This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Note on primitive disk complexes

Sangbum Cho Department of Mathematics Education, Hanyang University, Seoul 04763, Korea [email protected]  and  Jung Hoon Lee Department of Mathematics and Institute of Pure and Applied Mathematics, Jeonbuk National University, Jeonju 54896, Korea [email protected]
Abstract.

Given a Heegaard splitting of the 33-sphere, the primitive disk complex is defined to be the full subcomplex of the disk complex for one of the handlebodies of the splitting. It is an open question whether the primitive disk complex is connected or not when the genus of the splitting is greater than three. In this note, we prove that a quotient of the primitive disk complex, called the homotopy primitive disk complex, is connected.

Key words and phrases:
Heegaard splitting, disk complex, primitive disk
2020 Mathematics Subject Classification:
Primary: 57K30
The first-named author is supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (NRF-202100000002302).

1. Introduction

It is well known that every closed orientable 33-manifold can be decomposed into two handlebodies of the same genus. Such a decomposition is called a Heegaard splitting of the manifold, and the genus of the handlebodies is called the genus of the splitting. In particular, the 33-sphere admits a genus-gg Heegaard splitting for each g0g\geq 0. We denote by (V,W;Σ)(V,W;\Sigma) a genus-gg Heegaard splitting of the 33-sphere. That is, VV and WW are genus-gg handlebodies whose union is the 33-sphere, and Σ=VW=V=W\Sigma=V\cap W=\partial V=\partial W is a genus-gg closed orientable surface, called the Heegaard surface of the splitting.

An essential disk DD in VV is called a primitive disk if there exists an essential disk DD^{\prime} in WW such that D\partial D intersects D\partial D^{\prime} transversely in a single point. Such a disk DD^{\prime} is called a dual disk of DD. The disk DD^{\prime} is also primitive in WW with a dual disk DD. The primitive disk complex for the handlebody VV, denoted by 𝒫(V)\mathcal{P}(V), is defined as follows. The vertices of 𝒫(V)\mathcal{P}(V) are the isotopy classes of the primitive disks in VV, and k+1k+1 distinct vertices span a kk-simplex if they have pairwise disjoint representatives. Equivalently, the primitive disk complex is the full subcomplex of the disk complex for the handlebody VV spanned by the vertices whose representatives are primitive disks in VV. It is easy to see that the primitive disk complex 𝒫(V)\mathcal{P}(V) for a genus-gg splitting of the 33-sphere for g2g\geq 2 is a (3g4)(3g-4)-dimensional complex which is not locally finite. For the genus-0 splitting, 𝒫(V)\mathcal{P}(V) is empty and for the genus-11 splitting, 𝒫(V)\mathcal{P}(V) consists only of a single vertex.

For the genus-22 splitting of the 33-sphere, the combinatorial structure of 𝒫(V)\mathcal{P}(V) has been well understood from [1]. It is a 22-dimensional connected complex, even it is contractible. For the genus-33 splitting of the 33-sphere, Zupan [3] showed that the reducing sphere complex for the splitting is connected, which implies quickly that 𝒫(V)\mathcal{P}(V) is connected either. For the splittings of genus greater than 33, it is still an open question whether 𝒫(V)\mathcal{P}(V) is connected or not.

In this note, we define the homotopy primitive disk complex, denoted by 𝒫(V)\mathcal{HP}(V), which can be considered as a quotient complex of 𝒫(V)\mathcal{P}(V). We say that two primitive disks DD and EE are equivalent if D\partial D and E\partial E represent the same element of the free group π1(W)\pi_{1}(W). In other words, the boundary circles D\partial D and E\partial E in Σ=W\Sigma=\partial W are free homotopic in WW. Then vertices of 𝒫(V)\mathcal{HP}(V) are defined to be the equivalence classes of primitive disks in VV, and k+1k+1 distinct vertices span a kk-simplex if they have pairwise disjoint representatives. The main result is the following theorem.

Theorem 1.1.

Let DD and EE be any two primitive disks in VV. Then there exists a sequence D=E1,E2,,EnD=E_{1},E_{2},\ldots,E_{n} of primitive disks in VV satisfying that

  • EnE_{n} is equivalent to EE, and

  • for each i{1,2,,n1}i\in\{1,2,\ldots,n-1\}, either Ei=Ei+1E_{i}=E_{i+1}, or EiE_{i} and Ei+1E_{i+1} are disjoint from each other and further EiE_{i} and Ei+1E_{i+1} have dual disks EiE^{\prime}_{i} and Ei+1E^{\prime}_{i+1} respectively such that EiE^{\prime}_{i} is disjoint from Ei+1Ei+1E_{i+1}\cup E^{\prime}_{i+1} and Ei+1E^{\prime}_{i+1} is disjoint from EiEiE_{i}\cup E^{\prime}_{i}.

The theorem implies the following immediately.

Corollary 1.2.

The homotopy primitive disk complex 𝒫(V)\mathcal{HP}(V) is connected.

Throughout the paper, VV and WW will denote two genus-gg handlebodies of the Heegaard splitting (V,W;Σ)(V,W;\Sigma) of the 33-sphere for g2g\geq 2. It is known that the Heegaard splitting of the 33-sphere is unique up to isotopy for each genus [2]. Thus any orientation preserving homeomorphism of the splitting (V,W;Σ)(V,W;\Sigma) can be described as an isotopy from the identity map of the 33-sphere to the homeomorphism. We also remark that given any two primitive disks in VV there exists an orientation preserving homeomorphism of (V,W;Σ)(V,W;\Sigma) taking one to the other one.

A collection 𝒟={D1,D2,,Dg}\mathcal{D}=\{D_{1},D_{2},\ldots,D_{g}\} of pairwise disjoint, non-isotopic essential disks in VV is called a primitive system of VV if there exists a collection 𝒟={D1,D2,,Dg}\mathcal{D}^{\prime}=\{D^{\prime}_{1},D^{\prime}_{2},\ldots,D^{\prime}_{g}\} of pairwise disjoint, non-isotopic essential disks in WW such that |DiDj|=δij|\partial D_{i}\cap\partial D^{\prime}_{j}|=\delta_{ij}. The collection 𝒟\mathcal{D}^{\prime} is also a primitive system of WW. We say that 𝒟\mathcal{D} and 𝒟\mathcal{D}^{\prime} are dual systems to each other, and that {𝒟,𝒟}\{\mathcal{D},\mathcal{D}^{\prime}\} simply a dual pair. Note that D1D2DgD_{1}\cup D_{2}\cup\cdots\cup D_{g} and D1D2DgD^{\prime}_{1}\cup D^{\prime}_{2}\cup\cdots\cup D^{\prime}_{g} cut VV and WW into 33-balls respectively.

The fundamental group π1(W)\pi_{1}(W) of WW is the free group of rank gg. If any oriented simple closed curve CC in Σ\Sigma intersects D1D2Dg\partial D^{\prime}_{1}\cup\partial D^{\prime}_{2}\cup\cdots\cup\partial D^{\prime}_{g} transversely, then CC represents an element of π1(W)=x1,x2,,xg\pi_{1}(W)=\langle x_{1},x_{2},\ldots,x_{g}\rangle in a natural way. We first fix a base point inside WW and prepare an arc aa connecting the base point to a point on CC so that aa does not intersect each DiD^{\prime}_{i} for i{1,2,,g}i\in\{1,2,\ldots,g\}. Then assigning symbols x1,x2,,xgx_{1},x_{2},\ldots,x_{g} to the oriented circles D1,D2,,Dg\partial D^{\prime}_{1},\partial D^{\prime}_{2},\ldots,\partial D^{\prime}_{g} respectively, we obtain a word ww in terms of x1±1,x2±1,,xg±1x_{1}^{\pm 1},x_{2}^{\pm 1},\ldots,x_{g}^{\pm 1} by reading off consecutive intersections of the loop CaC\cup a with D1,D2,,Dg\partial D^{\prime}_{1},\partial D^{\prime}_{2},\ldots,\partial D^{\prime}_{g}. We say that such a word ww is determined by CC with respect to the system 𝒟\mathcal{D}^{\prime}. Of course, the words ww determined by CC depend on the choice of the arc aa, but they are all equivalent up to cyclic permutation. In particular, the circles D1,D2,,Dg\partial D_{1},\partial D_{2},\ldots,\partial D_{g} determine the generators x1,x2,,xgx_{1},x_{2},\ldots,x_{g} respectively.

2. Proof of the main theorem

It is well known that the automorphism group of π1(W)=x1,x2,,xg\pi_{1}(W)=\langle x_{1},x_{2},\ldots,x_{g}\rangle is generated by the elementary Nielsen transformations α,β,γ\alpha,\beta,\gamma, and δ\delta described as follows.

  • α(x1)=x1x2\alpha(x_{1})=x_{1}x_{2}, and α(xj)=xj\alpha(x_{j})=x_{j} for j=2,3,,gj=2,3,\ldots,g,

  • β(x1)=x11\beta(x_{1})=x^{-1}_{1} and β(xj)=xj\beta(x_{j})=x_{j} for j=2,3,,gj=2,3,\ldots,g,

  • γ(x1)=x2\gamma(x_{1})=x_{2}, γ(x2)=x1\gamma(x_{2})=x_{1}, and γ(xj)=xj\gamma(x_{j})=x_{j} for j=3,4,,gj=3,4,\ldots,g, and

  • δ(xj)=xj+1\delta(x_{j})=x_{j+1} for j=1,2,,g1j=1,2,\ldots,g-1, and δ(xg)=x1\delta(x_{g})=x_{1}.

Any orientation preserving homeomorphism of the splitting (V,W;Σ)(V,W;\Sigma) induces an automorphism of π1(W)\pi_{1}(W) in a natural way. Conversely, given an automorphism η\eta of π1(W)\pi_{1}(W), one can construct an orientation preserving homeomorphism ϕη\phi_{\eta} of (V,W;Σ)(V,W;\Sigma) that induces η\eta. More precisely, we have the following lemma.

Lemma 2.1.

Let η\eta be any automorphism of π1(W)\pi_{1}(W), and let DD be a primitive disk in VV. Then there exists an orientation preserving homeomorphism ϕη\phi_{\eta} of (V,W;Σ)(V,W;\Sigma) that induces η\eta, and furthermore there exists a sequence E1,E2,,EnE_{1},E_{2},\ldots,E_{n} of primitive disks in VV satisfying that

  • E1=DE_{1}=D and En=ϕη(D)E_{n}=\phi_{\eta}(D), and

  • for each i{1,2,,n1}i\in\{1,2,\ldots,n-1\}, either Ei=Ei+1E_{i}=E_{i+1}, or EiE_{i} and Ei+1E_{i+1} are disjoint from each other and they have dual disks EiE^{\prime}_{i} and Ei+1E^{\prime}_{i+1} respectively such that EiE^{\prime}_{i} is disjoint from Ei+1Ei+1E_{i+1}\cup E^{\prime}_{i+1} and Ei+1E^{\prime}_{i+1} is disjoint from EiEiE_{i}\cup E^{\prime}_{i}.

Proof.

We will describe an orientation preserving homeomorphism of (V,W;Σ)(V,W;\Sigma) as an isotopy from the identity to the homeomorphism. Choose a primitive system 𝒟={D1,D2,,Dg}\mathcal{D}=\{D_{1},D_{2},\ldots,D_{g}\} of VV such that D1=DD_{1}=D, together with a dual system 𝒟={D1,D2,,Dg}\mathcal{D}^{\prime}=\{D^{\prime}_{1},D^{\prime}_{2},\ldots,D^{\prime}_{g}\} of WW, and then assign symbols x1,x2,,xgx_{1},x_{2},\ldots,x_{g} to the circles D1,D2,,Dg\partial D^{\prime}_{1},\partial D^{\prime}_{2},\ldots,\partial D^{\prime}_{g} respectively. Fix a base point inside WW and choose pairwise disjoint arcs aia_{i}, for i{1,2,,g}i\in\{1,2,\ldots,g\}, connecting the base point to a point in Di\partial D_{i} respectively so that each aia_{i} is disjoint from D1D2DgD^{\prime}_{1}\cup D^{\prime}_{2}\cup\cdots\cup D^{\prime}_{g}. Then, with a choice of orientations of boundary circles of DiD_{i} and of DjD^{\prime}_{j}, for i,j{1,2,,g}i,j\in\{1,2,\ldots,g\}, the circles D1,D2,,Dg\partial D_{1},\partial D_{2},\ldots,\partial D_{g} represent the generators x1,x2,,xgx_{1},x_{2},\ldots,x_{g} of π1(W)=x1,x2,,xg\pi_{1}(W)=\langle x_{1},x_{2},\ldots,x_{g}\rangle respectively. See Figure 1.


\labellist\pinlabel

DgD^{\prime}_{g} [B] at -11 210 \pinlabelD1D^{\prime}_{1} [B] at 163 326 \pinlabelD2D^{\prime}_{2} [B] at 340 242 \pinlabelD3D^{\prime}_{3} [B] at 347 84 \pinlabelDg\partial D_{g} [B] at 86 153 \pinlabelD1\partial D_{1} [B] at 143 220 \pinlabelD2\partial D_{2} [B] at 225 209 \pinlabelD3\partial D_{3} [B] at 255 140 \pinlabelaga_{g} [B] at 126 176 \pinlabela1a_{1} [B] at 178 194 \pinlabela2a_{2} [B] at 220 162 \pinlabela3a_{3} [B] at 206 119 \pinlabelWW [B] at 143 60 \endlabellist[Uncaptioned image]

Figure 1.

First we construct the homeomorphisms ϕα,ϕβ,ϕγ\phi_{\alpha},\phi_{\beta},\phi_{\gamma}, and ϕδ\phi_{\delta} of the splitting (V,W;Σ)(V,W;\Sigma) that induce the four elementary Nielsen transformations α,β,γ\alpha,\beta,\gamma, and δ\delta respectively.

For the transformation α\alpha, we have the homeomorphism ϕα\phi_{\alpha} of (V,W;Σ)(V,W;\Sigma) described by the isotopy in Figure 2 (a). We drag a foot of the first handle along the second handle to make the circle ϕα(D1)\phi_{\alpha}(\partial D_{1}) determine the word x1x2x_{1}x_{2} with respect to 𝒟\mathcal{D}^{\prime}, while ϕα(Dj)\phi_{\alpha}(\partial D_{j}) determines xjx_{j} for each j{2,3,,g}j\in\{2,3,\ldots,g\}. After the isotopy for ϕα\phi_{\alpha}, all the disks in 𝒟𝒟\mathcal{D}\cup\mathcal{D}^{\prime} except D1D_{1} and D2D^{\prime}_{2} remain unchanged as subsets.

For the transformation β\beta, the homeomorphism ϕβ\phi_{\beta} is described as a π\pi-rotation of the first handle as in Figure 2 (b). The circle ϕβ(D1)\phi_{\beta}(\partial D_{1}) determines the word x11x^{-1}_{1} with respect to 𝒟\mathcal{D}^{\prime}, while ϕβ(Dj)\phi_{\beta}(\partial D_{j}) determines xjx_{j} for j{2,3,,g}j\in\{2,3,\ldots,g\}. After the isotopy for ϕβ\phi_{\beta}, all the disks in 𝒟𝒟\mathcal{D}\cup\mathcal{D}^{\prime} remain unchanged as subsets.

For the transformation γ\gamma, the homeomorphism ϕγ\phi_{\gamma} described by the isotopy in Figure 2 (c) exchanges the first and the second handles so that ϕγ(D1)\phi_{\gamma}(\partial D_{1}) and ϕγ(D2)\phi_{\gamma}(\partial D_{2}) determine x2x_{2} and x1x_{1} respectively with respect to 𝒟\mathcal{D}^{\prime}, while ϕγ(Dj)\phi_{\gamma}(\partial D_{j}) determines xjx_{j} for j{3,4,,g}j\in\{3,4,\ldots,g\}. The isotopy for ϕγ\phi_{\gamma} exchanges D1D_{1} and D1D^{\prime}_{1} with D2D_{2} and D2D^{\prime}_{2} respectively, but all other disks in 𝒟𝒟\mathcal{D}\cup\mathcal{D}^{\prime} remain unchanged.

Finally, the homeomorphism ϕδ\phi_{\delta} described by the isotopy in Figure 2 (d), which is a permutation of each handle to the next one, induces the transformation δ\delta. For each j{1,2,,g1}j\in\{1,2,\ldots,g-1\}, ϕδ(Dj)\phi_{\delta}(\partial D_{j}) determines xj+1x_{j+1}, and ϕδ(Dg)\phi_{\delta}(\partial D_{g}) determines x1x_{1} with respect to 𝒟\mathcal{D}^{\prime}. The isotopy for ϕδ\phi_{\delta} sends DjD_{j} and DjD^{\prime}_{j} to Dj+1D_{j+1} and Dj+1D^{\prime}_{j+1} respectively for j{1,2,,g1}j\in\{1,2,\ldots,g-1\}, and sends DgD_{g} and DgD^{\prime}_{g} to D1D_{1} and D1D^{\prime}_{1} respectively.


\labellist\pinlabel

(a) [B] at 230 400 \pinlabel(b) [B] at 230 270 \pinlabel(c) [B] at 230 132 \pinlabel(d) [B] at 230 0 \pinlabelϕα\phi_{\alpha} [B] at 229 442 \pinlabelϕβ\phi_{\beta} [B] at 229 313 \pinlabelϕγ\phi_{\gamma} [B] at 229 175 \pinlabelϕδ\phi_{\delta} [B] at 229 36

\pinlabel

D1D^{\prime}_{1} [B] at 59 480 \pinlabelD1D^{\prime}_{1} [B] at 59 347 \pinlabelD1D^{\prime}_{1} [B] at 59 210 \pinlabelD1D^{\prime}_{1} [B] at 59 71

\pinlabel

D2D^{\prime}_{2} [B] at 152 480 \pinlabelD2D^{\prime}_{2} [B] at 152 347 \pinlabelD2D^{\prime}_{2} [B] at 152 210 \pinlabelD2D^{\prime}_{2} [B] at 152 71

\pinlabel

D1\partial D_{1} [B] at 59 397 \pinlabelD1\partial D_{1} [B] at 59 265 \pinlabelD1\partial D_{1} [B] at 79 130 \pinlabelD1\partial D_{1} [B] at 84 -5

\pinlabel

D2\partial D_{2} [B] at 163 397 \pinlabelD2\partial D_{2} [B] at 163 265 \pinlabelD2\partial D_{2} [B] at 173 130 \pinlabelD2\partial D_{2} [B] at 177 -5

\pinlabel

ϕα(D1)\phi_{\alpha}(D^{\prime}_{1}) [B] at 308 483 \pinlabelϕβ(D1)\phi_{\beta}(D^{\prime}_{1}) [B] at 308 351 \pinlabelϕγ(D1)\phi_{\gamma}(D^{\prime}_{1}) [B] at 401 216 \pinlabelϕδ(D1)\phi_{\delta}(D^{\prime}_{1}) [B] at 401 74

\pinlabel

ϕα(D2)\phi_{\alpha}(D^{\prime}_{2}) [B] at 355 469 \pinlabelϕβ(D2)\phi_{\beta}(D^{\prime}_{2}) [B] at 404 351 \pinlabelϕγ(D2)\phi_{\gamma}(D^{\prime}_{2}) [B] at 309 214 \pinlabelϕδ(Dg)\phi_{\delta}(D^{\prime}_{g}) [B] at 309 75

\pinlabel

ϕα(D1)\phi_{\alpha}(\partial D_{1}) [B] at 279 397 \pinlabelϕβ(D1)\phi_{\beta}(\partial D_{1}) [B] at 314 269 \pinlabelϕγ(D1)\phi_{\gamma}(\partial D_{1}) [B] at 403 131 \pinlabelϕδ(D1)\phi_{\delta}(\partial D_{1}) [B] at 430 -7

\pinlabel

ϕα(D2)\phi_{\alpha}(\partial D_{2}) [B] at 465 460 \pinlabelϕβ(D2)\phi_{\beta}(\partial D_{2}) [B] at 407 269 \pinlabelϕγ(D2)\phi_{\gamma}(\partial D_{2}) [B] at 314 131 \pinlabelϕδ(Dg)\phi_{\delta}(\partial D_{g}) [B] at 338 -7

\endlabellist[Uncaptioned image]
Figure 2.

For each of the four homeomorphisms ϕα\phi_{\alpha}, ϕβ\phi_{\beta}, ϕγ\phi_{\gamma}, and ϕδ\phi_{\delta}, we make the union of arcs a1a2aga_{1}\cup a_{2}\cup\cdots\cup a_{g} remain invariant after the isotopies. Furthermore we observe that, if ϕ\phi is one of ϕα,ϕβ,ϕγ\phi_{\alpha},\phi_{\beta},\phi_{\gamma}, or ϕδ\phi_{\delta}, then either Dj=ϕ(Dj)D_{j}=\phi(D_{j}) or DjD_{j} and ϕ(Dj)\phi(D_{j}) are disjoint from each other for each j{1,2,,g}j\in\{1,2,\ldots,g\}, and in this case we can easily find dual disks EE^{\prime} and E′′E^{\prime\prime} of DjD_{j} and ϕ(Dj)\phi(D_{j}) respectively, satisfying that EE^{\prime} is disjoint from ϕ(Dj)E′′\phi(D_{j})\cup E^{\prime\prime} and E′′E^{\prime\prime} is disjoint from DjED_{j}\cup E^{\prime}. For example, for D1D_{1} and ϕα(D1)\phi_{\alpha}(D_{1}), we can take E=ϕα(D2)E^{\prime}=\phi_{\alpha}(D^{\prime}_{2}) and E′′=D2E^{\prime\prime}=D^{\prime}_{2}.

Simply we say that the homeomorphisms ϕα\phi_{\alpha}, ϕβ\phi_{\beta}, ϕγ\phi_{\gamma}, and ϕδ\phi_{\delta} are realizations of α\alpha, β\beta, γ\gamma, and δ\delta with respect to the dual pair {𝒟,𝒟}\{\mathcal{D},\mathcal{D}^{\prime}\}.

Next, let η\eta be any automorphism of π1(W)\pi_{1}(W). Then η\eta can be written as a finite product of elementary Nielsen transformations. That is, η=ηn1ηn2η1\eta=\eta_{n-1}\eta_{n-2}\cdots\eta_{1} for some n2n\geq 2 where ηi\eta_{i} is one of α\alpha, β\beta, γ\gamma, and δ\delta. For η1\eta_{1}, we find a realization ϕ1\phi_{1}, which is one of ϕα\phi_{\alpha}, ϕβ\phi_{\beta}, ϕγ\phi_{\gamma}, or ϕδ\phi_{\delta} with respect to the dual pair {𝒟,𝒟}\{\mathcal{D},\mathcal{D}^{\prime}\}, described in the above. For η2\eta_{2}, letting 𝒟1={ϕ1(D1),ϕ1(D2),,ϕ1(Dg)}\mathcal{D}_{1}=\{\phi_{1}(D_{1}),\phi_{1}(D_{2}),\ldots,\phi_{1}(D_{g})\} and 𝒟1={ϕ1(D1),ϕ1(D2),,ϕ1(Dg)}\mathcal{D}^{\prime}_{1}=\{\phi_{1}(D^{\prime}_{1}),\phi_{1}(D^{\prime}_{2}),\ldots,\phi_{1}(D^{\prime}_{g})\}, we have a realization ϕ2\phi_{2}, which is one of ϕα\phi_{\alpha}, ϕβ\phi_{\beta}, ϕγ\phi_{\gamma}, or ϕδ\phi_{\delta} with respect to the dual pair {𝒟1,𝒟1}\{\mathcal{D}_{1},\mathcal{D}^{\prime}_{1}\}. Continuing the process, we have a realization ϕi\phi_{i} of ηi\eta_{i} for each i{2,,n1}i\in\{2,\ldots,n-1\}, with respect to the dual pair {𝒟i1,𝒟i1}\{\mathcal{D}_{i-1},\mathcal{D}^{\prime}_{i-1}\}, where 𝒟i1\mathcal{D}_{i-1} and 𝒟i1\mathcal{D}^{\prime}_{i-1} consist of disks ϕi1ϕ2ϕ1(Dj)\phi_{i-1}\cdots\phi_{2}\phi_{1}(D_{j}) and ϕi1ϕ2ϕ1(Dj)\phi_{i-1}\cdots\phi_{2}\phi_{1}(D^{\prime}_{j}) respectively, for j{1,2,,g}j\in\{1,2,\ldots,g\}. Then the product ϕn1ϕn2ϕ1\phi_{n-1}\phi_{n-2}\cdots\phi_{1} is the desired homeomorphism ϕη\phi_{\eta} that induces η\eta, and letting D=E1D=E_{1}, and ϕi1ϕ2ϕ1(D)=Ei\phi_{i-1}\cdots\phi_{2}\phi_{1}(D)=E_{i} for i{2,3,,n}i\in\{2,3,\ldots,n\} we have the desired sequence of primitive disks in VV satisfying the condition. ∎

Proof of Theorem 1.1.

Choose any orientation preserving homeomorphism ϕ\phi of (V,W;Σ)(V,W;\Sigma) taking DD to EE. Let η\eta be the automorphism of π1(W)\pi_{1}(W) induced by ϕ\phi. By Lemma 2.1, there exists another orientation preserving homeomorphism ϕη\phi_{\eta} of (V,W;Σ)(V,W;\Sigma) that induces η\eta, together with a sequence E1,E2,,EnE_{1},E_{2},\ldots,E_{n} of primitive disks in VV satisfying the condition in the theorem. It is clear that E=ϕ(D)E=\phi(D) and En=ϕη(D)E_{n}=\phi_{\eta}(D) are equivalent to each other. ∎

References

  • [1] Cho, S., Homeomorphisms of the 33-sphere that preserve a Heegaard splitting of genus two, Proc. Amer. Math. Soc. 136 (2008), no. 3, 1113–1123.
  • [2] Waldhausen, F., Heegaard-Zerlegungen der 3-Sphäre, Topology 7 (1968), 195–203.
  • [3] Zupan, A., The Powell conjecture and reducing sphere complexes, J. Lond. Math. Soc. (2) 101 (2020), no. 1, 328–348.