ScalingStacks

3.1. Two-handles

The paper [25] contains a description of the skein lasagna module for 2-handlebodies (four-manifolds WW made of a 0-handle and some 2-handles), where the link L⊂∂WL\subset\partial W is empty, or at least local (contained in a 33-ball). The description is in terms of the Khovanov-Rozansky homology of cables of the attaching link KK.

In this subsection we extend that description to the case where we attach 2-handles to any four-manifold WW, to obtain a new manifold W′W^{\prime}. Moreover, we do not impose any restriction on the link L⊂∂WL\subset\partial W. The formula is very similar to that in [25]. The role of the Khovanov-Rozansky homology KhRN\operatorname{KhR}_{N} will be played by the skein lasagna module 𝒮0N​(W,−)\mathcal{S}_{0}^{N}(W;-), which can be thought of as a link homology for links in the boundary of WW. (When W=B4W=B^{4}, we have 𝒮0N​(W,L)=KhRN⁡(L)\mathcal{S}_{0}^{N}(W;L)=\operatorname{KhR}_{N}(L).)

Let K1,…,KnK_{1},\dots,K_{n} be the components of the framed link K⊂∂WK\subset\partial W along which the 2-handles are attached. The framing gives diffeomorphisms fif_{i} between tubular neighborhoods ν⁡(Ki)\nu(K_{i}) of each KiK_{i} and S1×D2S^{1}\times D^{2}. Given nn-tuples of nonnegative integers

k−=(k1−,…,kn−),k+=(k1+,…,kn+),k^{-}=(k_{1}^{-},\dots,k_{n}^{-}),\ \ \ k^{+}=(k_{1}^{+},\dots,k_{n}^{+}),

we let K⁡(k−,k+)K(k^{-},k^{+}) denote the framed, oriented cable of KK consisting of ki−k_{i}^{-} negatively oriented parallel strands to KiK_{i} and ki+k_{i}^{+} positively oriented parallel strands. Here, the notion of parallelism for the strands is determined by the framing, that is,

K⁡(k−,k+)=⋃ifi−1​(S1×{x1−,…,xki−−,x1+,…,xki++})K(k^{-},k^{+})=\bigcup_{i}f_{i}^{-1}(S^{1}\times\{x_{1}^{-},\dots,x_{k_{i}^{-}}^{-},x_{1}^{+},\dots,x_{k_{i}^{+}}^{+}\})

for fixed points x1−,…,xki−−,x1+,…,xki++∈D2.x_{1}^{-},\dots,x_{k_{i}^{-}}^{-},x_{1}^{+},\dots,x_{k_{i}^{+}}^{+}\in D^{2}.

After attaching 2-handles to WW along KK, we obtain the manifold W′W^{\prime}. Suppose we are given a framed link L⊂∂W′L\subset\partial W^{\prime}. Generically, we can assume that LL stays away from the attaching regions of the 2-handles, and therefore we can represent it as a link in ∂W\partial W, disjoint from (but possibly linked with) KK. (There are various ways of isotoping LL off of the attaching regions; the results of the calculation will be isomorphic.) We let

K⁡(k−,k+)∪LK(k^{-},k^{+})\cup L

be the union of K⁡(k−,k+)K(k^{-},k^{+}) and LL, where we do the cabling on the components of KK by choosing the tubular neighborhoods of KiK_{i} to be disjoint from LL. (Note that K⁡(k−,k+)∪LK(k^{-},k^{+})\cup L is not a split disjoint union.)

We seek to express the skein lasagna module 𝒮0N​(W′,L)\mathcal{S}_{0}^{N}(W^{\prime};L) in terms of 𝒮0N​(W,K⁡(k−,k+)∪L)\mathcal{S}_{0}^{N}(W;K(k^{-},k^{+})\cup L). To do this, we need to introduce a few more notions.

For each ii, let Bki−,ki+B_{k_{i}^{-},k_{i}^{+}} be the subgroup of the braid group on ki−+ki+k_{i}^{-}+k_{i}^{+} strands that consists of self-diffeomorphisms of D2D^{2} rel boundary (modulo isotopy rel boundary) taking the set {x1−,…,xki−−}\{x_{1}^{-},\dots,x_{k_{i}^{-}}^{-}\} to itself and the set {x1+,…,xki++}\{x_{1}^{+},\dots,x_{k_{i}^{+}}^{+}\} to itself. By taking the product with the identity on S1S^{1}, a braid element b∈Bki−,ki+b\in B_{k_{i}^{-},k_{i}^{+}} induces a self-diffeomorphism of D2×S1D^{2}\times S^{1}, which can be pulled back (via fif_{i}) to a self-diffeomorphism of ν⁡(Ki)\nu(K_{i}). This gives a group action

βi:Bki−,ki+→Aut⁡(𝒮0N​(W,K⁡(k−,k+)∪L)).\beta_{i}:B_{k_{i}^{-},k_{i}^{+}}\to\operatorname{Aut}(\mathcal{S}_{0}^{N}(W;K(k^{-},k^{+})\cup L)).

Let ei∈ℤne_{i}\in{\mathbb{Z}}^{n} denote the it​hi^{th} basis vector. Two strands parallel to KiK_{i}, if they have opposite orientations, co-bound a ribbon band RiR_{i} in S3S^{3}. By pushing RiR_{i} into S3×[0,1]S^{3}\times[0,1] so that it is properly embedded there, and taking the disjoint union with the identity cobordisms on the other strands, we obtain an oriented cobordism (still denoted RiR_{i}) from K⁡(k−,k+)∪LK(k^{-},k^{+})\cup L to K⁡(k−+ei,k++ei)∪LK(k^{-}+e_{i},k^{+}+e_{i})\cup L. For d=0,1,…,N−1d=0,1,\dots,N-1, we can decorate RiR_{i} with dd dots, and obtain a cobordism map

ψi[d]:𝒮0N​(W,K⁡(k−,k+)∪L)→𝒮0N​(W,K⁡(k−+ei,k++ei)∪L),\psi^{[d]}_{i}:\mathcal{S}_{0}^{N}(W;K(k^{-},k^{+})\cup L)\to\mathcal{S}_{0}^{N}(W;K(k^{-}+e_{i},k^{+}+e_{i})\cup L),

which changes the bigrading by (0,2​d)(0,2d).

Next, recall that we have a decomposition (2) for the skein lasagna module 𝒮0N​(W′,L)\mathcal{S}_{0}^{N}(W^{\prime};L), according to homology classes in H2L​(W′,ℤ)H_{2}^{L}(W^{\prime};{\mathbb{Z}}). Let us see how these homology classes are related to the similar ones in WW. Consider the tubular neighborhood ν(K)=∪iν(Ki)\nu(K)=\cup_{i}\nu(K_{i}), which is a union of solid tori. Express W′W^{\prime} as the union

W′=W∪C∪Z,W^{\prime}=W\cup C\cup Z,

where ZZ is the union of the new 2-handles, and C≅ν⁡(K)×[0,1]C\cong\nu(K)\times[0,1] is a connecting cylinder between WW and ZZ. Let also

C′=ν⁡(K)×{0,1}⊂C.C^{\prime}=\nu(K)\times\{0,1\}\subset C.

We identify ν⁡(K)\nu(K) with ν⁡(K)×{0}\nu(K)\times\{0\} and denote ν⁡(K)×{1}\nu(K)\times\{1\} by ∂−Z\partial_{-}Z (part of the boundary ∂Z\partial Z).

The Mayer-Vietoris sequence for W′W^{\prime} relative to the union of WW and Z∪LZ\cup L reads

⋯→H∗​(W′,W∩(Z∪L),ℤ)→H∗​(W′,W,ℤ)⊕H∗​(W′,Z∪L,ℤ)→H∗​(W′,W∪(Z∪L),ℤ)→⋯\cdots\to H_{*}(W^{\prime},W\cap(Z\cup L);{\mathbb{Z}})\to H_{*}(W^{\prime},W;{\mathbb{Z}})\oplus H_{*}(W^{\prime},Z\cup L;{\mathbb{Z}})\to H_{*}(W^{\prime},W\cup(Z\cup L);{\mathbb{Z}})\to\cdots

Observe that, by excision, H3​(W′,W∪(Z∪L),ℤ)≅H3​(C,C′,ℤ)=0H_{3}(W^{\prime},W\cup(Z\cup L);{\mathbb{Z}})\cong H_{3}(C,C^{\prime};{\mathbb{Z}})=0. From here we obtain an exact sequence

(8) 0→H2​(W′,L,ℤ)→H2​(Z,∂−Z,ℤ)⊕H2​(W,ν⁡(K)∪L,ℤ)→H2​(C,C′,ℤ).0\to H_{2}(W^{\prime},L;{\mathbb{Z}})\to H_{2}(Z,\partial_{-}Z;{\mathbb{Z}})\oplus H_{2}(W,\nu(K)\cup L;{\mathbb{Z}})\to H_{2}(C,C^{\prime};{\mathbb{Z}}).

Thus, an element in H2L​(W′,ℤ)⊆H2​(W′,L,ℤ)H_{2}^{L}(W^{\prime};{\mathbb{Z}})\subseteq H_{2}(W^{\prime},L;{\mathbb{Z}}) can be identified with its image in H2​(Z,∂−Z,ℤ)⊕H2​(W,ν⁡(K)∪L,ℤ)H_{2}(Z,\partial_{-}Z;{\mathbb{Z}})\oplus H_{2}(W,\nu(K)\cup L;{\mathbb{Z}}), which we write as a pair (α,η)(\alpha,\eta).

Let us further identify H2​(Z,∂−Z,ℤ)H_{2}(Z,\partial_{-}Z;{\mathbb{Z}}) with ℤn{\mathbb{Z}}^{n} by letting the iith handle correspond to the coordinate vector eie_{i}. Then, we write

α=(α1,…,αn)∈ℤn\alpha=(\alpha_{1},\dots,\alpha_{n})\in{\mathbb{Z}}^{n}

and let α+\alpha^{+} denote its positive part and α−\alpha^{-} its negative part; i.e., αi+=max⁡(αi,0)\alpha^{+}_{i}=\operatorname{max}(\alpha_{i},0) and αi−=min⁡(αi,0)\alpha^{-}_{i}=\operatorname{min}(\alpha_{i},0). We also let |α|=∑i|αi||\alpha|=\sum_{i}|\alpha_{i}|.

Let r∈ℕnr\in{\mathbb{N}}^{n} and consider the cable K⁡(r−α−,r+α+)K(r-\alpha^{-},r+\alpha^{+}). The fact that (α,η)∈ℤn⊕H2​(W,ν⁡(K)∪L,ℤ)(\alpha,\eta)\in{\mathbb{Z}}^{n}\oplus H_{2}(W,\nu(K)\cup L;{\mathbb{Z}}) is in the kernel of the map to H2​(C,C′,ℤ)≅ℤnH_{2}(C,C^{\prime};{\mathbb{Z}})\cong{\mathbb{Z}}^{n} in (8) implies the existence of a (unique) class

ηr∈H2L∪K⁡(r−α−,r+α+)​(W,ℤ)⊆H2​(W,L∪K⁡(r−α−,r+α+),ℤ)\eta^{r}\in H_{2}^{L\cup K(r-\alpha^{-},r+\alpha^{+})}(W;{\mathbb{Z}})\subseteq H_{2}(W,L\cup K(r-\alpha^{-},r+\alpha^{+});{\mathbb{Z}})

which is sent to η\eta by the natural map to H2​(W,L∪ν⁡(K),ℤ).H_{2}(W,L\cup\nu(K);{\mathbb{Z}}).

From now on, using the deformation retraction from ν⁡(K)\nu(K) to KK, let us think of η\eta as a class in H2​(W,K∪L,ℤ)H_{2}(W,K\cup L;{\mathbb{Z}}).

0NFR

Definition 3.1. The cabled skein lasagna module of K⊂∂WK\subset\partial W at level α\alpha and in class η\eta is

𝒮¯0N,α(W;K,L,η)=(⨁r∈ℕn𝒮0N(W;K(r−α−,r+α+)∪L,ηr){(1−N)(2|r|+|α|)})/∼\underline{\mathcal{S}\mkern-3.0mu}\mkern 3.0mu_{0}^{N,\alpha}(W;K,L,\eta)=\Bigl(\bigoplus\limits_{r\in{\mathbb{N}}^{n}}\mathcal{S}_{0}^{N}(W;K(r-\alpha^{-},r+\alpha^{+})\cup L,\eta^{r})\{(1-N)(2|r|+|\alpha|)\}\Bigr)/\sim

where the equivalence ∼\sim is the transitive and linear closure of the relations

(9) βi​(b)​v∼v,ψi[d]​(v)∼0​ for ​d<N−1,ψi[N−1]​(v)∼v\beta_{i}(b)v\sim v,\ \ \psi^{[d]}_{i}(v)\sim 0\text{ for }d<N-1,\ \ \psi^{[N-1]}_{i}(v)\sim v

for all i=1,…,ni=1,\dots,n; b∈Bki−,ki+b\in B_{k_{i}^{-},k_{i}^{+}}, and v∈𝒮0N​(W,K⁡(r−α−,r+α+)∪L,ηr).v\in\mathcal{S}_{0}^{N}(W;K(r-\alpha^{-},r+\alpha^{+})\cup L,\eta^{r}).

0NFS

Theorem 3.2. Let WW be a four-manifold and L⊂∂WL\subset\partial W be a framed link. Let W′W^{\prime} be obtained from WW by attaching 2-handles along a framed link KK disjoint from LL. Then, for each (α,η)∈H2L​(W′,ℤ)(\alpha,\eta)\in H_{2}^{L}(W^{\prime};{\mathbb{Z}}), we have an isomorphism

Φ:𝒮¯0N,α​(W,K,L,η)→≅𝒮0N​(W′,L,(α,η)).\Phi:\underline{\mathcal{S}\mkern-3.0mu}\mkern 3.0mu_{0}^{N,\alpha}(W;K,L,\eta)\xrightarrow{\phantom{a}\cong\phantom{a}}\mathcal{S}_{0}^{N}(W^{\prime};L,(\alpha,\eta)).
0NFT

Proof. An element v∈𝒮0N​(W,K⁡(r−α−,r+α+)∪L,ηr)v\in\mathcal{S}_{0}^{N}(W;K(r-\alpha^{-},r+\alpha^{+})\cup L,\eta^{r}) is represented by a linear combination of lasagna fillings (Σ,{(Bi,Li,vi})(\Sigma,\{(B_{i},L_{i},v_{i}\}) in WW, where ∂Σ=K(r−α−,r+α+)∪L∪(∪iLi)\partial\Sigma=K(r-\alpha^{-},r+\alpha^{+})\cup L\cup(\cup_{i}L_{i}). We define Φ⁡(v)\Phi(v) to be the class of the linear combination of lasagna fillings with the same input data {(Bi,Li,vi}\{(B_{i},L_{i},v_{i}\} as vv, but with the surfaces given by attaching to each Σ\Sigma (along its boundary) the disjoint union of ri−αi−r_{i}-\alpha_{i}^{-} negatively oriented discs parallel to the core of it​hi^{th} 2-handle and ri+αi+r_{i}+\alpha_{i}^{+} positively oriented such discs (union over all ii).

We also define a map Φ−1\Phi^{-1} in the opposite direction, as follows. Let FF be a lasagna filling in W′W^{\prime} with surface Σ\Sigma. We isotope the input balls of FF to be inside WW, and isotope the surface Σ\Sigma such that its intersection with the 2-handles consists of several disks parallel to their cores. Removing these disks produces a lasagna filling of WW with boundary on a link of the form K⁡(r−α−,r+α+)∪LK(r-\alpha^{-},r+\alpha^{+})\cup L. We let this be Φ−1​(F)\Phi^{-1}(F).

The proofs that Φ\Phi and Φ−1\Phi^{-1} are well-defined and inverse to each other are similar to the proof of Theorem 1.1 in [25], which dealt with the case W=B4W=B^{4} and L=∅L=\emptyset. The extension to arbitrary WW and LL is obtained by replacing the Khovanov-Rozansky homologies KhRN\operatorname{KhR}_{N} with the skein lasagna modules in WW. (In the formulation here, the proof of the statement is even slightly clearer since it relates lasagna skein modules with lasagna skein modules. In particular, we do not have to choose standard lasagna fillings with “slighly smaller input balls”, as these were only required when comparing 𝒮0N​(B4,−)\mathcal{S}_{0}^{N}(B^{4},-) with KhRN\operatorname{KhR}_{N}.) ∎

0NFU

Remark 3.3. In some cases it is known that the braid group actions on the link homology of cabled links factor through the symmetric group. For Khovanov homology of links in ℝ3\mathbb{R}^{3}, this was shown by Grigsby–Licata–Wehrli [12, Theorem 2]. For the 𝔤​𝔩N\mathfrak{gl}_{N} homology of links in ℝ3\mathbb{R}^{3} (or S3S^{3}) a similar argument works in the case of parallelly oriented strands [11, Section 6.1]. We have no reason to doubt that the same could be true for anti-parallel strands, i.e. in the situation relevant for 𝒮0N\mathcal{S}_{0}^{N}, but we do not currently know how to prove it.

We will primarily be using the results from this subsection in the case where the role of WW is played by

W1:=♮m​(S1×B3),W_{1}:=\natural^{m}(S^{1}\times B^{3}),

a manifold obtained from a 0-handle by attaching some 1-handles. We denote W′W^{\prime} by W2W_{2}. Then, H2​(W1,ℤ)=0H_{2}(W_{1};{\mathbb{Z}})=0, so H2L​(W1,ℤ)=0H_{2}^{L}(W_{1};{\mathbb{Z}})=0 for any null-homologous LL, and the decomposition (2) for skein lasagna modules of links in W1W_{1} is trivial (consists of a single summand). Moreover, in this case an element (α,η)∈H2L​(W2,ℤ)⊆H2​(W2,L,ℤ)(\alpha,\eta)\in H_{2}^{L}(W_{2};{\mathbb{Z}})\subseteq H_{2}(W_{2},L;{\mathbb{Z}}) is uniquely determined by its image α\alpha in H2​(W2,W1,ℤ)≅ℤnH_{2}(W_{2},W_{1};{\mathbb{Z}})\cong{\mathbb{Z}}^{n}. Indeed, the exact sequence

0=H2​(W1,ℤ)→H2​(W1,L∪ν⁡(K),ℤ)→H1​(L∪ν⁡(K),ℤ)0=H_{2}(W_{1};{\mathbb{Z}})\to H_{2}(W_{1},L\cup\nu(K);{\mathbb{Z}})\to H_{1}(L\cup\nu(K);{\mathbb{Z}})

show that the component η\eta is determined by its image in

H1​(L∪ν⁡(K),ℤ)=H1​(L,ℤ)⊕H1​(ν⁡(K),ℤ).H_{1}(L\cup\nu(K);{\mathbb{Z}})=H_{1}(L;{\mathbb{Z}})\oplus H_{1}(\nu(K);{\mathbb{Z}}).

The part in H1​(L,ℤ)H_{1}(L;{\mathbb{Z}}) has to be the fundamental class [L][L], while the part in H1​(ν⁡(K),ℤ)≅ℤnH_{1}(\nu(K);{\mathbb{Z}})\cong{\mathbb{Z}}^{n} is the image of α\alpha under the isomorphisms

H2​(W2,W1,ℤ)→≅H2​(Z,∂−Z,ℤ)→≅H1​(∂−Z,ℤ)→≅H1​(ν⁡(K),ℤ).H_{2}(W_{2},W_{1};{\mathbb{Z}})\xrightarrow{\cong}H_{2}(Z,\partial_{-}Z;{\mathbb{Z}})\xrightarrow{\cong}H_{1}(\partial_{-}Z;{\mathbb{Z}})\xrightarrow{\cong}H_{1}(\nu(K);{\mathbb{Z}}).

Therefore, in this case the class η\eta is redundant (being determined by α\alpha), so we simply drop it from the notation, writing for example α\alpha instead of (α,η)(\alpha,\eta) for the classes in H2L​(W2,ℤ)H_{2}^{L}(W_{2};{\mathbb{Z}}). With this in mind, the isomorphism from Theorem 3.2 is written as

(10) Φ:𝒮¯0N,α​(W1,K,L)→≅𝒮0N​(W2,L,α).\Phi:\underline{\mathcal{S}\mkern-3.0mu}\mkern 3.0mu_{0}^{N,\alpha}(W_{1};K,L)\xrightarrow{\phantom{a}\cong\phantom{a}}\mathcal{S}_{0}^{N}(W_{2};L,\alpha).

Original mathematics by the credited authors. Source collection and HTML conversion remain in progress.

Ciprian Manolescu, Kevin Walker, Paul Wedrich

Original source: arXiv:2206.04616v2