold , any hypersurface of transverse to is -coisotropic. 2. b)
Let be a symplectic manifold, whose non-degenerate Poisson structure we denote , and let be a log-symplectic manifold with singular locus . Then is log-symplectic with singular locus . Given a Poisson map transverse to , we have that is -coisotropic. As a concrete example, consider for instance
[TABLE]
We will now prove Proposition 2.8. We start recalling some facts from [9, §2.1]. Given a manifold and a closed submanifold of codimension , one can construct a new manifold by replacing with the projectivization of its normal bundle. The resulting manifold , the real projective blow-up of along , comes with a map
[TABLE]
which restricts to a diffeomorphism . Further, let be a submanifold which intersects cleanly , i.e. is a submanifold with . Then can be “lifted” to a submanifold of , namely the closure of the inverse image of under :
[TABLE]
Now let be log-symplectic manifolds, for . The product is not log-symplectic in general111 However it fits in a slight generalization of the notion of log-symplectic structure used in this note: indeed is a normal crossing divisor, and vector fields tangent to it give rise to a Lie algebroid to which the Poisson structure on lifts in a non-degenerate way (we thank Aldo Witte for pointing this out to us). One can check that if is a Poisson map transverse to , then intersects transversely both and . This statement generalizes Example 2.7 b) and can be viewed as an analogue of Proposition 2.8. , but [20], [9, §2.2]
[TABLE]
is log-symplectic with singular locus the exceptional divisor , and the blow-down map is Poisson, where denotes .
Proposition 2.8**.**
Let be a Poisson map with . Then
[TABLE]
is a -coisotropic submanifold of the log-symplectic manifold defined in (14).
Proof.
The intersection is clean, since it coincides with thanks to the assumption . Hence can be “lifted” to .
The resulting submanifold is coisotropic: is coisotropic in because is a Poisson map, so is coisotropic in (since is a Poisson diffeomorphism away from the exceptional divisor), and the same holds for its closure.
To finish the proof, we have to show that is transverse to the exceptional divisor . Let be local coordinates on such that , and similarly let be local coordinates on such that . Then
[TABLE]
are local coordinates on that are adapted , but also to
[TABLE]
due to the hypothesis . Hence we can apply Lemma 2.9, which yields the desired transversality. ∎
The proof of Proposition 2.8 uses the following statement, for which we could not find a reference in the literature.
Lemma 2.9**.**
Let be non-negative integers. Consider with standard coordinates , and the subspaces
[TABLE]
where and . Then, in the blow-up , the submanifold interesects transversely the exceptional divisor .
Proof.
We have
[TABLE]
where denotes the class in projective space. Notice that by taking the closure we are adding exactly the exceptional divisor
[TABLE]
We have
[TABLE]
By taking the closure we are adding exactly
[TABLE]
For every point there is a curve of the form
[TABLE]
lying in with , and clearly . Since and has codimension , we obtain . ∎
Remark 2.10*.*
One can show that for any pair of submanifolds and intersecting cleanly, around any point of the intersection there exist local coordinates of the ambient manifold that are simultaneously adapted to both submanifolds. Lemma 2.9 than implies that, with the notation of the beginning of this subsection, intersects transversely the hypersurface of .
2.3. -coisotropic embeddings and the -Gotay theorem
If is -coisotropic, then Proposition 2.4 implies that is -presymplectic, i.e. the -two-form is closed of constant rank. Conversely, in this subsection we prove that any -presymplectic manifold embeds -coisotropically into a -symplectic manifold, which is unique up to neighborhood equivalence. In other words, we show a version of Gotay’s theorem for -coisotropic submanifolds. For Lagrangian submanifolds, this becomes a version of Weinstein’s tubular neighborhood theorem, which was already obtained in [13, Theorem 5.18].
As a consequence, a -coisotropic submanifold determines (up to -symplectomorphism) in a neighborhood of . Notice that arbitrary coisotropic submanifolds of the log-symplectic manifold do not satisfy this property: for instance is a coisotropic (even Poisson) submanifold, and by [11] the additional data consisting of a certain element of is necessary in order to determine the -symplectic structure in a neighborhood of .
Definition 2.11**.**
A -presymplectic form on a -manifold is a -two-form which is closed and of constant rank.
Definition 2.12**.**
Let be a -manifold endowed with a -presymplectic form . A -coisotropic embedding of into a -symplectic manifold is a -map such that is an embedding and
- i)
. 2. ii)
is -coisotropic in .
We will prove the following Gotay theorem in the -symplectic setting.
Theorem 2.13** (The -Gotay theorem).**
Let be a -manifold with a -presymplectic form . We then have the following:
- a)
* embeds -coisotropically into a -symplectic manifold,*
- b)
the embedding is unique up to -symplectomorphism in a tubular neighborhood of , fixing pointwise.
We divide the proof of Theorem 2.13 into several steps. We roughly follow the reasoning from the symplectic case, presented in [8]. We start by constructing a -symplectic thickening of the -presymplectic manifold , from which item of Theorem 2.13 will follow.
Proposition 2.14**.**
Denote by the vector bundle . Then there is a -symplectic structure on a neighborhood of the zero section .
Proof.
Fix a complement to in , and let be the induced inclusion. It is clear that . Since both the bundle projection and the inclusion are -maps, we can define a -two-form on by
[TABLE]
Here denotes the canonical -symplectic form on as in Example 1.17, and the subscript is used to stress that the definition depends on the choice of complement .
We want to show that is -symplectic on a neighborhood of . As is clearly -closed, it suffices to prove that is non-degenerate at points .
Claim 0:
Under the decomposition
[TABLE]
of Lemma 1.12, the canonical -symplectic form is the usual pairing
[TABLE]
This claim can be checked writing in cotangent coordinates and noticing that is a linear coordinate on each fiber , i.e. .
Consider now the decomposition
[TABLE]
given by Lemma 1.12. Using Lemma 1.13 we have . Hence under the decomposition (17) we have
[TABLE]
using the above claim and recalling that . In matrix notation,
[TABLE]
for some matrix of full rank. Similarly we have , applying Lemma 1.13 to (regarded as a vector bundle map). Therefore, under (17) we get
[TABLE]
so that we get a matrix representation of the form
[TABLE]
where we also use that . Note that the matrix in (19) is of full rank since the restriction of to is non-degenerate. Combining (18) and (19), we have that
[TABLE]
which is of maximal rank. Therefore, is non-degenerate at points . ∎
Proof of item a) of Theorem 2.13.
We show that the inclusion is indeed a -coisotropic embedding, i.e.
- i)
, 2. ii)
.
We have . Note that is the inclusion of into , so that since is -Lagrangian in .
To check ii), we let and choose lying in . Let be arbitrary. Thanks to (20), we then have
[TABLE]
which forces that due to non-degeneracy of on . Hence lies in , as desired. ∎
The uniqueness statement of Theorem 2.13 is an immediate consequence of the following proposition, to which we devote the rest of this subsection.
Proposition 2.15**.**
Let be a -symplectic manifold and a -coisotropic submanifold, with induced -presymplectic form . Let and fix a splitting . Then there is a -symplectomorphism between a neighborhood of and a neighborhood of , with .
Proof.
Since is non-degenerate, we have a decomposition as symplectic vector bundles. Note that is a Lagrangian subbundle of , since
[TABLE]
We fix a Lagrangian complement to in , i.e. .
The idea of the proof is to construct a -diffeomorphism between neighborhoods of in and – obtained as a composition of -diffeomorphisms to a neighborhood in – whose -derivative at points of pulls back to , and then apply a Moser argument.
We start by establishing a -geometry version of the tubular neighborhood theorem, in which plays the role of the normal bundle to .
Claim 1:
There is a -diffeomorphism between a neighborhood of in and a neighborhood of in , satisfying .
We will construct this map in two steps:
[TABLE]
Step 1. Let denote the anchor map of and notice that its restriction to is injective. To see this, recall the decomposition
[TABLE]
and the fact that by Lemma 1.8, so that intersects trivially. As such, we get a -diffeomorphism .
Step 2. The distribution is complementary to , i.e.
[TABLE]
Indeed, by Step 1, we have at any point
[TABLE]
and moreover, if is such that , then . Now fix a Riemannian metric on such that is totally geodesic (e.g. [19, Lemma 6.8]). The corresponding exponential map takes a neighborhood of diffeomorphically onto a neighborhood of . Moreover the fibers of over are mapped into , since for and is totally geodesic. Therefore, the map222Alternatively, one can apply [2, Example 3.3.9, p. 88-89] (see also [21, Theorem 2]). is a -diffeomorphism between neighborhoods of .
We now show that has the claimed property. That is, we show that is the identity map on , by checking that it acts as the identity on sections. We will need the commutative diagram
[TABLE]
which implicitly uses of Lemma 1.13. We will also use that for all the ordinary derivative reads
[TABLE]
For a section we now compute
[TABLE]
using (23) in the first equality and (24) in the second. Since the anchor is injective on sections, this implies that , as desired. Claim 1 is proved.
Next, the map
[TABLE]
is an isomorphism of vector bundles covering , whence a -diffeomorphism between the total spaces (For the injectivity, note that implies that as in (21), so that ). The composition is a -diffeomorphism between neighborhoods of , with .
Claim 2:
This -diffeomorphism satisfies .
As before, let denote the bundle projection, and let be the inclusion induced by the splitting . Since is a vector bundle morphism covering , by Lemma 1.13 b) we have that
[TABLE]
equals . Furthermore by Claim 1. Therefore, for and , we have
[TABLE]
Recalling equation (15) and applying Lemma 1.13 as in the proof of Proposition 2.14, we expand the right hand side of (25) as follows:
[TABLE]
using equation (16) in the second equality, writing , and using in the last equality that is a Lagrangian subbundle of . This finishes the proof of Claim 2.
Applying Proposition 1.19 (relative -Moser) yields a -diffeomorphism , defined on a neighborhood of , such that and . So setting finishes the proof. ∎
3. Strong -coisotropic submanifolds and -symplectic reduction
We consider a subclass of -coisotropic submanifolds in -symplectic manifolds, namely, the coisotropic submanifolds that are transverse to the symplectic leaves they meet. The main observation is that their characteristic distribution has constant rank, and the quotient (whenever smooth) by this distribution inherits a -symplectic form (Proposition 3.6).
3.1. Strong -coisotropic submanifolds
In Subsection 2.1 we have seen that a -coisotropic submanifold comes with a characteristic distribution
[TABLE]
In general, fails to be regular. To force that has constant rank, we have to impose a condition on that is stronger than -coisotropicity.
Definition 3.1**.**
A submanifold of a log-symplectic manifold is called strong -coisotropic if it is coisotropic (with respect to ) and transverse to all the symplectic leaves of it meets.
To justify this definition, we note that
[TABLE]
where denotes the symplectic leaf through . The last equation is exactly the transversality condition of Definition 3.1. Consequently, we have:
Proposition 3.2**.**
Let be a coisotropic submanifold. Then is strong -coisotropic iff the characteristic distribution of is regular, with rank equal to .
Lemma 2.5 immediately implies:
Corollary 3.3**.**
Let be strong -coisotropic. Then its characteristic distribution is tangent to , and corresponds to under the bijection of Lemma 1.10 b).
Remark 3.4*.*
If is a strong -coisotropic submanifold of intersecting , then necessarily . Indeed, if denotes the symplectic leaf through , then we have
[TABLE]
where the last inequality holds since is a coisotropic subspace of the -dimensional vector space . Alternatively, one can observe that a middle-dimensional -coisotropic submanifold is -Lagrangian (i.e. ). Its characteristic distribution satisfies
[TABLE]
so that cannot be strong -coisotropic whenever it intersects , due to Proposition 3.2.
3.2. Coisotropic reduction in -symplectic geometry
In this subsection we adapt coisotropic reduction to the -symplectic category. It is well-known that, given a coisotropic submanifold of a Poisson manifold , its quotient by the characteristic distribution is again a Poisson manifold, provided it is smooth. More precisely, the vanishing ideal is a Poisson subalgebra of , and denoting by its Poisson normalizer, we have that is a Poisson algebra. As an algebra it is canonically isomorphic to the algebra of smooth functions on the quotient , so it endows the latter with a Poisson structure, called the reduced Poisson structure.
Remark 3.5*.*
When the Poisson structure on is non-degenerate, i.e. corresponds to a symplectic form , the reduced Poisson structure on is also non-degenerate. Indeed [22], it corresponds to the symplectic form on obtained by symplectic coisotropic reduction, i.e. the unique one that satisfies , where is the projection and is the inclusion.
Proposition 3.6** (Coisotropic reduction).**
Let be a strong -coisotropic submanifold of a -symplectic manifold . Then is a (constant rank) involutive distribution on . Assume that has a smooth manifold structure, such that the projection is a submersion. Then inherits a -symplectic structure , determined by
[TABLE]
where is the inclusion. Its corresponding log-symplectic structure is exactly the reduced Poisson structure on obtained from .
Proof.
We know that has constant rank, by Proposition 3.2. As for involutivity, first note that is generated by Hamiltonians of functions . On such generators, we have
[TABLE]
where due to coisotropicity of . Hence is involutive.
The quotient is a smooth submanifold of , since for every slice in transverse to , the intersection is a smooth slice in transverse to . The leaf space is a -manifold, and the projection is a -map. For , we have an exact sequence
[TABLE]
which corresponds with an exact sequence on the level of -tangent spaces
[TABLE]
To see this, consider the canonical splitting of the anchor , as constructed in Lemma 1.10 , and notice that
[TABLE]
where the first equality holds by Corollary 1.11 and the third by Corollary 3.3.
Since is a surjective submersion, it admits sections, hence for every sufficiently small open subset there is a submanifold transverse to such that is a diffeomorphism. At points we have
[TABLE]
due to the sequence (28). This implies that is a -symplectic form on , where is the inclusion and is the restriction of to . Denote by the inverse of . Then is -symplectic on . Away from , this -2-form agrees with the symplectic form obtained by symplectic coisotropic reduction from . Denote by the non-degenerate -bivector on corresponding to . Away from , the log-symplectic structure agrees with the reduced Poisson structure, by Remark 3.5. By continuity, the same is true on the whole of . As was arbitrary, the reduced Poisson structure on is log-symplectic, and the above reasoning shows that the corresponding -symplectic form satisfies equation (27). ∎
Examples 3.7*.*
- a)
Let be a -submanifold. A quick check in coordinates shows333The converse is also true. If is strong -coisotropic in , then is transverse to , which implies that is transverse to , i.e. that is a -submanifold. that is strong -coisotropic in . Its quotient is canonically -symplectomorphic to . To see this, consider the surjective submersion
[TABLE]
and notice that the fibers of coincide with the leaves of the characteristic distribution on . So we get a -diffeomorphism . To see that this is in fact a -symplectomorphism, we note that the tautological -one-forms on and are related by
[TABLE]
where is the inclusion. Recall that the -symplectic form on is determined by the relation , where is the projection (cf. (27)). Hence to conclude that is -symplectic, we have to show that . But this is immediate from (29) since . 2. b)
Given a -manifold , let be a distribution on tangent to . Thanks to Lemma 1.10 a) we can view as a subbundle of . Its annihilator is strong - coisotropic in , and the quotient is , whenever is smooth. We give a proof of this fact in the particular case of a Hamiltonian group action, see Corollary 3.13.
3.3. Moment map reduction in -symplectic geometry
Recall that, given an action of a Lie group on a Poisson manifold , a moment map is a Poisson map satisfying
[TABLE]
Here is the -component of , the vector field is the infinitesimal generator of the action corresponding with , i.e.
[TABLE]
and is endowed with its canonical Lie-Poisson structure [4, Section 3]. A -equivariant map satisfying (30) is automatically Poisson [23, Proposition 7.30].
In view of Proposition 3.6, we recall a general fact about equivariant moment maps.
Lemma 3.8**.**
Let be a Lie group acting on a Poisson manifold with equivariant moment map . Assume the action is free on . Then
- a)
* is a coisotropic submanifold of .* 2. b)
* is transverse to all symplectic leaves of it meets.* 3. c)
the characteristic distribution on coincides with the tangent distribution to the orbits of .
Remark 3.9*.*
(i) When is a log-symplectic manifold, Lemma 3.8 implies that the level set is a strong -coisotropic submanifold.
(ii) When a torus, there is a more flexible notion of moment map [12, Definition 22] for log-symplectic manifolds. The smooth level sets of such moment maps are not strong -coisotropic submanifolds in general. Indeed they can even fail to be transverse to the degeneracy locus (see [12, Example 23] for an instance where itself is such a level set).
For the sake of for completeness we provide a proof of Lemma 3.8. Items a) and c) also follow from well-known facts in symplectic geometry, by restricting the -action to each symplectic leaf (whenever is connected) and using item b).
Proof.
- a)
We show that [math] is a regular value of . Choosing , it is enough to prove that the restriction is surjective. To this end, assume that annihilates d_{p}J(\text{Im}\big{(}\Pi_{p}^{\sharp})\big{)}. We then get for all that
[TABLE]
and therefore . Since the action is free, this implies that . It follows that d_{p}J\big{(}\text{Im}(\Pi_{p}^{\sharp})\big{)}=\mathfrak{g}^{*}, so [math] is indeed a regular value of . In particular, is a submanifold of . The coisotropicity of follows since it is the preimage of a symplectic leaf under a Poisson map. 2. b)
Let denote the symplectic leaf through . By the computation (3.1), it suffices to prove that is injective. Since [math] is a regular value, this annihilator is given by We now have a composition of maps
[TABLE]
that is injective by freeness of . In particular, is injective. 3. c)
We have
[TABLE]
which is exactly the tangent space of the -orbit through .
∎
Combining Proposition 3.6 with Lemma 3.8, we obtain a moment map reduction statement in the -symplectic category. The case was already addressed in [10, Proposition 7.8].
Corollary 3.10** (Moment map reduction).**
Consider an action of a connected Lie group on a -symplectic manifold with equivariant moment map . Assume the action is free and proper on . Then is a strong -coisotropic submanifold, and its reduction is -symplectic.
Remark 3.11*.*
The fact that is -symplectic follows already from [16, Theorem 3.11], taking there. (The hypothesis made there, that has contant rank for all , is satisfied since is transverse to ). In that reference the authors develop a reduction theory for level sets of arbitrary regular values satisfying the contant rank hypothesis, their statement is thus more general than the reduction statement in our Corollary 3.10.
3.3.1. Exact -symplectic forms
As a particular case of the previous construction, we consider the -symplectic analog of a well-known fact in symplectic geometry. Recall that, if a Lie group acts on an exact symplectic manifold and is invariant under the action, then defined by
[TABLE]
is an equivariant moment map for the action (in the sense of (30)). For a proof, see for instance [1, Theorem 4.2.10]. A similar result holds in -symplectic geometry.
Lemma 3.12** (Exact -symplectic forms).**
Suppose is a -manifold with exact -symplectic form . If is a Lie group action preserving and , then an equivariant moment map is given by . Here is the lift of the infinitesimal generator under the anchor .
Proof.
Clearly is a smooth map. Restricting the action to the symplectic manifold , we know that admits a moment map given by . Hence the equality holds on the dense subset , and as both sides are smooth on , it holds on the whole of . Similarly, since is equivariant, it follows that itself is equivariant. ∎
An example of Corollary 3.10 and Lemma 3.12 is -cotangent bundle reduction. Let us recall the picture in symplectic geometry: given an action , its cotangent lift preserves the tautological one-form and therefore it comes with an equivariant moment map given by (31)
[TABLE]
Here is the infinitesimal generator of corresponding with . If the action is free and proper, then symplectic reduction gives . Indeed, in some detail, there is a well-defined map
[TABLE]
where denotes the projection and
[TABLE]
Since the fibers of coincide with the orbits of , there is an induced bijection , which is in fact a symplectomorphism (see [17, Theorem 2.2.2]).
Corollary 3.13** (Group actions on -cotangent bundles).**
Given a -manifold and a connected Lie group , assume that is a free and proper action that preserves . Denote by the -cotangent lift of this action, that is
[TABLE]
for and . Note that the action is also free and proper, and that it preserves the hypersurface . The action has a canonical equivariant moment map , and is canonically -symplectomorphic to .
Proof.
Denote the infinitesimal generators of by and those of by , where . One checks that they are related via
[TABLE]
where denotes the projection. Since the action preserves the tautological -one form , Lemma 3.12 gives an equivariant moment map defined by . Explicitly, one has
[TABLE]
where the last equality uses (32) and Lemma 1.6. Denoting by the tangent distribution to the orbits of and by the splitting of the anchor obtained via Lemma 1.10 , the equality (33) shows that
[TABLE]
We now perform reduction on as in Corollary 3.10. Because the projection map is a -submersion with kernel , Corollary 1.11 implies that , and therefore
[TABLE]
It is now clear from (34) and (35) that -covectors in descend to , i.e. we get a well-defined map
[TABLE]
where
[TABLE]
It is easy to check that is a surjective submersion with connected fibers. From symplectic geometry we know that the fibers of and the orbits of the -action coincide on the open dense subset of . By continuity, the corresponding tangent distributions must agree on all of , and so the same holds for the foliations integrating them. Therefore, the map descends to a smooth bijective -map
[TABLE]
Being a bijective submersion between manifolds of the same dimension, is a diffeomorphism. The restriction of to the complement of , endowed with the symplectic structure obtained by symplectic (i.e. coisotropic) reduction, is a symplectomorphism onto its image. Hence, by Proposition 3.6, is a -symplectomorphism . ∎
3.3.2. Circle bundles
We find examples for Proposition 3.6 and Corollary 3.10 by “reverse engineering”.
Proposition 3.14**.**
Let be a -symplectic manifold, which for simplicity we assume to be compact. Let be a principal -bundle, with connection . Denote by the closed 2-form satisfying .
- (i)
The following is a is -symplectic manifold:
[TABLE]
Here is an interval around with coordinate , and the projection. 2. (ii)
* is a strong -coisotropic submanifold, and the reduced -symplectic manifold (as in Proposition 3.6) is isomorphic to .*
We make a few observations about . The summand of containing is necessary to ensure that is -closed. In the special case that is the trivial -bundle , choosing for the angle “coordinate” on (so ), the above lemma delivers the product of the -symplectic manifold and of the symplectic manifold .
In the special case that equals the closed 2-form , we have , which can be interpreted as the prequantization of when the latter is symplectic.
Remark 3.15*.*
By the above proposition, we actually recover by moment map reduction, as in Corollary 3.10. Indeed, acts on (trivially on the second factor) preserving the -symplectic form (since is -invariant). An equivariant moment map is , hence .
Proof.
(i) To check that is -closed, notice that its first two summands can be written as , which is closed since is closed.
For every real number sufficiently close to , is a -symplectic form on , so its -th power (where ) is a nowhere-vanishing element of . This implies that is a nowhere-vanishing element of , shrinking if necessary. Hence is -symplectic.
(ii) Denote by the singular hypersurface of . Then the singular hypersurface of is , which is transverse to . Therefore the latter is a -submanifold, and is coisotropic since it has codimension one. If denotes the inclusion, then we have . One consequence is that . Applying the anchor , we obtain that the characteristic distribution of is given by . Since the latter has constant rank one, by Proposition 3.2 we conclude that is a strong -coisotropic submanifold. A second consequence is that the reduced -symplectic manifold is isomorphic to . ∎
A concrete instance of the construction of Proposition 3.14 is the following.
Corollary 3.16**.**
Let be any smooth function on that vanishes transversely along a hypersurface. On consider the differential forms (twice the standard symplectic form) and , and denote by the radius.
- (i)
In a neighborhood of the unit sphere , the following is a -symplectic form:
[TABLE]
where is the projection. 2. (ii)
The unit sphere is a strong -coisotropic submanifold, and the reduced -symplectic manifold is where is twice the Fubini-Study symplectic form.
Remark 3.17*.*
The diagonal action of on the above neighborhood of the unit sphere in preserves and has moment map given by . This follows from Remark 3.15 and the proof below.
Proof.
On we consider the 1-form . Notice that we have . Consider the unit sphere . Let be the principal bundle given by the diagonal action of (the Hopf fibration). Then is a connection 1-form on , where is the inclusion. Then , where is the symplectic form on obtained from by coisotropic reduction. Consider the -symplectic form on . Applying Proposition 3.14 to yields a -symplectic form on , defined by
[TABLE]
We now make more explicit. Denote by the projection , let denote the radius function . Then , since the Euler vector field satisfies and . Hence, using and we obtain
[TABLE]
Using and we get . If we now use the identification between and a neighborhood of in (so ), then the expression (37) becomes (36). ∎
Remark 3.18*.*
We show directly from its definition (36) that satisfies the transversality condition required for -symplectic forms. As vanishes, one obtains . The dual 4-vector field is thus transverse to the zero section, in a neighborhood of the unit sphere .
Example 3.19*.*
We display an example of a function on which vanishes on the circle . The function on is -invariant, hence descends to a function on , which is readily seen to vanish exactly on . It vanishes linearly there: using homogeneous the coordinate on the open subset of , we have444To see this, first notice that on we have , and then divide numerator and denominator by . , which vanishes with non-zero derivative on . Since is quadratic, we have , hence the coefficient in equation (36) reads
[TABLE]
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] R. Abraham and J. Marsden. Foundations of Mechanics . Addison-Wesley, 2nd edition, 1987.
- 2[2] A. Candel and L. Conlon. Foliations. I , volume 23 of Graduate Studies in Mathematics . American Mathematical Society, Providence, RI, 2000.
- 3[3] A. Cannas da Silva. Lectures on symplectic geometry , volume 1764 of Lecture Notes in Mathematics . Springer-Verlag, Berlin, 2001.
- 4[4] A. Cannas da Silva and A. Weinstein. Geometric Models for Noncommutative Algebras , volume 10 of Berkeley Mathematics Lecture Notes series . American Mathematical Society, 1999.
- 5[5] G. Cavalcanti and R. Klaasse. Fibrations and log-symplectic structures. Journal of Symplectic Geometry , 17:603–638, 2019.
- 6[6] G. R. Cavalcanti. Examples and counter-examples of log-symplectic manifolds. J. Topol. , 10(1):1–21, 2017.
- 7[7] S. Geudens and M. Zambon. Deformations of Lagrangian submanifolds in b 𝑏 b -symplectic geometry.. In preparation.
- 8[8] M. J. Gotay. On coisotropic imbeddings of presymplectic manifolds. Proc. Amer. Math. Soc. , 84(1):111–114, 1982.
