Exotic families of symplectic manifolds with Milnor fibers of -type
Abstract
In this paper, we give infinitely many diffeomorphic families of different Weinstein manifolds. The diffeomorphic families consist of well-known Weinstein manifolds which are the Milnor fibers of -type, and Weinstein manifolds constructed by taking the end connected sums of Milnor fibers of -type. In order to distinguish them as Weinstein manifolds, we study how to measure the number of connected components of wrapped Fukaya categories.
Keywords Exotic Weinstein structures, Milnor fiber of simple singularity, Weinstein handle, Lefschetz fibration, Wrapped Fukaya category, Symplectic cohomology
Mathematics Subject Classification 53D37 (Symplectic aspects of mirror symmetry, homological mirror symmetry, and Fukaya category), 57R17 (Symplectic and contact topology in high or arbitrary dimension), 32Q28 (Stein manifolds)
1 Introduction
1.1 Introduction
The constructions of exotic Weinstein manifolds, i.e., diffeomorphic manifolds having different Weinstein structures, have been studied extensively. See, for example, [McL09] or [AS10]. In this paper, we investigate Weinstein manifolds which are exotic to the Milnor fibers of simple singularities.
We construct different Weinstein manifolds by attaching the same Weinstein handles to the same Weinstein manifold in different ways. However, as smooth handles, they are attached in the same way. Thus, the construction gives diffeomorphic, but different Weinstein manifolds.
We note that the idea has been used historically. See [May09], [MS10], and [AS10]. The main difference between the previous works and the current paper is that, for some particular Weinstein manifolds, we will identify plumbings and end connected sums procedures as the same smooth handle attachments. However, they are different as Weinstein handle attachments. Thus, one can show that, for example, the Milnor fiber of singularity and the end connected sum of the Milnor fiber of singularity and a cotangent bundle of the sphere are diffeomorphic but different as Weinstein manifolds.
What allows us to distinguish a plumbing and an end connected sum as Weinstein manifolds, is the number of connected components of the wrapped Fukaya categories. Since the end connected sum procedure increases the number of connected components, but the plumbing procedure does not, two procedures lead to the different Weinstein manifolds. Thus, we can distinguish different Weinstein manifolds without using vanishing wrapped Fukaya categories or without using symplectic homologies. The first is used in [May09], [MS10], and the second is used in [McL09], [AS10].
The present paper contains two generalizations of the above construction of exotic pairs. More specific results will be stated in Section 1.2.
1.2 Results
In Sections 3 – 5, we observe that a plumbing procedure and an end connected sum procedure can be realized as the same smooth handle attachments, but different Weinstein handle attachments.
More precisely, we prove Theorem 1.1.
Theorem 1.1 (Technical statement is Theorem 5.2).
Let be an odd integer. The Milnor fiber of type of dimension , which is obtained by plumbing to the Milnor fiber of type, is diffeomorphic to the end connected sum of the Milnor fiber of type and . Meanwhile, they are different as Weinstein manifolds.
After proving Theorem 1.1, we generalize it in two different ways. The first generalization is to consider the end connected sums of multiple Milnor fibers of type.
Theorem 1.2 (Technical statement is Theorem 6.3).
-
1.
If , then
is an exotic family of different Weinstein manifolds where is the dimensional Milnor fiber of type, and where means the end connected sum.
-
2.
If is odd, then
is an exotic family of different Weinstein manifolds.
The second generalization is obtained by comparing Milnor fibers of different simple singularities. The detailed results are given in Theorems 1.3 and 1.4.
Theorem 1.3.
Let . Then, the following families of dimensional manifolds have a same diffeomorphism type, but they are pairwise different as Weinstein manifolds.
-
•
The Milnor fibers of and -singularities.
-
•
The Weinstein manifold and the Milnor fibers of , , and -singularities.
-
•
The Milnor fibers of and -singularities.
-
•
For any , the Minor fiber of -singularity and the Weinstein manifold .
-
•
For , the Milnor fibers of -singularities and the Weinstein manifold .
In Section 7, will be defined as an end connected sum of the Milnor fiber of singularity and .
Theorem 1.4.
Let be an even number. Then, the following pairs of dimensional manifolds have a same diffeomorphic type, but they are pairwise different as Weinstein manifolds.
-
•
The Milnor fiber of -singularity and .
-
•
The Milnor fibers of and -singularities.
-
•
The Milnor fiber of -singularity and .
-
•
The Milnor fiber of -singularity and .
-
•
The Milnor fibers of and -singularities.
The decomposability of wrapped Fukaya categories cannot prove Theorems 1.3 and 1.4 completely. Instead, we compare the symplectic cohomologies of the given Weinstein manifolds.
Remark 1.5.
We would like to point out that one can generalize the results of this paper easily. For example, by using the same technique, one can prove the following: Let be a Weinstein manifold obtained by plumbing multiple along a tree such that
-
•
is odd, and
-
•
has the Dynkin tree of type as a subtree.
Then, there are two Weinstein manifolds and such that
-
•
the end connected sum is exotic to ,
-
•
one can break the tree into two sub trees and so that is a plumbing of along a tree for .
We omit these generalizations for the sake of conciseness.
1.3 Acknowledgment
The authors appreciate Youngjin Bae and Hanwool Bae for the helpful discussions. We also appreciate Yanki Lekili for his comments on the draft of the present paper.
The first (resp. last) named author has been supported by a KIAS individual grant (MG079401) (resp. the Institute for Basic Science (IBS-R003-D1)).
2 Preliminaries
Sections 2.1 and 2.2 contain the definition of Lefschetz fibrations, the definition of abstract Lefschetz fibrations, the notion of stabilization, and the notion of Hurwitz moves.
We note that in the paper, we give only a brief explanation. However, in the literature, there are lots of references. Some of the references are [GP17], [Cou, Section 2], [BEE12, Section 8], and [Sei08]. We also refer the reader to [CM19, Section 3.1], especially for the notion of stabilization of Lefschetz fibrations.
In the last subsection of Section 2, the definition of end connected sum is given.
2.1 Lefschetz fibration
Our main tool in the current paper is the notion of Lefschetz fibration, which is defined as follows:
Definition 2.1.
Let be a finite type Liouville manifold. See [CE12] for the definition of Liouville manifold (of finite type). A Lefschetz fibration on is a map satisfying the following properties:
-
•
(Lefschetz type critical points.) There are only finitely many points where is not surjective, and for any such critical point , there exist complex Darboux coordinates centered at so that . Moreover, there is at most one critical point in each fiber of .
-
•
(Symplectic fiber.) Away from the critical points, is non-degenerate on the fibers of .
-
•
(Triviality near the horizontal boundary.) There exists a contact manifold , an open set such that is proper and a codimension zero embedding such that and where is the symplectization, , and means the polar coordinates of .
-
•
(Transversality to the vertical boundary.) There exists such that the Liouville vector filed lifts the vector field near the region .
We note that it would be more precise to use the term ‘Liouville Lefschetz fibration’ in Definition 2.1 because there are Lefschetz fibrations of other types. However, in this paper, this is the only type which we considered. Thus, we omit the adjective for convenience.
Let be a Lefschetz fibration defined on a Weinstein manifold . Then, it is well-known that induces a decomposition of into two parts, one is a subcritical part, and the other is a collection of Weinstein critical handles. See [Sei08, Lemma 16.9] or [BEE12, Section 8]. We note that the subcritical part is given as a product of the regular fiber of and . In order to attaching critical handles to the subcritical part, one needs a collection of Legendrian attaching spheres which one can obtain from
-
•
the cyclic order of the critical values of , and
-
•
the vanishing cycles corresponding to the critical values of .
Conversely, if one has such decomposition data of , then, one can construct a Lefschetz fibration defined on . We give a brief explanation after Definition 2.2.
Based on the converse direction, one has an alternative definition for Definition 2.1.
Definition 2.2.
An abstract (Weinstein) Lefschetz fibration is a tuple
consisting of a Weinstein domain (the “central fiber”) along with a finite sequence of exact parameterized Lagrangian spheres (the “vanishing cycles”).
Definitions 2.1 and 2.2 are interchangeable. In the rest of Section 2.1, we explain the reason for that. For more details, we refer the reader to [Sei08] and [BEE12, Section 8].
Let be a given abstract Weinstein Lefschetz fibration. Then, one can construct a Weinstein domain as follows: First, we consider the product of and . We remark that the product is not a Weinstein domain, because the product is a manifold with corner. To be more precise, we should consider a Weinstein manifold which is the product of symplectic completions of and . Then, we should consider a corresponding Weinstein domain, i.e., a Weinstein domain whose symplectic completion is the product Weinstein manifold. However, for convenience, we consider .
The vertical boundary has a natural contact structure. Moreover, the vanishing cycle can be lifted to a Legendrian near . We do not give the lifting procedure explicitly, but it is easily achieved by using the product structure on . We note that by assuming that the disk has a sufficiently large radius, one could assume that the projection images of onto the factor are disjoint to each other.
Finally, one could attach critical Weinstein handles along for all . Then, the completion of the resulting Weinstein domain admits a Lefschetz fibration satisfying that the regular fiber is , and that there are exactly singular values.
2.2 Equivalent abstract Lefschetz fibrations
By [GP17], it is known that every Weinstein manifold admits a Lefschetz fibration. Also, it is known that, for a Weinstein manifold, there are infinitely many different Lefschetz fibrations. Then, it would be natural to ask that if one has two different Lefschetz fibrations of the same total space, is there any relation between these two Lefschetz fibrations? The above questions is partially answered as follows.
Let
be a given abstract Lefschetz fibration. It is known that there are four moves producing another abstract Lefschetz fibration from so that the total spaces of two abstract Lefschetz fibrations are Weinstein homotopic. We list the four moves here.
-
•
Deformation means a simultaneous Weinstein deformation of and exact Lagrangian isotopy of .
-
•
Cyclic permutation is to replace the ordered collection with . In other words,
The equivalence means that their total spaces are equivalent.
-
•
Hurwitz moves. Let denote the symplectic Dehn twist around . Hurwitz move is to replace with either or , i.e.,
-
•
Stabilization. Let , or equivalently, the total space is of dimension . For a parameterized Lagrangian disk with Legendrian boundary such that where is the Liouville -form, replace with , obtained by attaching a dimensional Weinstein -handle to along , and replace with , where is obtained by gluing together and the core of the handle. In other words,
Remark 2.3.
It is natural to ask that the above four moves are enough to connect any two Lefschetz fibrations of the same total space. As far as we know, this question is still open.
2.3 End connected sum
The goal of Section 2.3 is to define the notion of end connected sum. Since the notion of end connected sum is defined as an attachment of index Weinstein handle, we start the current subsection by reviewing the notion of Weinstein handle attachment.
In [Wei91], Weinstein explained how to attach a Weinstein handle to a Weinstein manifold. A rough explanation on that is the following: In order to a Weinstein handle to a Weinstein manifold , one needs
-
•
an isotropic embedding of the attaching sphere of onto the asymptotic boundary , and
-
•
a conformal symplectic normal bundle of where is the isotropic image of the above embedding.
We note that a conformal symplectic normal bundle of means a conformal symplectic structure on the bundle where means the “symplectic orthogonal operation” in the tangent bundle of . We refer the reader to [Wei91] for more details.
From the above arguments, one can induce Lemma 2.4.
Lemma 2.4.
Let be a connected Weinstein manifold of dimension . Then, the attaching of Weinstein handle to is unique, up to Weinstein homotopy.
Proof.
Let be an index Weinstein handle of dimension . Then, the attaching sphere of is homeomorphic to , i.e., two points. Thus, the embedding of the attaching sphere is to choose two points from the asymptotic boundary of . This implies that any two isotropic embedding of the attaching sphere are isotopic if is connected.
Since is a Weinstein manifold, admits a Weinstein handle decomposition. Thus, is given as a union of dimensional handles whose indices are . This implies that does not contain an index handle, should be connected.
To be more precise, let be the contact structure given on . Also, let be the contact one form on induced from the Weinstein structure of . If the embedding of the attaching sphere is , then a conformal symplectic bundle on the attaching sphere would be a choice of symplectic structure on .
Since is a point, is the zero vector space. Thus, . Then, defines a unique symplectic structure on , it completes the proof. ∎
Let and be connected Weinstein manifolds of the same dimension . Then, it is easy to check that there is a unique way, up to Weinstein homotopic, to construct a connected Weinstein manifold by attaching a Weinstein handle of index to , where means the disjoint union.
Definition 2.5.
Let and be the connected Weinstein manifolds of dimension . Then, the end connected sum of and is the connected Weinstein manifold obtained by attaching an index Weinstein handle to . Let denote the end connected sum of and .
3 Construction of an exotic pair ()
3.1 Construction
As mentioned in Section 2, from a Weinstein manifold and a cyclically ordered finite collection of exact Lagrangian spheres in , one could construct another Weinstein manifold equipped with a Lefschetz fibration such that
-
•
the regular fiber of is , and
-
•
the number of critical values of is the same as the number of exact Lagrangians in the cyclically ordered collection.
We will construct and for by using the above method.
First, we define a notation.
Definition 3.1.
Let be a type plumbing of , i.e., plumbings of copies of , whose plumbing graph is the Dynkin diagram of type.
It is well-known that admits a Lefschetz fibration such that
-
•
the regular fiber of is ,
-
•
has singular values, and
-
•
the corresponding vanishing cycles for singular values are the zero section of the regular fiber .
The existence, and the properties of are well-known in the literature. We refer the reader to [KS02] and [Wu14] for more details.
Remark 3.2.
We note that if , then exists. This is the reason why we consider the case of .
Let . Since is a plumbing of two , one can find two exact Lagrangian spheres which are the zero sections of two . Let those Lagrangian spheres be denoted by and . Then, and are given by curves connecting two singular values of . Figure 1 is the picture of the base of .
Since and are Lagrangian spheres in , there are generalized Dehn twists and along and .
With above arguments, one can define and as follows:
Definition 3.3.
-
1.
Let be a Weinstein manifold which is the total space of the abstract Lefschetz fibration
(3.1) The total number of in Equation (3.1) is , and the total number of exact Lagrangians in the cyclically ordered collection is .
-
2.
Let be a Weinstein manifold which is the total space of the abstract Lefschetz fibration
(3.2) The total number of in Equation (3.2) is , and the total number of exact Lagrangians in the cyclically ordered collection is .
3.2 Wrapped Fukaya category of
In Section 3.1, we constructed and as total spaces of abstract Lefschetz fibrations. Before going further, we investigate the wrapped Fukaya categories of and in Sections 3.2 and 3.3. By studying those wrapped Fukaya categories, we can have a hint for proving that and are not equivalent as Weinstein manifolds.
We start with Lemma 3.4.
Lemma 3.4.
The Weinstein manifold defined in Definition 3.3 is Weinstein homotopic to .
Proof.
First, we can operate Hurwitz moves times, which are moving to left in Equation (3.1). Since, on the left side of , there are many , we obtain the following:
Second, we can move the first to the right end of the collection of exact Lagrangians. This is because the collection is cyclically ordered. Then, we have
After that, we move to the left once. It concludes that
By using the property of , one can easily check that
Thus,
We note that the number of in the above abstract Lefschetz fibration is .
From the definition of the operation ‘stabilization’, it is easy to show that the right side of the above equation is obtained by stabilizing the following abstract Lefschetz fibration
which is the well-known abstract Lefschetz fibration of . This proves that . ∎
Lemma 3.5 follows naturally.
Lemma 3.5.
We have
where means the wrapped Fukaya category of .
Remark 3.6.
For the definition of wrapped Fukaya categories we used in the present paper, see [GPS20]. Roughly, wrapped Fukaya categories can be defined as -categories whose objects are exact Lagrangians with cylindrical ends. In order to be precise, we note that the notion of equivalence between wrapped Fukaya categories in this paper is the pretriangulated equivalence, i.e., quasi-equivalence after taking pretriangulated closures. In the rest of the paper, we simply say ‘equivalence’ rather than ‘pretriangulated equivalence’.
3.3 Wrapped Fukaya category of
In the next subsection, we prove the following Lemma.
Lemma 3.7.
The wrapped Fukaya category of is equivalent to the coproduct of the wrapped Fukaya categories of and , i.e.,
Because of the length of the proof, we give a sketch of the proof in the present subsection.
In order to prove Lemma 3.7, we construct a Weinstein sectorial covering of a Weinstein domain which is Weinstein homotopic to . In the construction of the Weinstein sectorial covering, we need to use the notions of Weinstein homotopy, symplectic completion, Lagrangian skeleta, etc. The construction would take the most part of Section 3.3.
The constructed Weinstein sectorial covering satisfies that and . The notion of equivalences here is Weinstein homotopic. Also, after a proper modification of the Liouville structures, the intersection is equivalent to whose wrapped Fukaya category vanishes.
3.4 Proof of Lemma 3.7
We note that in the current subsection, we modify Weinstein structures via Weinstein homotopies. For the notion of Weinstein homotopies, see [CE12]. Also, we refer the reader to [Sta18].
Before studying , we recall the construction of . By Equation (3.2), is obtained by attaching Weinstein critical handles to the subcritical part. Let denote the critical handles which are attached to the subcritical part of such that
-
•
are attached along the Legendrians induced from , and
-
•
and are attached along the Legendrians induced from .
We discuss a decomposition of the subcritical part of . Since the subcritical part is given as a product of the fiber and a disk , we will decompose and .
Remark 3.8.
To be more precise, we should say that the subcritical part of is a Weinstein domain whose symplectic completion is . Then, is the symplectic completion of a Weinstein domain which is obtained by attaching to the subcritical part. However, for convenience, we omit the words ‘up to symplectic completion’ in the current paper.
In order to decompose , we recall that is a union of and , as mentioned in Section 3.1. This induces that admits a Weinstein handle decomposition which consists of one -dimensional index -handle and two -dimensional index -handles. Moreover, the union of the unique -handle and one -handle is , and the union of the -handle and the other -handle is . We use to denote the index -handle, the index -handle making , the index -handle making of the fiber , respectively.
For a decomposition of , we consider a Weinstein sectorial covering of . In order to do this, we start from a Weinstein handle decomposition of , consisting of three index -handles and two index -handles. Figure 2, describes the handle decomposition.
From the Weinstein handle decomposition, the centers of two -handles have curves as their unstable manifolds with respect to the Liouville flow. Then, those two curves divide into three parts. This decomposition of is given in Figure 2, . Let be the piece of the decomposition such that the boundary of contains the both unstable manifolds. Also, let denote the other two pieces.
Based on the above arguments, the subcritical part admits a decomposition into a union of nine pieces
One can observe that and are equivalent to -dimensional index -Weinstein handles, up to Weinstein homotopy. In other words, we can construct the subcritical part by attaching two Weinstein handles and to , instead of attaching and . The notation (resp. ) denotes the Weinstein handle replacing (resp. ).
We note that by attaching and to , we obtain a Weinstein domain which does not admit a product Liouville structure. In other words, by replacing (resp. ) with (resp. ), we modify the Liouville structure on them, and the modified structures do not respect the product structure.
To summarize, we decompose the subcritical part as a union of the following five pieces,
(3.3) |
We note that the decomposition is not a Weinstein handle decomposition. This is because the first three pieces are not Weinstein handles.
From the above descriptions, we have a decomposition of into the union of five pieces in (3.3) and for . For the critical handles, without loss of generality, one can assume that the critical handles (resp. ) are attached to the (resp. ). Figure 3 describes it.
For the later use, we modify the Weinstein domain which we obtained by attaching five pieces in (3.3). We observe that the attaching regions of and are contained in . However, we can modify it so that the attaching region of (resp. ) is contained in (resp. ). We note that the modification does not change the symplectic completion of the resulting Weinstein domain, up to Weinstein homotopy.
By the above modification, one can observe that the resulting Weinstein domain, even after attaching critical handles, does not change up to Weinstein homotopy. However, the Lagrangian skeleton changes. Figure 4 describes the change on Lagrangian skeleta.
Thanks to the modification, we can set
(3.4) | |||
(3.5) | |||
(3.6) |
Then, is equivalent to the original up to Weinstein homotopy. We note that Equations (3.5) and (3.6) make sense since the attaching region of (resp. ) is contained in (resp. ).
This induces that
(3.7) |
Moreover, since is a Weinstein sectorial covering of , the result of [GPS18b] and Equation (3.7) conclude that
In order to complete the proof, we consider the wrapped Fukaya categories of Weinstein sectors , and .
For , we observe that
Moreover, from the definition of , one can easily check that as a Weinstein sector, is equivalent to . Thus, we have
Since is quasi-equivalent to the zero category, so is the wrapped Fukaya category of the intersection.
For (resp. ), we consider the convex completion of the Weinstein sector (resp. ). For the notion of convex completion, see [GPS20, Lemma 2.32].
By taking the convex completion of (resp. ), one obtains
(3.8) | |||
(3.9) |
together with a stop. Since the boundary of the Liouville sector (resp. ) is given by a product of and an dimensional curve, the stop is a Legendrian disk. We note that the dimensional curve above is given in Figure 2, as a part of the boundary of (resp. ), contained in the interior disk.
Since the stop is a disk, we have the followings.
Thus, we consider (resp. ) instead of (resp. ).
In order to study (resp. ), we consider the construction of (resp. ). Equation (3.8) (resp. Equation (3.9)) means that (resp. ) is obtained by attaching and critical handles to a Weinstein domain
Since , and since is a -handle, it is equivalent to attach and to the dimensional index -handle.
From the above argument, it is easy to check that is a Weinstein domain which is equivalent to the total space of an abstract Lefschetz fibration
where the number of above is . It is known that the total space of the abstract Lefschetz fibration is . Thus, it concludes that
Similarly, is equivalent to the total space of
which is .
It concludes that
i.e.,
∎
Remark 3.9.
We remark the following four facts before going further.
-
1.
From the proof of Lemma 3.7, one can observe that is constructed from and , by a ‘gluing construction’ using a hypersurface. The gluing construction is described in [Eli18, Section 3.1]. Also, see [Avd12]. The hypersurface which used to glue and is a -dimensional disk, which corresponds to in .
In other words, is constructed using the notion of end connected sum. Then, one can observe that is obtained by taking an end connected sum of and . The notion of end connected sum is used in [McL09], in order to construct exotic symplectic structures on for .
-
2.
One can observe that both of and can be constructed from . More precisely, since is equivalent to , is obtained by plumbing to . From the view point of Weinstein handle decomposition, plumbing is equivalent to add a critical Weinstein handle. Similarly, one can observe that the construction of in Definition 3.3 is equivalent to add a critical Weinstein handle and a canceling pair of Weinstein handles of index and to .
One also can observe that the construction of in Definition 3.3 is also equivalent to add a critical Weinstein handle and a canceling pair of Weinstein handles of index and to , in a different way from the case of . The argument in Section 3.4 explains a way to convert the canceling pair in as another canceling pair of Weinstein handles of index and . Then, the added critical handle and the index handle in the canceling pair construct , and the index handle in the canceling pair becomes the index handle ‘gluing’ and .
-
3.
The above arguments give a way to construct a Lefschetz fibration of where is obtained by a end connected sum of Weinstein manifolds. For example, if is the end connected sum of three , then is a total space of the abstract Lefschetz fibration
where is the plumbing of three cotangent bundles of spheres , , and .
-
4.
One can observe that the same technique, i.e., to use Lefschetz fibrations in order to construct Weinstein sectorial coverings, works if we calculate the wrapped Fukaya category of a total space of a Lefschetz fibration. However, taking the homotopy colimit for a general Lefschetz fibration would not as simple as the case of . For example, one can compute by using the same method. However, for taking the homotopy colimit for , we should care the plumbing sector which is described in [GPS18a, Section 6.2]. Then, the homotopy colimit formula in [KL21] will give the resulting wrapped Fukaya category.
4 Different symplectic structures on and
In Section 3, we constructed and . Also, we proved Lemmas 3.5 and 3.7. By using them, we prove that and have different wrapped Fukaya categories in Section 4.
In order to distinguish their symplectic structures, we focus on the result of Lemma 3.7. Lemma 3.7 says that can be written as a coproduct of two nontrivial categories, up to equivalence. On the other hand, Lemma 3.5 says that . Thus, if cannot be written as a coproduct in a nontrivial way, i.e., a coproduct of itself and something equivalent to the empty category. This completes the proof.
In Section 4.1, we prove Lemma 4.4 describing a property of coproduct, and in Section 4.2, we show that cannot be a coproduct by using Lemma 4.4.
4.1 Decomposability of dg categories
Let be the coefficient ring. For a -linear dg category , we write for the morphism complex where are objects in , and we write for its cohomology. Let be the dg category of (one-sided) twisted complexes in . For the above definitions, see [BK90].
Definition 4.1.
Let be a dg category. We say that the set of objects of is a generating set of if
up to quasi-equivalence, where is the full dg subcategory of with the set objects , and is an indexing set.
Proposition 4.2.
We have the quasi-equivalence
where the objects of are the direct sums of the objects of and .
Proof.
If , there exists a twisted complex for consisting of the objects of and . Since there is no nonzero morphisms between and , we can rearrange the twisted complex in such a way that all the objects of are on the left, and all the objects of are on the right. This means that
for some and . Similarly, the converse is true. ∎
Definition 4.3.
Let be a dg category. We call decomposable if is pretriangulated equivalent to a coproduct of nonempty dg categories, i.e.
up to quasi-equivalence for some nonempty dg categories and . Otherwise, is called indecomposable. We call decomposable with -components, if
up to quasi-equivalence for some nonempty dg categories such that is indecomposable for all . This is well-defined.
Lemma 4.4.
Let be a dg category. Assume that is a generating set for satisfying
for all , and
for all . Then is indecomposable.
Proof.
We will prove the contrapositive. Assume is decomposable, and
for all . We will show that for some .
Since is decomposable, there exist nonempty dg categories and such that
up to quasi-equivalence. For all , since , we have
for some and . We note that
for any .
Case 1: Assume there exists such that and . Then,
Since and , we have
Hence, we get
and consequently,
which is what we wanted to show.
Case 2: Assume Case 1 is not true. Then either for all or for all . If not, there exist with such that
and
This implies that
which contradicts with the assumption on at the start of the proof. Thus, without loss of generality, we can assume for all . Then for all . Since is nonempty, there exists such that . However,
for all . This means that , since generates and . This is a contradiction. Hence, Case 1 holds. ∎
4.2 Distinguishing wrapped Fukaya categories of and
We prove the following proposition in Section 4.2.
Proposition 4.6.
The Weinstein manifolds and are not exact deformation equivalent.
We prove Proposition 4.6 by using Lemma 4.4. More precisely, we will prove that is not decomposable by Lemma 4.4. In order to do this, we would like to point out that Lemma 4.4 works when is an -category. The reason is as follows: If is an -category, we can consider the -Yoneda embedding
where is the dg category of -modules over . See [Sei08]. Hence, is quasi-equivalent to the dg category . Consequently, Lemma 4.4 applies on , and hence on since is the same in both categories for any objects and .
Proof of Proposition 4.6.
We note that admits a Lefschetz fibration such that
-
•
the regular fiber is , and
-
•
the number of singular values are .
This Lefschetz fibration is described in Section 3.2.
From the above Lefschetz fibration, we can choose a set of Lefschetz thimbles . Then, it is well-known that the set of Lefschetz thimbles is a generating set of .
In the current literature, is already computed for any . We refer the reader to [BK21] or [LU20]. Since our case satisfies the condition of [BK21, Proposition 1.5], for all ,
where the base ring is .
This concludes that cannot be a coproduct of two non-empty categories. However, Lemma 3.7 says that can be written as a coproduct. It concludes that and are not equivalent. Thus, and are not equivalent as Weinstein manifolds. ∎
Lemma 4.7.
Let be Lefschetz thimbles defined above. Then,
Proof.
We use the index-positivity argument described in [BK21], in order to show that . Let us use a following explicit description of , and of the Lefschetz fibration on
The critical values of the fibration are where . Let
denote rays in , emanating from each critical points. The Lagrangians we are considering are thimbles over each ;
We note that is defined geometrically, by using the relation of Hamiltonian chords from to .
A sequence of admissible Hamiltonians we are using is of the form , where is a smooth function such that
-
•
only depends on ,
-
•
is -small on a compact region, and
-
•
when .
Hamiltonian chords from to of comes in families because the set of vanishing cycles of the thimbles have rotational symmetries. Whenever we have a Hamiltonian chords of from to on the base, we get an -family of corresponding chords of from to . Using a Morse-Bott type perturbation, we get two non-degenerate Hamiltonian chords coming from Morse homology of a sphere . We label the generators of the Floer complex as
where a non-negative number indicates the winding number of the corresponding base chord, and where min/max denotes the generators corresponding to a fundamental/point class of . In this setup, we have
when and . See Proposition 1.2 in [BK21] for more details. In particular, we conclude that there is no non-trivial Floer differential hitting . (In fact, the spectral sequence degenerates at this page whenever .) This implies that . ∎
5 Diffeomorphism between and
Proposition 5.1.
If , the Weinstein manifolds and are diffeomorphic. If is even, then and are diffeomorphic.
We note that the main idea of Proposition 5.1 is given in [May09, Proposition 4.0.1] and [MS15]. For more details, we refer the reader to them.
Proof of Proposition 5.1.
In order to prove this, let us recall the construction of and .
In Section 3.1, and are given as total spaces of abstract Lefschetz fibrations in (3.1) and (3.2). Section 3.3 gives more detailed construction of , which is a Weinstein domain obtained by attaching critical handles to the subcritical part . Similarly, is equivalent to the Weinstein domain obtained by attaching critical handles to the same subcritical part . For more detail on the constructions, we refer the reader to [Sei08, Lemma 16.9], [BEE12, Section 8]. See also [GS99].
From (3.1) and (3.2), without loss of generality, one could assume that and are attached along the same Legendrian of , if . Thus, in order to prove Proposition 5.1, it is enough to consider the attaching Legendrian spheres of and .
The attaching Legendrian spheres of and are induced by Lagrangian spheres and . More precisely, these Lagrangian spheres induce Legendrian spheres with the canonical formal Legendrian structures as defined in [Mur12]. We note that when one attaches a critical handle along a Legendrian sphere, the formal Legendrian structure on the Legendrian sphere determines a diffeomorphism class of the resulting space. This is because the formal Legendrian structure affects on the framing of the attached critical handle.
In order to prove Proposition 5.1, we need to compare the Legendrian spheres induced from and together with their formal Legendrian structures for , and to compare the Legendrian spheres induced from and together with their formal Legendrian structures for even .
Construction of smooth isotopy: First, we construct a smooth Lagrangian isotopy between and inside for any even . Then, it induces an isotopy between and , naturally. This construction is originally given in [May09, Section 5].
By using the Lefschetz fibration on , (resp. ) is given as a union of vanishing cycles in fibers of along a curve in . Let and be the curves corresponding to and , respectively. See Figure 5.
We note that the regular fiber of is , and every vanishing cycle of is the zero section. From the structure of , when is even, we can lift the zero section in a Lagrangian isotopic way, but not Hamiltonian isotopic way. This is because admits a non-vanishing vector field. The lift for the case of is given in Figure 6, . We note that we can control how much the vanishing cycle is lifted.
After that, we consider the union of lifted vanishing cycles inside . The union runs over . The union is over the fibers projected on . How much the vanishing cycle in lifted is determined by the distance between and . If a point on the base is close enough to , then we do not lift the vanishing cycle in . And, if a point is not close to , then we lift the vanishing cycles in . If is further from than , then we lift more the vanishing cycle in than that in . Then, this union of lifted vanishing cycles is a Lagrangian sphere which is isotopic to . Figure 6, describes the case of .
On the base of , two curves and are isotopic to each other. Along the isotopy of curves connecting and , we can construct a smooth family of Lagrangian spheres, whose each member is a union of lifted vanishing cycles. Also, we consider the lifted vanishing cycles, the family of Lagrangian spheres does not touch the singular point. Then, this family gives us an isotopy connecting and .
Formal Legendrian structures: Notice that for a given Legendrian sphere of dimension , the set of formal Legendrian structures is given by , where stands for the Stiefel manifold. For more detail, see [Mur12, Appendix].
For the case of , [Mur12, Proposition A.4] says that all formal embeddings of are formally Legendrian isotopic, or equivalently, is trivial. Thus, the isotopy between and which is constructed above induces a smooth isotopy connecting the induced Legendrian spheres with formal Legendrian structures. This prove the first half of Proposition 5.1.
For the case of even , [MS15] shows that the constructed isotopy between and induces a smooth isotopy connecting the induced Legendrian spheres with formal Legendrian structures. In order to show this, the authors recalled that
(5.1) |
If we measure the formal Legendrian structures given on and , the difference must lies in the group by Equation (5.1). Therefore we might get a non-trivial element. However, when we consider and , the difference is the trivial element in because we take the same operation twice to cancel out the difference inside . This completes the proof. ∎
We are now ready to prove Theorem 5.2.
Theorem 5.2.
Let and be symplectic manifolds constructed in Section 3. Then, for (resp. even ) the pair (resp. ) satisfies the followings:
-
•
and (resp. and ) are diffeomorphic,
-
•
and are not exact deformation equivalent, and
-
•
and are not vanishing.
Remark 5.3.
Before going further, we remark that in Section 7, we extend the exotic pair to diffeomorphic families of different Weinstein manifolds. However, in the next section, we use different notation (resp. ) in order to denote (resp. ). Since Proposition 5.1 is a part of Lemmas 7.3 and 7.4, it would seem more reasonable to use and , instead of and . Moreover, instead of , we can directly say that it is the Milnor fiber of -singularity, as proven in Lemma 3.4. In this remark, we would like to clarify the reason of using the notations and .
We would like to point out that and are different as Weinstein manifolds since has a decomposable Fukaya category, but does not. Similarly, if another Weinstein manifold has non-decomposable Fukaya category, then and are different. Based on this, for example, we can expect that the arguments in Sections 3 – 5 distinguish and the Milnor fiber of -type as Weinstein manifolds even if they are diffeomorphic to each other. Actually, this corresponds to the fourth families in Theorems 1.3 and 1.4. Moreover, we can consider to replace with another plumbing space in the same way.
By using a letter which is the conventional letter for a variable, we would like to implicitly mention that one can choose different space for the position of in the exotic pair, i.e., .
6 The first generalization
In Sections 3 – 5, we constructed exotic pairs of Weinstein manifolds. Section 6 extends the construction to families of diffeomorphic, but different Weinstein manifolds.
In order to prepare the extension, we introduce notation first. Let be the type plumbing of as defined in Definition 3.1. Then, there are Lagrangian spheres which are zero sections of . Let denote the Lagrangian spheres so that and intersect at one point. Since is obtained by plumbing a cotangent bundle of to , the following relations make sense by abuses of notation.
For a Lagrangian sphere , there is a Dehn twist along . Let denote the Dehn twist along .
With the above notation, one can define Definition 6.1.
Definition 6.1.
For natural numbers , let denote the total space of the following abstract Lefschetz fibration:
(6.1) |
where the number of in (6.1) is .
It is easy to check that is equivalent to as a Weinstein manifold by the existence of a Lefschetz fibration given below of Definition 3.1. Similarly, is equivalent to .
Lemma 6.2.
-
1.
If , then two Weinstein manifolds
are diffeomorphic to each other.
-
2.
If is even, then two Weinstein manifolds
are diffeomorphic to each other.
Proof.
We prove the second case of Lemma 6.2. The first case can be proven in the same way.
By definition, is the total space of
Then, by the stabilization given in Section 2.2, one can modify the above and can obtain
By a Hurwitz move, one can move to the right, i.e., one has
(6.2) |
We note that in (6.2), the number of is .
One can move the last to the right by a Hurewitz move. Then, becomes
One can easily check the equality by using the Lefschetz fibration of , which is given right below of Definition 3.1. Then, as a result of the Hurewitz move, one obtains
Similarly, one can operate the similar Hurwitz move times. Then, it gives
(6.3) |
We note that in (6.3), the numbers of and are and , respectively.
After that, one can move to the left by operating Hurwitz moves times. Then, one obtains
(6.4) |
Theorem 6.3.
-
1.
If , the following family
(6.6) are diffeomorphic families of pairwise different Weinstein manifolds.
-
2.
If is even, the following family
(6.7) are diffeomorphic families of pairwise different Weinstein manifolds.
Proof.
By Lemma 6.2 induces that the families in (6.6) and (6.7) are diffeomorphic families. Thus, it is enough to show that the members of the families are pairwise different as Weinstein manifolds.
We note that by definition, is a Weinstein manifold obtained by taking the end connected sum of , see the third item in Remark 3.9. This induces that the wrapped Fukaya category of can be written as a coproduct of nontrivial categories.
7 The second generalization
In Sections 3 – 5, we consider an exotic pair of Weinstein manifolds (resp. ) for and for (resp. even ). We distinguish them as Weinstein manifolds by using their wrapped Fukaya categories.
In the current section, we extend the pairs to diffeomorphic families of different Weinstein manifolds. The diffeomorphic families are listed at the end of Section 7.1. We prove that the given families are diffeomorphic by using Lefschetz fibrations as similar to Section 5. However, we prove that they are different as Weinstein manifolds by using their symplectic cohomologies, which is a different method from the previous case.
7.1 Construction of diffeomorphic families
In Section 7.1, we give diffeomorphic families of Weinstein manifolds. In order to do this, we need Definition 7.1.
Definition 7.1.
-
1.
For any , and any , let denote the tree which is given in Figure 7.
Figure 7: Tree . -
2.
For any tree , let denote the Weinstein manifold obtained by plumbing along the plumbing pattern .
- 3.
Remark 7.2.
We note that .
Lemma 7.3.
If , then, and are diffeomorphic to each other. If is even, then and are diffeomorphic to each other.
Proof.
In order to prove Lemma 7.3, we would like to use a Lefschetz fibration on . We note that [Lee21, Theorem 11.3] gives an algorithm producing a Lefschetz fibration for a plumbing space for any . Thus, by applying [Lee21, Theorem 11.3] to , we can produce a desired Lefschetz fibration whose total space is . The resulting Lefschetz fibration is
(7.2) |
such that
In the current paper, we omit a detailed proof for the statement that the total space of (7.2) is . However, we prove the statement for a special case. The special case we consider is .
What we want to show is that the total space of
is . Figure 8 describes the base of the above Lefschetz fibration.
For convenience, let denote the total space of the above abstract Lefschetz fibration.
We consider two submanifolds by using the Lefschetz fibration on . In order to define , we consider the inverse image of the interior of the red circle in Figure 8 under the Lefschetz fibration. Roughly speaking, when we consider the Weinstein handle decomposition of corresponding to the abstract Lefschetz fibration, is a Weinstein domain obtained by ‘deleting’ a critical handle from . Moreover, since could be seen as a total space of an abstract Lefschetz fibration
is equivalent to the Milnor fiber of -type. For the detail, see the proof of Lemma 3.4.
Before defining , we note that the fiber can be seen as a union of two , or and with the notation in Section 3. Then, we define as the total space of the sub-fibration, whose fiber is , and whose base is the interior of the blue circle in Figure 8.
As similar to , one could see as a union of corresponding Weinstein handles. Then, one could conclude that is a Weinstein domain equivalent to .
One could easily check that is a neighborhood of the Lagrangian skeleton of . In order to obtain the Lagrangian skeleton, we use the Liouville structure which the abstract Lefschetz fibration induces. It means that is equivalent to up to symplectic completion. Thus, it is enough to show that is equivalent to .
We note that is obtained by plumbing to where is the Dynkin diagram of -type. Since , it is enough to show that is obtained by plumbing and . One can prove this easily by using the handle movements on and . We omit the details, but instead we give Figure 9 describing a Lefschetz fibration type picture after handle slides. We note that Figure 9 is different from a general Lefschetz fibration picture, especially, the singular values are not lying on a circle whose center is the origin. This is because we slide critical handles ‘onto’ another critical handles. This is not allowed for a general Lefschetz fibration picture.
Lemma 7.4.
If , then, and are diffeomorphic. If is even, and are diffeomorphic.
Proof.
From Lemmas 7.3 and 7.4, we obtain diffeomorphic families such that some of whose members are well-studied plumbing spaces. The list of the families is given in Corollaries 7.5 and 7.6.
Corollary 7.5.
We have a following list of Weinstein manifolds which are diffeomorphic to each other, if their dimensions are , i.e., .
-
•
The Milnor fibers of and -singularities are diffeomorphic, since they are and , respectively.
-
•
The Weinstein manifold and the Milnor fibers of , , and -singularities are diffeomorphic, since the Milnor fibers are , and , respectively.
-
•
The Milnor fibers of and -singularities are diffeomorphic, since they are and , respectively.
-
•
For any , the Weinstein manifold and the Minor fiber of -singularities are diffeomorphic, since the Milnor fiber is .
-
•
For , the Weinstein manifold and the Milnor fibers of -singularities are diffeomorphic, since the Milnor fibers are and , respectively.
Corollary 7.6.
We have a following list of Weinstein manifolds which are diffeomorphic to each other if their dimensions are with even .
-
•
The Milnor fiber of -singularity and are diffeomorphic, since the Milnor fiber is .
-
•
The Milnor fibers of and -singularities are diffeomorphic, since the Milnor fibers are and , respectively.
-
•
The Milnor fiber of -singularity and are diffeomorphic, since the Milnor fiber is .
-
•
The Milnor fiber of -singularity and are diffeomorphic, since the Milnor fiber is .
-
•
The Milnor fibers of and -singularities are diffeomorphic, since the Milnor fibers are and , respectively.
Remark 7.7.
When we consider the Milnor fibers of simple singularities having dimension , then we can check Corollaries 7.5 and 7.6, without using Lefschetz fibrations. Because of the dimension reason, we only need to compare their Euler characteristics and the numbers of boundary components. For example, and have the same Euler characteristics, and the numbers of their boundary components are . Thus, they are diffeomorphic to each other.
Moreover, it is also simple to compare them as Weinstein manifolds if the dimension is two. However, for the case of higher dimension, it would be not simple to compare them as Weinstein manifolds. In the next section, we will show that they are different as Weinstein manifolds in dimension .
We note that we can have bigger diffeomorphic families from Lemmas 7.3 and 7.4, but we only consider smaller families in Corollaries 7.5 and 7.6. This is because, in order to compare them as Weinstein manifolds, we need to compute their symplectic invariant. For the named spaces which are contained in the smaller families, the computations are well-studied, and we would like to use the well-studied computations.
For the other plumbing spaces which are not contained in the smaller families, we are working on comparing them as Weinstein manifolds.
7.2 Distinguishing symplectic cohomologies
In this section, we compare the Weinstein structures of the listed Weinstein manifolds. We note that for this subsection, we do not need to separate the case of and the case of even . Moreover, even for the case of odd , the results in this subsection hold.
Before proving Theorem 1.3, we note that Milnor fibers of ADE-types can be described by Milnor fibers of invertible polynomials. In other words, there is an invertible matrix so that we can write the Milnor fibers as where
If we define its transpose as a Milnor fiber of an invertible polynomial with the exponent matrix , then we obtain the following list of polynomials for Milnor fibers of -types and their transposes.
-
•
-
•
-
•
-
•
-
•
Proposition 7.8.
(Theorem 1.2 [LU20]) Let be a polynomial of -type for . Then,
-
1.
we have an equivalence of categories
The right handed side is a category of -equivariant matrix factorizations. Here is a maximal group of abelian symmetries such that becomes a semi-invariant, i.e.,
-
2.
Also, one can compute using the second isomorphism combined with results of [BFK14].
We do not prove Proposition 7.8, but an interested reader can find a detailed computation in [LU20, Section 5].
Proof of Theorem 1.3..
We would like to point out that members in the families are Milnor fibers of -types and . The symplectic cohomology rings of Milnor fibers of simple singularities have been computed using homological mirror symmetry as mentioned in Proposition 7.8. On the other hand, Lemma 3.7 implies that
We note that is the Milnor fiber of -type. Thus, one can compute symplectic cohomologies of all Weinstein manifolds in Corollaries 7.5 and 7.6.
From [LU20], we can check the following facts.
-
•
Milnor fibers of and -types have different symplectic cohomologies, because the contributions from twisted sectors of each are different from each other.
-
•
A symplectic cohomology carries extra information of weights. For matrix factorizations, weights are coming from the weights of variables of . A mirror -action on a symplectic cohomology ring can be found in [SS12]. The weights of the each generators of are equal or less than their degrees. There is only one family of generators (other then unit) of such that whose weight and degree are the same. Every generator in the family has weight and degree .
-
•
In , there is an element of degree and weight . The element is a product of elements in and whose degrees and weights are . Moreover, there are no such elements inside , or .
References
- [AS10] Mohammed Abouzaid and Paul Seidel, Altering symplectic manifolds by homologous recombination, arXiv preprint arXiv:1007.3281 (2010).
- [Avd12] Russell Avdek, Liouville hypersurfaces and connect sum cobordisms, arXiv preprint arXiv:1204.3145 (2012).
- [BEE12] Frédéric Bourgeois, Tobias Ekholm, and Yasha Eliashberg, Effect of Legendrian surgery, Geom. Topol. 16 (2012), no. 1, 301–389, With an appendix by Sheel Ganatra and Maksim Maydanskiy. MR 2916289
- [BFK14] Matthew Ballard, David Favero, and Ludmil Katzarkov, A category of kernels for equivariant factorizations and its implications for Hodge theory, Publications mathématiques de l’IHÉS 120 (2014), no. 1, 1–111.
- [BK90] A. I. Bondal and M. M. Kapranov, Framed triangulated categories, Mat. Sb. 181 (1990), no. 5, 669–683. MR 1055981
- [BK21] Hanwool Bae and Myeonggi Kwon, A computation of the ring structure in wrapped Floer homology, Math. Z. 299 (2021), no. 1-2, 1155–1196. MR 4311633
- [CE12] Kai Cieliebak and Yakov Eliashberg, From Stein to Weinstein and back, American Mathematical Society Colloquium Publications, vol. 59, American Mathematical Society, Providence, RI, 2012, Symplectic geometry of affine complex manifolds. MR 3012475
- [CM19] Roger Casals and Emmy Murphy, Legendrian fronts for affine varieties, Duke Math. J. 168 (2019), no. 2, 225–323. MR 3909897
- [Cou] Sylvain Courte, Liouville and Weinstein manifolds.
- [Eli18] Yakov Eliashberg, Weinstein manifolds revisited, Modern geometry: a celebration of the work of Simon Donaldson, Proc. Sympos. Pure Math., vol. 99, Amer. Math. Soc., Providence, RI, 2018, pp. 59–82. MR 3838879
- [GP17] Emmanuel Giroux and John Pardon, Existence of Lefschetz fibrations on Stein and Weinstein domains, Geom. Topol. 21 (2017), no. 2, 963–997. MR 3626595
- [GPS18a] Sheel Ganatra, John Pardon, and Vivek Shende, Microlocal Morse theory of wrapped Fukaya categories, arXiv preprint arXiv:1809.08807 (2018).
- [GPS18b] , Sectorial descent for wrapped Fukaya categories, arXiv preprint arXiv:1809.03427 (2018).
- [GPS20] , Covariantly functorial wrapped Floer theory on Liouville sectors, Publ. Math. Inst. Hautes Études Sci. 131 (2020), 73–200. MR 4106794
- [GS99] Robert E. Gompf and András I. Stipsicz, -manifolds and Kirby calculus, Graduate Studies in Mathematics, vol. 20, American Mathematical Society, Providence, RI, 1999. MR 1707327
- [KL21] Dogancan Karabas and Sangjin Lee, Homotopy colimits of semifree dg categories and Fukaya categories of cotangent bundles of lens spaces, arXiv preprint arXiv:2109.03411 (2021).
- [KS02] Mikhail Khovanov and Paul Seidel, Quivers, Floer cohomology, and braid group actions, J. Amer. Math. Soc. 15 (2002), no. 1, 203–271. MR 1862802
- [Lee21] Sangjin Lee, Lefschetz fibrations on cotangent bundles and some plumbings, arXiv preprint arXiv:2105.11200 (2021).
- [LU20] Yanki Lekili and Kazushi Ueda, Homological mirror symmetry for milnor fibers of simple singularities, arXiv preprint arXiv:2004.07374 (2020).
- [May09] Maksim Maydanskiy, Exotic symplectic manifolds from Lefschetz fibrations, ProQuest LLC, Ann Arbor, MI, 2009, Thesis (Ph.D.)–Massachusetts Institute of Technology. MR 2717731
- [McL09] Mark McLean, Lefschetz fibrations and symplectic homology, Geom. Topol. 13 (2009), no. 4, 1877–1944. MR 2497314
- [MS10] Maksim Maydanskiy and Paul Seidel, Lefschetz fibrations and exotic symplectic structures on cotangent bundles of spheres, J. Topol. 3 (2010), no. 1, 157–180. MR 2608480
- [MS15] , Corrigendum: Lefschetz fibrations and exotic symplectic structures on cotangent bundles of spheres [ MR2608480], J. Topol. 8 (2015), no. 3, 884–886. MR 3394320
- [Mur12] Emmy Murphy, Loose Legendrian Embeddings in High Dimensional Contact Manifolds, ProQuest LLC, Ann Arbor, MI, 2012, Thesis (Ph.D.)–Stanford University. MR 4172336
- [Sei08] Paul Seidel, Fukaya categories and Picard-Lefschetz theory, Zurich Lectures in Advanced Mathematics, European Mathematical Society (EMS), Zürich, 2008. MR 2441780
- [SS12] Paul Seidel and Jake P Solomon, Symplectic cohomology and q-intersection numbers, Geometric and Functional Analysis 22 (2012), no. 2, 443–477.
- [Sta18] Laura Starkston, Arboreal singularities in Weinstein skeleta, Selecta Math. (N.S.) 24 (2018), no. 5, 4105–4140. MR 3874691
- [Wei91] Alan Weinstein, Contact surgery and symplectic handlebodies, Hokkaido Math. J. 20 (1991), no. 2, 241–251. MR 1114405
- [Wu14] Weiwei Wu, Exact Lagrangians in -surface singularities, Math. Ann. 359 (2014), no. 1-2, 153–168. MR 3201896