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

An infinite presentation for the twist subgroup of the mapping class group of a compact non-orientable surface

Ryoma Kobayashi Department of General Education,National Institute of Technology, Ishikawa College,Tsubata, Ishikawa, 929-0392, Japan kobayashi_ryoma@ishikawa-nct.ac.jp  and  Genki Omori Department of Mathematics,Faculty of Science and Technology,Tokyo University of Science,Noda, Chiba, 278-8510, Japan omori_genki@ma.noda.tus.ac.jp
Abstract.

A finite presentation for the subgroup of the mapping class group of a compact non-orientable surface generated by all Dehn twists was given by Stukow [26]. In this paper, we give an infinite presentation for this group, mainly using the presentation given by Stukow [26] and Birman exact sequences on mapping class groups of non-orientable surfaces.

Key words and phrases:
mapping class group, twist subgroup, presentation.
2010 Mathematics Subject Classification:
Primary 20F05, Secondary 57M07.
The first author was supported by JSPS KAKENHI Grant Number JP19K14542 and 22K13920. The second author was supported by JSPS KAKENHI Grant Number JP19K23409 and JP21K13794.
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.

1. Introduction

1.1. Background

For g1g\geq 1 and n0n\geq 0, let Ng,nN_{g,n} denote a compact non-orientable surface of genus gg with nn boundary components, that is, Ng,nN_{g,n} is a surface obtained by removing nn open disks from a connected sum of gg real projective planes. For g0g\geq 0 and n0n\geq 0, let Σg,n\Sigma_{g,n} denote a compact orientable surface of genus gg with nn boundary components, that is, Σg,n\Sigma_{g,n} is a surface obtained by removing nn open disks from a connected sum of gg tori. As shown in Figure 1, we can regard Ng,nN_{g,n} as a surface obtained by attaching g2hg-2h Möbius bands to g2hg-2h boundary components of Σh,n+g2h\Sigma_{h,n+g-2h}, for 0h<g2\displaystyle 0\leq{h}<\frac{g}{2}. We call these attached Möbius bands crosscaps.

Refer to caption
Figure 1. A model of a non-orientable surface Ng,nN_{g,n}.

The mapping class group of Ng,nN_{g,n}, denoted by (Ng,n)\mathcal{M}(N_{g,n}), is the group consisting of isotopy classes of all diffeomorphisms of Ng,nN_{g,n} which fix the boundary pointwise. The mapping class group of Σg,n\Sigma_{g,n}, denoted by (Σg,n)\mathcal{M}(\Sigma_{g,n}), is the group consisting of isotopy classes of all orientation-preserving diffeomorphisms of Σg,n\Sigma_{g,n} which fix the boundary pointwise. It is well known that (Σg,n)\mathcal{M}(\Sigma_{g,n}) is generated by only Dehn twists (see [3, 17, 4]). On the other hand, (Ng,n)\mathcal{M}(N_{g,n}) can not be generated by only Dehn twists. As generators of (Ng,n)\mathcal{M}(N_{g,n}), other than Dehn twists, crosscap slides or crosscap transpositions are needed (see [16, 18]). Let us consider the subgroup of (Ng,n)\mathcal{M}(N_{g,n}) generated by all Dehn twists, denoted by 𝒯(Ng,n)\mathcal{T}(N_{g,n}). We call 𝒯(Ng,n)\mathcal{T}(N_{g,n}) the twist subgroup of (Ng,n)\mathcal{M}(N_{g,n}).

We now explain about the history of studies on presentations for (Σg,n)\mathcal{M}(\Sigma_{g,n}), (Ng,n)\mathcal{M}(N_{g,n}) and 𝒯(Ng,n)\mathcal{T}(N_{g,n}). Finite presentations for (Σg,n)\mathcal{M}(\Sigma_{g,n}) were given by Hatcher-Thurston [10] and Harer [9], and subsequently simplified by Wajnryb [28] and Matsumoto [20] for n1n\leq 1. Gervais [8] and Labruère-Paris [15] gave finite presentations of (Σg,n)\mathcal{M}(\Sigma_{g,n}) for n2n\geq 2. Gervais [7] gave an infinite presentation for (Σg,n)\mathcal{M}(\Sigma_{g,n}) by using the presentation for (Σg,n)\mathcal{M}(\Sigma_{g,n}) given in [9, 28], and then Luo [19] simplified its presentation. Finite presentations for (N2,0)\mathcal{M}(N_{2,0}), (N2,1)\mathcal{M}(N_{2,1}), (N3,0)\mathcal{M}(N_{3,0}) and (N4,0)\mathcal{M}(N_{4,0}) ware given by [16], [23], [2] and [27] respectively. Note that (N1,0)\mathcal{M}(N_{1,0}) and (N1,1)\mathcal{M}(N_{1,1}) are trivial (see [5]). Paris-Szepietowski [22] gave a finite presentation of (Ng,n)\mathcal{M}(N_{g,n}) for g+n>3g+n>3 with n1n\leq 1. Stukow [25] gave another finite presentation of (Ng,n)\mathcal{M}(N_{g,n}) for g+n>3g+n>3 with n1n\leq 1, applying Tietze transformations for the presentation of (Ng,n)\mathcal{M}(N_{g,n}) given in [22]. The second author [21] gave an infinite presentation of (Ng,n)\mathcal{M}(N_{g,n}) for g1g\geq 1 and n1n\leq 1, using the presentation of (Ng,n)\mathcal{M}(N_{g,n}) given in [25], and then, following this work, the authors [12] gave an infinite presentation of (Ng,n)\mathcal{M}(N_{g,n}) for g1g\geq 1 and n2n\geq 2. It is known that 𝒯(Ng,n)\mathcal{T}(N_{g,n}) is the index 22 subgroup of (Ng,n)\mathcal{M}(N_{g,n}) (see [18]). Stukow [26] gave a finite presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) for g+n>3g+n>3 with n1n\leq 1, applying the Reidemeister-Schreier method for the presentation of (Ng,n)\mathcal{M}(N_{g,n}) given in [25] (see Theorems 2.5 and 2.6).

In this paper, we give an infinite presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) for g1g\geq 1 and n0n\geq 0 (see Theorem 1.1), mainly using the presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) given in [26] and Birman exact sequences on mapping class groups of non-orientable surfaces.

Through this paper, the product gfgf of mapping classes ff and gg means that we apply ff first and then gg. Moreover we do not distinguish a loop from its isotopy class.

1.2. Main result

For a simple closed curve cc of Ng,nN_{g,n}, a regular neighborhood of cc is either an annulus or a Möbius band. We call cc a two sided or a one sided simple closed curve respectively. For a two sided simple closed curve cc, we can take two orientations +c+_{c} and c-_{c} of a regular neighborhood of cc. The right handed Dehn twist tc;θt_{c;\theta} about a two sided simple closed curve cc with respect to θ{+c,c}\theta\in\{+_{c},-_{c}\} is the isotopy class of the map described as shown in Figure 2. tc;θt_{c;\theta} does not depend on a choice of a representative curve of the isotopy class of cc and its regular neighborhood. We remark that although the Dehn twist was defined for an oriented simple closed curve in [21, 12], in this paper we do not consider an orientation of a simple closed curve in the definition of the Dehn twist. We write tc;θ=tct_{c;\theta}=t_{c} if the orientation θ\theta is given explicitly. That is, the direction of the twist is indicated by an arrow written beside cc as shown in Figure 2.

Refer to caption
Figure 2. The Dehn twist tc;θ=tct_{c;\theta}=t_{c} about cc with respect to θ{+α,α}\theta\in\{+_{\alpha},-_{\alpha}\}.

We denote by f(θ)f_{\ast}(\theta) the orientation of a regular neighborhood of f(c)f(c) induced from θ{+c,c}\theta\in\{+_{c},-_{c}\}, for a two sided simple closed curve cc of Ng,nN_{g,n} and f(Ng,n)f\in\mathcal{M}(N_{g,n}). Let c1,,ckc_{1},\dots,c_{k}, c0c_{0}, c0c_{0}^{\prime} and d1,,d7d_{1},\dots,d_{7} be simple closed curves with arrows as shown in Figure 3. (Ng,n)\mathcal{M}(N_{g,n}) admits following relations.

  1. (1)

    tc;θ=1t_{c;\theta}=1 if cc bounds a disk or a Möbius band.

  2. (2)

    tc;c1=tc;+ct_{c;-_{c}}^{-1}=t_{c;+_{c}}.

  3. (3)

    ftc;θf1=tf(c);f(θ)ft_{c;\theta}f^{-1}=t_{f(c);f_{\ast}(\theta)} for f(Ng,n)f\in\mathcal{M}(N_{g,n}).

  4. (4)

    (tc1tc2tck)k+1=tc0tc0(t_{c_{1}}t_{c_{2}}\cdots{}t_{c_{k}})^{k+1}=t_{c_{0}}t_{c_{0}^{\prime}} if kk is odd,
    (tc1tc2tck)2k+2=tc0(t_{c_{1}}t_{c_{2}}\cdots{}t_{c_{k}})^{2k+2}=t_{c_{0}} if kk is even.

  5. (5)

    td1td2td3=td4td5td6td7t_{d_{1}}t_{d_{2}}t_{d_{3}}=t_{d_{4}}t_{d_{5}}t_{d_{6}}t_{d_{7}}.

We can check the relations (1) and (2) easily. The relations (3), (4) and (5) are called a conjugation relation, a kk-chain relation, a lantern relation respectively. These are famous relations on mapping class groups. In the relation (3), if f=tc;θf=t_{c^{\prime};\theta^{\prime}}, |cc|=0|c\cap{}c^{\prime}|=0 or 11, and the orientations θ\theta and θ\theta^{\prime} are compatible, then the relation can be rewritten as a commutativity relation tc;θtc;θ=tc;θtc;θt_{c;\theta}t_{c^{\prime};\theta^{\prime}}=t_{c^{\prime};\theta^{\prime}}t_{c^{;}\theta} or a braid relation tc;θtc;θtc;θ=tc;θtc;θtc;θt_{c;\theta}t_{c^{\prime};\theta^{\prime}}t_{c;\theta}=t_{c^{\prime};\theta^{\prime}}t_{c^{;}\theta}t_{c^{\prime};\theta^{\prime}} respectively.

Refer to caption
Figure 3.

Our main result is as follows.

Theorem 1.1.

For g1g\geq 1 and n0n\geq 0, 𝒯(Ng,n)\mathcal{T}(N_{g,n}) admits a presentation with a generating set

X={tc;θ|cis a two sided simple closed curve ofNg,nandθis an orientation of a regular neighborhood ofc.}.X=\left\{t_{c;\theta}\left|\begin{array}[]{l}c~{}\textrm{is a two sided simple closed curve of}~{}N_{g,n}~{}\textrm{and}\\ \theta~{}\textrm{is an orientation of a regular neighborhood of}~{}c.\end{array}\right.\right\}.

The defining relations are

  1. (1)

    tc;θ=1t_{c;\theta}=1 if cc bounds a disk or a Möbius band,

  2. (2)

    tc;c1=tc;+ct_{c;-_{c}}^{-1}=t_{c;+_{c}},

  3. (3)

    all the conjugation relations ftc;θf1=tf(c);f(θ)ft_{c;\theta}f^{-1}=t_{f(c);f_{\ast}(\theta)} for fXf\in{X},

  4. (4)

    all the 22-chain relations,

  5. (5)

    all the lantern relations,

Remark 1.2.

If a two sided simple closed curve cc is a non-separating curve of N2,0N_{2,0}, then we can check that tc;+ct_{c;+_{c}} and tc;ct_{c;-_{c}} are the same elements in XX of Theorem 1.1. Hence by the relation (2) of Theorem 1.1, we have the relation tc;θ2=1t_{c;\theta}^{2}=1.

Remark 1.3.

Applying Theorem in [19] to a regular neighborhood of i=1kci\cup_{i=1}^{k}c_{i} in Figure 3, it follows that any chain relation is obtained from the relations (1), (3), (4) and (5) of Theorem 1.1. We assign the label (4)(4)^{\prime} to all the chain relations.

Here is the outline of this paper. In Section 2, we explain what we need to prove our main result. In Section 3, we prove that the infinitely presented group with the presentation in Theorem 1.1 is isomorphic to 𝒯(Ng,n)\mathcal{T}(N_{g,n}) with a presentation which is already known, for n1n\leq 1. In Section 4, we complete the proof of Theorem 1.1, by induction on nn.

2. Preliminaries

In this section, we explain what we need to prove our main result Theorem 1.1. In Section 2.1, we define the capping map, the point pushing map, the forgetful map and the crosscap pushing map on mapping class groups for non-orientable surfaces. In Section 2.2, we introduce a finite presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) for g1g\geq 1 and n1n\leq 1.

2.1. Homomorphisms on mapping class groups of non-orientable surfaces

We first define the capping map, the point pushing map and the forgetful map on mapping class groups of non-orientable surfaces. Take a point \ast in the interior of Ng,n1N_{g,n-1}. Let (Ng,n1,)\mathcal{M}(N_{g,n-1},\ast) denote the group consisting of isotopy classes of all diffeomorphisms of Ng,n1N_{g,n-1} which fix \ast and the boundary pointwise. We can regard Ng,nN_{g,n} as a complement of an open disk neighborhood of \ast in Ng,n1N_{g,n-1}. The natural embedding Ng,nNg,n1N_{g,n}\hookrightarrow{}N_{g,n-1} induces the homomorphism

𝒞:(Ng,n)(Ng,n1,)\mathcal{C}:\mathcal{M}(N_{g,n})\to\mathcal{M}(N_{g,n-1},\ast)

which is called the capping map. The point pushing map

𝒫:π1(Ng,n1,)(Ng,n1,)\mathcal{P}_{\ast}:\pi_{1}(N_{g,n-1},\ast)\to\mathcal{M}(N_{g,n-1},\ast)

is defined as follows. For any loop xπ1(Ng,n1,)x\in\pi_{1}(N_{g,n-1},\ast), 𝒫(x)\mathcal{P}_{\ast}(x) is described by pushing \ast once along xx. The forgetful homomorphism

:(Ng,n1,)(Ng,n1)\mathcal{F}:\mathcal{M}(N_{g,n-1},\ast)\to\mathcal{M}(N_{g,n-1})

is defined naturally, that is, (f)\mathcal{F}(f) does not necessarily fix \ast.

Next, we consider exact sequences on mapping class groups of non-orientable surfaces. Let +(Ng,n1,)\mathcal{M}^{+}(N_{g,n-1},\ast) denote the subgroup of (Ng,n1,)\mathcal{M}(N_{g,n-1},\ast) consisting of elements which preserve a local orientation of \ast, and π1+(Ng,n1,)\pi_{1}^{+}(N_{g,n-1},\ast) the subgroup of π1(Ng,n1,)\pi_{1}(N_{g,n-1},\ast) generated by two sided simple loops. We have the exact sequences

(1) 1(Ng,n)𝒞+(Ng,n1,)1,\displaystyle 1\to\mathbb{Z}\to\mathcal{M}(N_{g,n})\overset{\mathcal{C}}{\to}\mathcal{M}^{+}(N_{g,n-1},\ast)\to 1,
(2) π1+(Ng,n1,)𝒫+(Ng,n1,)(Ng,n1)1.\displaystyle\pi_{1}^{+}(N_{g,n-1},\ast)\overset{\mathcal{P}_{\ast}}{\to}\mathcal{M}^{+}(N_{g,n-1},\ast)\overset{\mathcal{F}}{\to}\mathcal{M}(N_{g,n-1})\to 1.

The second sequence is called the Birman exact sequence, introduced by Birman [1]. Note that 𝒫\mathcal{P}_{\ast} is injective except for the case (g,n)=(2,1)(g,n)=(2,1). For details see [1, 14, 24, 6, 22].

Remark 2.1.

For a simple loop απ1+(Ng,n1,)\alpha\in\pi_{1}^{+}(N_{g,n-1},\ast), we take an orientation θ\theta of a regular neighborhood of α\alpha. Let γ1\gamma_{1} and γ2\gamma_{2} be the right side boundary curve and the left side boundary curve of the regular neighborhood of α\alpha for the direction of α\alpha, respectively, where the left and right sides are determined by θ\theta. Then we have 𝒫(α)=tγ1;θ1tγ2;θ21\mathcal{P}_{\ast}(\alpha)=t_{\gamma_{1};\theta_{1}}t_{\gamma_{2};\theta_{2}}^{-1}, where θ1\theta_{1} and θ2\theta_{2} are the orientations compatible with θ\theta (see Figure 4).

Refer to caption
Figure 4. The boundary curves γ1\gamma_{1} and γ2\gamma_{2} of a regular neighborhood of α\alpha.
Remark 2.2.

Let α\alpha, β\beta and γπ1+(Ng,n1,)\gamma\in\pi_{1}^{+}(N_{g,n-1},\ast) be simple loops with αβ=γ\alpha\beta=\gamma. If α\alpha and β\beta intersect transversally at only \ast as shown in Figure 5, then the relation 𝒫(γ)=𝒫(β)𝒫(α)\mathcal{P}_{\ast}(\gamma)=\mathcal{P}_{\ast}(\beta)\mathcal{P}_{\ast}(\alpha) is obtained from the relation (3) of Theorem 1.1 (see Lemma 5.4 in [12]). If α\alpha and β\beta intersect tangentially at only \ast as shown in Figure 5, then the relation 𝒫(γ)=𝒫(β)𝒫(α)\mathcal{P}_{\ast}(\gamma)=\mathcal{P}_{\ast}(\beta)\mathcal{P}_{\ast}(\alpha) is obtained from the relation (5) of Theorem 1.1 and a relation tc;θ=1t_{c;\theta}=1, where cc is a simple closed curve bounding a disk neighborhood of \ast (see Lemma 5.2 in [12]). We call this relation the extended lantern relation.

Refer to caption
Figure 5.

Finally, we define the crosscap pushing map. Let SS be a surface obtained by shrinking some crosscap of Ng,nN_{g,n} into a point. We call this operation the blowdown with respect to the crosscap. Note that SS is diffeomorphic to either Ng1,nN_{g-1,n} or Σg12,n\Sigma_{\frac{g-1}{2},n} (see Figure 6). Let \ast be the shrinked point of SS. Conversely, we can obtain Ng,nN_{g,n} from SS, and call this operation the blowup with respect to \ast. The crosscap pushing map

𝒫:π1(S,)(Ng,n)\mathcal{P}_{\otimes}:\pi_{1}(S,\ast)\to\mathcal{M}(N_{g,n})

is defined as follows. For xπ1(S,)x\in\pi_{1}(S,\ast), let x~\tilde{x} be an oriented loop of Ng,nN_{g,n} induced from xx by the blowup with respect to \ast. 𝒫(x)\mathcal{P}_{\otimes}(x) is described by pushing the crosscap, which is obtained by the blowup with respect to \ast, once along x~\tilde{x} (see Figure 7).

Refer to caption
Figure 6.
Refer to caption
Figure 7. The crosscap pussing map 𝒫\mathcal{P}_{\otimes}.

Similar to Remarks 2.1 and 2.2, we have the followings.

Remark 2.3.

For a two sided simple loop απ1(S,)\alpha\in\pi_{1}(S,\ast), we take an orientation θ\theta of a regular neighborhood of α\alpha. Let γ1\gamma_{1} and γ2\gamma_{2} be the right side boundary curve and the left side boundary curve of the regular neighborhood of α\alpha for the direction of α\alpha, respectively, where the left and right sides are determined by θ\theta. We also denote by γ~1\tilde{\gamma}_{1} and γ~2\tilde{\gamma}_{2} loops of Ng,nN_{g,n} induced from γ1\gamma_{1} and γ2\gamma_{2} by the blowup with respect to \ast, respectively. Then we have 𝒫(α)=tγ~1;θ~1tγ~2;θ~21\mathcal{P}_{\otimes}(\alpha)=t_{\tilde{\gamma}_{1};\tilde{\theta}_{1}}t_{\tilde{\gamma}_{2};\tilde{\theta}_{2}}^{-1}, where θ~1\tilde{\theta}_{1} and θ~2\tilde{\theta}_{2} are the orientations of regular neighborhoods of γ~1\tilde{\gamma}_{1} and γ~2\tilde{\gamma}_{2} induced by the blowup with respect to \ast from the orientations of regular neighborhoods of γ1\gamma_{1} and γ2\gamma_{2} which are compatible with θ\theta, respectively.

Remark 2.4.

Let α\alpha, β\beta and γπ1(S,)\gamma\in\pi_{1}(S,\ast) be two sided simple loops with αβ=γ\alpha\beta=\gamma. If α\alpha and β\beta intersect transversally at only \ast as shown in Figure 5, then the relation 𝒫(γ)=𝒫(β)𝒫(α)\mathcal{P}_{\otimes}(\gamma)=\mathcal{P}_{\otimes}(\beta)\mathcal{P}_{\otimes}(\alpha) is obtained from the relarions (3) of Theorem 1.1 (see Lemma 5.4 in [12]). If α\alpha and β\beta intersect tangentially at only \ast as shown in Figure 5, then the relation 𝒫(γ)=𝒫(β)𝒫(α)\mathcal{P}_{\otimes}(\gamma)=\mathcal{P}_{\otimes}(\beta)\mathcal{P}_{\otimes}(\alpha) is obtained from the relations (1) and (5) of Theorem 1.1 (see Lemma 5.2 in [12]).

2.2. A finite presentation for 𝒯(Ng,n)\mathcal{T}(N_{g,n})

For g2g\geq 2 and n0n\geq 0 we have the short exact sequence

(3) 1𝒯(Ng,n)(Ng,n)/21\displaystyle 1\to\mathcal{T}(N_{g,n})\to\mathcal{M}(N_{g,n})\to\mathbb{Z}/{2\mathbb{Z}}\to 1

(see [16, 18]). Using the Reidemeister Schreier method for a presentation of (Ng,n)\mathcal{M}(N_{g,n}), we can obtain a presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}).

For n1n\leq 1, let α1,,αg1\alpha_{1},\dots,\alpha_{g-1}, β\beta, γ\gamma, ε\varepsilon and ζ\zeta be simple closed curves of Ng,nN_{g,n} with arrows as shown in Figure 8, and if gg is even, β0,,βg22\beta_{0},\dots,\beta_{\frac{g-2}{2}}, β¯g62\bar{\beta}_{\frac{g-6}{2}}, β¯g42\bar{\beta}_{\frac{g-4}{2}} and β¯g22\bar{\beta}_{\frac{g-2}{2}} simple closed curves of Ng,nN_{g,n} with arrows as shown in Figure 8. Let α\alpha be an oriented simple loop of Ng1,nN_{g-1,n} based at \ast induced from α1\alpha_{1} by the blowdown with respect to the first crosscap, as shown in Figure 8. For simplicity, we denote tαi=ait_{\alpha_{i}}=a_{i}, tβ=bt_{\beta}=b, tγ=ct_{\gamma}=c, tε=et_{\varepsilon}=e, tζ=ft_{\zeta}=f, tβi=bit_{\beta_{i}}=b_{i}, tβ¯i=b¯it_{\bar{\beta}_{i}}=\bar{b}_{i} and 𝒫(α)=y\mathcal{P}_{\otimes}(\alpha)=y. Note that y2=tδy^{2}=t_{\delta}, where δ\delta is a simple close curve of Ng,nN_{g,n} with an arrow as shown in Figure 8. For g1g\geq 1 and n1n\leq 1, 𝒯(Ng,n)\mathcal{T}(N_{g,n}) admits a finite presentation given in Theorems 2.5, 2.6 or Lemma 2.7.

Refer to caption
Figure 8.
Theorem 2.5 (Theorem 3.1 in [26]).

If g3g\geq 3 is odd or g=4g=4, then 𝒯(Ng,1)\mathcal{T}(N_{g,1}) admits a presentation with generators a1,,ag1a_{1},\dots,a_{g-1}, ee, ff, y2y^{2} and bb, cc for g4g\geq 4. The defining relations are

  • (A1)

    aiaj=ajaia_{i}a_{j}=a_{j}a_{i} for g4g\geq 4, |ij|>1|i-j|>1,

  • (A2)

    aiai+1ai=ai+1aiai+1a_{i}a_{i+1}a_{i}=a_{i+1}a_{i}a_{i+1} for i=1,,g2i=1,\dots,g-2,

  • (A3)

    aib=baia_{i}b=ba_{i} for g4g\geq 4, i4i\neq 4,

  • (A4)

    ba4b=a4ba4ba_{4}b=a_{4}ba_{4} for g5g\geq 5,

  • (A5)

    (a2a3a4b)10=(a1a2a3a4b)6(a_{2}a_{3}a_{4}b)^{10}=(a_{1}a_{2}a_{3}a_{4}b)^{6} for g5g\geq 5,

  • (A6)

    (a2a3a4a5a6b)12=(a1a2a3a4a5a6b)9(a_{2}a_{3}a_{4}a_{5}a_{6}b)^{12}=(a_{1}a_{2}a_{3}a_{4}a_{5}a_{6}b)^{9} for g7g\geq 7,

  • (A1¯1\overline{\mathrm{A1}}_{1})

    eaj=ajeea_{j}=a_{j}e for g5g\geq 5, j4j\geq 4,

  • (A1¯2\overline{\mathrm{A1}}_{2})

    faj=ajffa_{j}=a_{j}f for g5g\geq 5, j4j\geq 4,

  • (A2¯1\overline{\mathrm{A2}}_{1})

    a1ea1=ea1ea_{1}ea_{1}=ea_{1}e,

  • (A2¯2\overline{\mathrm{A2}}_{2})

    a31ea31=ea31ea_{3}^{-1}ea_{3}^{-1}=ea_{3}^{-1}e for g4g\geq 4,

  • (A2¯3\overline{\mathrm{A2}}_{3})

    a1fa1=fa1fa_{1}fa_{1}=fa_{1}f,

  • (A3¯1\overline{\mathrm{A3}}_{1})

    a1c=ca1a_{1}c=ca_{1} for g=4g=4, 55,

  • (A3¯2\overline{\mathrm{A3}}_{2})

    ec=ceec=ce for g=4g=4, 55,

  • (A4¯\overline{\mathrm{A4}})

    ca4c=a4ca4ca_{4}c=a_{4}ca_{4} for g=5g=5, 66,

  • (A5¯\overline{\mathrm{A5}})

    (e1a3a4c)10=(a11e1a3a4c)6(e^{-1}a_{3}a_{4}c)^{10}=(a_{1}^{-1}e^{-1}a_{3}a_{4}c)^{6} for g=5g=5, 66,

  • (A6¯\overline{\mathrm{A6}})

    (e1a3a4a5a6c)12=(a11e1a3a4a5a6c)9(e^{-1}a_{3}a_{4}a_{5}a_{6}c)^{12}=(a_{1}^{-1}e^{-1}a_{3}a_{4}a_{5}a_{6}c)^{9} for g=7g=7, 88,

  • (B1¯\overline{\mathrm{B1}})

    (a2a3a1a2ea1a31e)(a2a3a1a2fa1a31f)=1(a_{2}a_{3}a_{1}a_{2}ea_{1}a_{3}^{-1}e)(a_{2}a_{3}a_{1}a_{2}fa_{1}a_{3}^{-1}f)=1 for g4g\geq 4,

  • (B2¯1\overline{\mathrm{B2}}_{1})

    y2=a2a1ea1a2a1a2a1a2fa1a2y^{2}=a_{2}a_{1}ea_{1}a_{2}a_{1}a_{2}a_{1}a_{2}fa_{1}a_{2},

  • (B2¯2\overline{\mathrm{B2}}_{2})

    (a2a1ea1a2a1a2a1a2fa1a2)(a2a1fa1a2a1a2a1a2ea1a2)=1(a_{2}a_{1}ea_{1}a_{2}a_{1}a_{2}a_{1}a_{2}fa_{1}a_{2})(a_{2}a_{1}fa_{1}a_{2}a_{1}a_{2}a_{1}a_{2}ea_{1}a_{2})=1,

  • (B3¯\overline{\mathrm{B3}})

    y2a3=a3y2y^{2}a_{3}=a_{3}y^{2} for g4g\geq 4,

  • (B4¯1\overline{\mathrm{B4}}_{1})

    ea2=a2eea_{2}=a_{2}e,

  • (B4¯2\overline{\mathrm{B4}}_{2})

    fa2=a2ffa_{2}=a_{2}f,

  • (B6¯1\overline{\mathrm{B6}}_{1})

    bc=(a1a2a3f1a31a21a11)(a21a31e1a3a2)bc=(a_{1}a_{2}a_{3}f^{-1}a_{3}^{-1}a_{2}^{-1}a_{1}^{-1})(a_{2}^{-1}a_{3}^{-1}e^{-1}a_{3}a_{2}) for g4g\geq 4,

  • (B6¯2\overline{\mathrm{B6}}_{2})

    c(y2by2)=(a11e1a3a2a31ea1)(ea31y2a2y2a3e1)c(y^{2}by^{-2})=(a_{1}^{-1}e^{-1}a_{3}a_{2}a_{3}^{-1}ea_{1})(ea_{3}^{-1}y^{2}a_{2}y^{-2}a_{3}e^{-1}) for g=4g=4, 55,

  • (B7¯1\overline{\mathrm{B7}}_{1})

    (a4a5a3a4a2a3a1a2ea1a31ea41a31a51a41)c=b(a4a5a3a4a2a3a1a2ea1a31ea41a31a51a41)(a_{4}a_{5}a_{3}a_{4}a_{2}a_{3}a_{1}a_{2}ea_{1}a_{3}^{-1}ea_{4}^{-1}a_{3}^{-1}a_{5}^{-1}a_{4}^{-1})c\\ =b(a_{4}a_{5}a_{3}a_{4}a_{2}a_{3}a_{1}a_{2}ea_{1}a_{3}^{-1}ea_{4}^{-1}a_{3}^{-1}a_{5}^{-1}a_{4}^{-1}) for g6g\geq 6,

  • (B7¯2\overline{\mathrm{B7}}_{2})

    (a21a11a31a21a41a31a51a41)b(a4a5a3a4a2a3a1a2)y2=y2(a21a11a31a21a41a31a51a41)b(a4a5a3a4a2a3a1a2)(a_{2}^{-1}a_{1}^{-1}a_{3}^{-1}a_{2}^{-1}a_{4}^{-1}a_{3}^{-1}a_{5}^{-1}a_{4}^{-1})b(a_{4}a_{5}a_{3}a_{4}a_{2}a_{3}a_{1}a_{2})y^{2}\\ =y^{2}(a_{2}^{-1}a_{1}^{-1}a_{3}^{-1}a_{2}^{-1}a_{4}^{-1}a_{3}^{-1}a_{5}^{-1}a_{4}^{-1})b(a_{4}a_{5}a_{3}a_{4}a_{2}a_{3}a_{1}a_{2}) for g6g\geq 6,

  • (B8¯1\overline{\mathrm{B8}}_{1})

    (a1ea31a41ca4a3e1a11)(a11a21a31a41b1a4a3a2a1)=a41(a31a21e1a3a4a31ea2a3)a21e1(a_{1}ea_{3}^{-1}a_{4}^{-1}ca_{4}a_{3}e^{-1}a_{1}^{-1})(a_{1}^{-1}a_{2}^{-1}a_{3}^{-1}a_{4}^{-1}b^{-1}a_{4}a_{3}a_{2}a_{1})\\ =a_{4}^{-1}(a_{3}^{-1}a_{2}^{-1}e^{-1}a_{3}a_{4}a_{3}^{-1}ea_{2}a_{3})a_{2}^{-1}e^{-1} for g5g\geq 5,

  • (B8¯2\overline{\mathrm{B8}}_{2})

    (a11a21a31a41ba4a3a2a1)(a1fa31a41y2c1y2a4a3f1a11)=a41(a31fa2a3a4a31a21f1a3)fa2(a_{1}^{-1}a_{2}^{-1}a_{3}^{-1}a_{4}^{-1}ba_{4}a_{3}a_{2}a_{1})(a_{1}fa_{3}^{-1}a_{4}^{-1}y^{-2}c^{-1}y^{2}a_{4}a_{3}f^{-1}a_{1}^{-1})\\ =a_{4}^{-1}(a_{3}^{-1}fa_{2}a_{3}a_{4}a_{3}^{-1}a_{2}^{-1}f^{-1}a_{3})fa_{2} for g=5g=5, 66.

If g6g\geq 6 is even, then 𝒯(Ng,1)\mathcal{T}(N_{g,1}) admits a presentation with generators a1,,ag1a_{1},\dots,a_{g-1}, ee, ff, y2y^{2}, bb, cc and additionally b0,,bg22b_{0},\dots,b_{\frac{g-2}{2}}, b¯g62\bar{b}_{\frac{g-6}{2}}, b¯g42\bar{b}_{\frac{g-4}{2}}, b¯g22\bar{b}_{\frac{g-2}{2}}. The defining relations are relations (A1)(\mathrm{A1})-(A6)(\mathrm{A6}), (A1¯1)(\overline{\mathrm{A1}}_{1})-(A6¯)(\overline{\mathrm{A6}}), (B1¯)(\overline{\mathrm{B1}})-(B8¯2)(\overline{\mathrm{B8}}_{2}) and additionally

  • (A7)

    b0=a1b_{0}=a_{1}, b1=bb_{1}=b,

  • (A8)

    bi+1=(bi1a2ia2i+1a2i+2a2i+3bi)5(bi1a2ia2i+1a2i+2a2i+3)6b_{i+1}=(b_{i-1}a_{2i}a_{2i+1}a_{2i+2}a_{2i+3}b_{i})^{5}(b_{i-1}a_{2i}a_{2i+1}a_{2i+2}a_{2i+3})^{-6} for 1ig421\leq{i}\leq\frac{g-4}{2},

  • (A9a)

    b2b=bb2b_{2}b=bb_{2} for g=6g=6,

  • (A9b)

    bg22ag5=ag5bg22b_{\frac{g-2}{2}}a_{g-5}=a_{g-5}b_{\frac{g-2}{2}} for g8g\geq 8,

  • (A7a¯\overline{\mathrm{A7a}})

    b¯0=a11\bar{b}_{0}=a_{1}^{-1}, b¯1=c\bar{b}_{1}=c for g=6g=6,

  • (A7b¯\overline{\mathrm{A7b}})

    b¯1=c\bar{b}_{1}=c for g=8g=8,

  • (A7c¯\overline{\mathrm{A7c}})

    b¯i=zg1bizg11\bar{b}_{i}=z_{g-1}b_{i}z_{g-1}^{-1} where i=g62i=\frac{g-6}{2}, g42\frac{g-4}{2}, i2i\geq 2 and zg1=(ag1agag2ag1a3a4e1a3a11e1)(a21a11ag11ag21ag1ag11)z_{g-1}=(a_{g-1}a_{g}a_{g-2}a_{g-1}\cdots{}a_{3}a_{4}e^{-1}a_{3}a_{1}^{-1}e^{-1})(a_{2}^{-1}a_{1}^{-1}\cdots{}a_{g-1}^{-1}a_{g-2}^{-1}a_{g}^{-1}a_{g-1}^{-1}),

  • (A8a¯\overline{\mathrm{A8a}})

    b¯2=(b¯0e1a3a4a5b¯1)5(b¯0e1a3a4a5)6\bar{b}_{2}=(\bar{b}_{0}e^{-1}a_{3}a_{4}a_{5}\bar{b}_{1})^{5}(\bar{b}_{0}e^{-1}a_{3}a_{4}a_{5})^{-6} for g=6g=6,

  • (A8b¯\overline{\mathrm{A8b}})

    b¯g22=(b¯g62ag4ag3ag2ag1b¯g42)5(b¯g62ag4ag3ag2ag1)6\bar{b}_{\frac{g-2}{2}}=(\bar{b}_{\frac{g-6}{2}}a_{g-4}a_{g-3}a_{g-2}a_{g-1}\bar{b}_{\frac{g-4}{2}})^{5}(\bar{b}_{\frac{g-6}{2}}a_{g-4}a_{g-3}a_{g-2}a_{g-1})^{-6} for g8g\geq 8,

  • (A9a¯\overline{\mathrm{A9a}})

    b¯2c=cb¯2\bar{b}_{2}c=c\bar{b}_{2} for g=6g=6,

  • (A9b¯\overline{\mathrm{A9b}})

    b¯g22ag5=ag5b¯g22\bar{b}_{\frac{g-2}{2}}a_{g-5}=a_{g-5}\bar{b}_{\frac{g-2}{2}} for g8g\geq 8.

Theorem 2.6 (Theorem 3.2 in [26]).

If g5g\geq 5 is odd, then the group 𝒯(Ng,0)\mathcal{T}(N_{g,0}) is isomorphic to the quotient of the group 𝒯(Ng,1)\mathcal{T}(N_{g,1}) with the presentation given in Theorem 2.5 obtained by adding a generator ϱ\varrho and relations

  • (C1a\mathrm{C1a})

    (a1a2ag1)g=ϱ(a_{1}a_{2}\cdots{}a_{g-1})^{g}=\varrho,

  • (C1a¯\overline{\mathrm{C1a}})

    (a11e1a3ag1)g=y2ϱ(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g}=y^{2}\varrho,

  • (C2\mathrm{C2})

    aiϱ=ϱaia_{i}\varrho=\varrho{}a_{i} for 1ig11\leq{i}\leq{}g-1,

  • (C2¯\overline{\mathrm{C2}})

    ϱe=fϱ\varrho{}e=f\varrho,

  • (C5¯1\overline{\mathrm{C5}}_{1})

    ϱy2=y2ϱ\varrho{}y^{2}=y^{-2}\varrho,

  • (C3\mathrm{C3})

    ϱ2=1\varrho^{2}=1,

  • (C4a¯\overline{\mathrm{C4a}})

    (a2a3ag1e1a3ag1)g12=1(a_{2}a_{3}\cdots{}a_{g-1}e^{-1}a_{3}\cdots{}a_{g-1})^{\frac{g-1}{2}}=1.

Moreover, relations (A1¯2)(\overline{\mathrm{A1}}_{2}), (B2¯2)(\overline{\mathrm{B2}}_{2}), (B4¯2)(\overline{\mathrm{B4}}_{2}) are superfluous.

If g4g\geq 4 is even, then the group 𝒯(Ng,0)\mathcal{T}(N_{g,0}) is isomorphic to the quotient of the group 𝒯(Ng,1)\mathcal{T}(N_{g,1}) with the presentation given in Theorem 2.5 obtained by adding a generator ϱ¯\bar{\varrho} and relations

  • (C1b\mathrm{C1b})

    (a1a2ag1)g=1(a_{1}a_{2}\cdots{}a_{g-1})^{g}=1,

  • (C2¯1\overline{\mathrm{C2}}_{1})

    ϱ¯a1=a11ϱ¯\bar{\varrho}a_{1}=a_{1}^{-1}\bar{\varrho},

  • (C2¯2\overline{\mathrm{C2}}_{2})

    ϱ¯ai=aiϱ¯\bar{\varrho}a_{i}=a_{i}\bar{\varrho} for 3ig13\leq{i}\leq{}g-1,

  • (C2¯3\overline{\mathrm{C2}}_{3})

    ϱ¯a2=e1ϱ¯\bar{\varrho}a_{2}=e^{-1}\bar{\varrho},

  • (C5¯2\overline{\mathrm{C5}}_{2})

    ϱ¯y2=y2ϱ¯\bar{\varrho}y^{2}=y^{-2}\bar{\varrho},

  • (C3¯\overline{\mathrm{C3}})

    ϱ¯2=1\bar{\varrho}^{2}=1,

  • (C4¯\overline{\mathrm{C4}})

    (ϱ¯a2a3ag1)g1=1(\bar{\varrho}a_{2}a_{3}\cdots{}a_{g-1})^{g-1}=1.

Moreover, relations (A1¯1)(\overline{\mathrm{A1}}_{1}), (A2¯1)(\overline{\mathrm{A2}}_{1}), (A2¯2)(\overline{\mathrm{A2}}_{2}) are superfluous.

Lemma 2.7.
  1. (1)

    𝒯(N1,0)\mathcal{T}(N_{1,0}) and 𝒯(N1,1)\mathcal{T}(N_{1,1}) are trivial.

  2. (2)

    𝒯(N2,0)=a1a12\mathcal{T}(N_{2,0})=\langle{a_{1}}\mid{a_{1}^{2}}\rangle.

  3. (3)

    𝒯(N2,1)=a1,y2a1y2a11y2\mathcal{T}(N_{2,1})=\langle{a_{1},y^{2}}\mid{a_{1}y^{2}a_{1}^{-1}y^{-2}}\rangle.

  4. (4)

    𝒯(N3,0)=a1,a2a1a2a1a21a11a21,(a1a2)6\mathcal{T}(N_{3,0})=\langle{a_{1},a_{2}}\mid{a_{1}a_{2}a_{1}a_{2}^{-1}a_{1}^{-1}a_{2}^{-1},(a_{1}a_{2})^{6}}\rangle.

Proof.

Note that /2\mathbb{Z}/{2\mathbb{Z}} is generated by the image of yy, in the sequence (3) at the beginning of Section 2.2.

  1. (1)

    (N1,0)\mathcal{M}(N_{1,0}) and (N1,1)\mathcal{M}(N_{1,1}) are trivial (see Theorem 3.4 in [5]). Thus 𝒯(N1,0)\mathcal{T}(N_{1,0}) and 𝒯(N1,1)\mathcal{T}(N_{1,1}) are also trivial.

  2. (2)

    We have the presentation

    (N2,0)=a1,ya12,y2,(a1y)2\mathcal{M}(N_{2,0})=\langle{a_{1},y}\mid{a_{1}^{2},y^{2},(a_{1}y)^{2}}\rangle

    (see Lemma 5 in [16]). Using the Reidemeister Schreier method for this presentation, we obtain the presentation

    𝒯(N2,0)\displaystyle\mathcal{T}(N_{2,0}) =\displaystyle= a1,y2a12,y2,a11y2a1\displaystyle\langle{a_{1},y^{2}}\mid{a_{1}^{2},y^{2},a_{1}^{-1}y^{2}a_{1}}\rangle
    =\displaystyle= a1a12.\displaystyle\langle{a_{1}}\mid{a_{1}^{2}}\rangle.
  3. (3)

    We have the presentation

    (N2,1)=a1,yya1y1a1\mathcal{M}(N_{2,1})=\langle{a_{1},y}\mid{ya_{1}y^{-1}a_{1}}\rangle

    (see Theorem A.7 in [23]). Using the Reidemeister Schreier method for this presentation, we obtain the presentation

    𝒯(N2,1)=a1,y2a1y2a11y2.\mathcal{T}(N_{2,1})=\langle{a_{1},y^{2}}\mid{a_{1}y^{2}a_{1}^{-1}y^{-2}}\rangle.
  4. (4)

    We have the presentation

    (N3,0)=a1,a2,ya1a2a1a21a11a21,(a1a2)6,y2,(a1y)2,(a2y)2\mathcal{M}(N_{3,0})=\langle{a_{1},a_{2},y}\mid{a_{1}a_{2}a_{1}a_{2}^{-1}a_{1}^{-1}a_{2}^{-1},(a_{1}a_{2})^{6},y^{2},(a_{1}y)^{2},(a_{2}y)^{2}}\rangle

    (see Theorem 3 in [2]). Using the Reidemeister Schreier method for this presentation, we obtain the presentation

    𝒯(N3,0)\displaystyle\mathcal{T}(N_{3,0}) =\displaystyle= a1,a2,y2a1a2a1a21a11a21,(a1a2)6,y2,a11y2a1,a21y2a2\displaystyle\langle{a_{1},a_{2},y^{2}}\mid{a_{1}a_{2}a_{1}a_{2}^{-1}a_{1}^{-1}a_{2}^{-1},(a_{1}a_{2})^{6},y^{2},a_{1}^{-1}y^{2}a_{1},a_{2}^{-1}y^{2}a_{2}}\rangle
    =\displaystyle= a1,a2a1a2a1a21a11a21,(a1a2)6.\displaystyle\langle{a_{1},a_{2}}\mid{a_{1}a_{2}a_{1}a_{2}^{-1}a_{1}^{-1}a_{2}^{-1},(a_{1}a_{2})^{6}}\rangle.

3. Proof of Theorem 1.1 for g1g\geq 1 and n1n\leq 1

In this section, we prove Theorem 1.1 for g1g\geq 1 and n1n\leq 1, using the presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) given in Theorems 2.5, 2.6 and Lemma 2.7.

Let XY\langle{X}\mid{Y}\rangle be the infinitely presented group with the presentation given in Theorem 1.1, and X0Y0\langle{X_{0}}\mid{Y_{0}}\rangle the presentation for 𝒯(Ng,n)\mathcal{T}(N_{g,n}) given in Theorems 2.5, 2.6 or Lemma 2.7. Denote by F(X0)F(X_{0}) the free group freely generated by X0X_{0}. Let p:F(X0)X0Y0p:F(X_{0})\to\langle{X_{0}}\mid{Y_{0}}\rangle be the natural projection and η:F(X0)XY\eta:F(X_{0})\to\langle{X}\mid{Y}\rangle the homomorphism defined as η(x)=x\eta(x)=x for any xX0x\in{}X_{0}. We consider a correspondence

ψ:X0Y0XY\psi:\langle{X_{0}}\mid{Y_{0}}\rangle\to\langle{X}\mid{Y}\rangle

satisfying ψp=η\psi\circ{}p=\eta. Showing the following proposition, we will obtain Theorem 1.1 for g1g\geq 1 and n1n\leq 1.

Proposition 3.1.

For g1g\geq 1 and n1n\leq 1, ψ\psi is an isomorphism.

Let φ:XYX0Y0=𝒯(Ng,n)\varphi:\langle{X}\mid{Y}\rangle\to\langle{X_{0}}\mid{Y_{0}}\rangle=\mathcal{T}(N_{g,n}) be the homomorphism defined as φ(tc;θ)=tc;θ\varphi(t_{c;\theta})=t_{c;\theta} for any tc;θXt_{c;\theta}\in{X}. Since φ(Y)=1\varphi(Y)=1 in 𝒯(Ng,n)\mathcal{T}(N_{g,n}) clearly, φ\varphi is well-defined. By the definitions of ψ\psi and φ\varphi, it is clear that φψ\varphi\circ\psi is the identity map if ψ\psi is a homomorphism. Hence in order to prove Proposition 3.1, what we need is to show well-definedness and surjectivity of ψ\psi. In Sections 3.1 and 3.2, we see well-definedness and surjectivity of ψ\psi respectively.

3.1. Well-definedness of ψ\psi

In order to see well-definedness of ψ\psi, it suffices to show that ψ(r)=1\psi(r)=1 in XY\langle{X}\mid{Y}\rangle for any rY0r\in{}Y_{0}. More precisely, we check that any relation and relator of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) in Theorems 2.5, 2.6 and Lemma 2.7 are obtained from the relations (1)-(5) of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) in Theorem 1.1 and the relation (4)(4)^{\prime} assigned in Remark 1.3.

First we notice that the relations (A7)(\mathrm{A7}), (A7a¯)(\overline{\mathrm{A7a}}) and (A7b¯)(\overline{\mathrm{A7b}}) hold since each of the pairs (b0,a1)(b_{0},a_{1}), (b1,b)(b_{1},b), (b¯0,a11)(\bar{b}_{0},a_{1}^{-1}) for g=6g=6, and (b¯1,c)(\bar{b}_{1},c) for g=6g=6, 88 coincides as mapping classes from the definition of loops in Figure 8. We may also treat the relation (C1a)(\mathrm{C1a}) as a definition of ϱ\varrho. It is easy to check that any relator of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) in Lemma 2.7 is obtained from the relations (1)-(4) in Theorem 1.1. In particular, for the relator a2a^{2} of 𝒯(N2,0)\mathcal{T}(N_{2,0}), recall Remark 1.2. The relations (A1)(\mathrm{A1}), (A2)(\mathrm{A2}), (A3)(\mathrm{A3}), (A4)(\mathrm{A4}), (A1¯1)(\overline{\mathrm{A1}}_{1}), (A1¯2)(\overline{\mathrm{A1}}_{2}), (A2¯1)(\overline{\mathrm{A2}}_{1}), (A2¯2)(\overline{\mathrm{A2}}_{2}), (A2¯3)(\overline{\mathrm{A2}}_{3}), (A3¯1)(\overline{\mathrm{A3}}_{1}), (A3¯2)(\overline{\mathrm{A3}}_{2}), (A4¯)(\overline{\mathrm{A4}}), (B3¯)(\overline{\mathrm{B3}}), (B4¯1)(\overline{\mathrm{B4}}_{1}), (B4¯2)(\overline{\mathrm{B4}}_{2}), (B7¯1)(\overline{\mathrm{B7}}_{1}), (B7¯2)(\overline{\mathrm{B7}}_{2}), (A9a)(\mathrm{A9a}), (A9b)(\mathrm{A9b}), (A7c¯)(\overline{\mathrm{A7c}}), (A9a¯)(\overline{\mathrm{A9a}}), (A9b¯)(\overline{\mathrm{A9b}}) (C2)(\mathrm{C2}), (C2¯)(\overline{\mathrm{C2}}), (C5¯1)(\overline{\mathrm{C5}}_{1}), (C2¯1)(\overline{\mathrm{C2}}_{1}), (C2¯2)(\overline{\mathrm{C2}}_{2}), (C2¯3)(\overline{\mathrm{C2}}_{3}) and (C5¯2)(\overline{\mathrm{C5}}_{2}) are obtained by repeating the relation (3) in Theorem 1.1. The relations (A5)(\mathrm{A5}), (A6)(\mathrm{A6}), (A5¯)(\overline{\mathrm{A5}}), (A6¯)(\overline{\mathrm{A6}}), (A8)(\mathrm{A8}), (A8a¯)(\overline{\mathrm{A8a}}) and (A8b¯)(\overline{\mathrm{A8b}}) come from relations of mapping class groups of orientable surfaces (see Theorem 1.3 in [20]), and hence these relations are obtained from the relations (3), (4) and (5) in Theorem 1.1 by Theorem in [19]. The relations (C3)(\mathrm{C3}) and (C1b)(\mathrm{C1b}) are obtained from the relations (1) in Theorem 1.1 and (4)(4)^{\prime} assigned in Remark 1.3. Thus it suffices to show that the relations (B1¯)(\overline{\mathrm{B1}}), (B2¯1)(\overline{\mathrm{B2}}_{1}), (B2¯2)(\overline{\mathrm{B2}}_{2}), (B6¯1)(\overline{\mathrm{B6}}_{1}), (B6¯2)(\overline{\mathrm{B6}}_{2}), (B8¯1)(\overline{\mathrm{B8}}_{1}), (B8¯2)(\overline{\mathrm{B8}}_{2}), (C1a¯)(\overline{\mathrm{C1a}}), (C4a¯)(\overline{\mathrm{C4a}}), (C3¯)(\overline{\mathrm{C3}}) and (C4¯)(\overline{\mathrm{C4}}) are satisfied in XY\langle{X\mid{}Y}\rangle.

Recall the simple closed curves and the simple loop as shown in Figure 8.

3.1.1. On the relation (B1¯)(\overline{\mathrm{B1}})

The relation (B1¯)(\overline{\mathrm{B1}}) can be rewritten as follows.

(a2a3a1a2ea1a31e)(a2¯(B4¯1)a3a1¯(A1)a2fa1a31f)=1\displaystyle(a_{2}a_{3}a_{1}a_{2}ea_{1}a_{3}^{-1}\underset{(\overline{\mathrm{B4}}_{1})}{\underline{e)(a_{2}}}\underset{(\mathrm{A1})}{\underline{a_{3}a_{1}}}a_{2}fa_{1}a_{3}^{-1}f)=1
\displaystyle\iff a2a3¯a1(a2ea1a31a2ea1a3a2fa1a31fa2¯(B4¯2)a3a1¯(A1))a11a31a21¯=1\displaystyle\underline{a_{2}a_{3}}a_{1}(a_{2}ea_{1}a_{3}^{-1}a_{2}ea_{1}a_{3}a_{2}fa_{1}a_{3}^{-1}\underset{(\overline{\mathrm{B4}}_{2})}{\underline{fa_{2}}}\underset{(\mathrm{A1})}{\underline{a_{3}a_{1}}})a_{1}^{-1}\underline{a_{3}^{-1}a_{2}^{-1}}=1
\displaystyle\iff a1{(a2ea1)a31(a2ea1)a3(a2fa1)a31(a2fa1)a3}a11=1.\displaystyle a_{1}\{(a_{2}ea_{1})a_{3}^{-1}(a_{2}ea_{1})a_{3}(a_{2}fa_{1})a_{3}^{-1}(a_{2}fa_{1})a_{3}\}a_{1}^{-1}=1.

Similarly, the relation (B2¯1)(\overline{\mathrm{B2}}_{1}) can be rewritten as follows.

y2\displaystyle y^{2} =\displaystyle= a2a1ea1¯(A2¯1)a2a1a2a1a2f¯(B4¯2)a1a2\displaystyle a_{2}\underset{(\overline{\mathrm{A2}}_{1})}{\underline{a_{1}ea_{1}}}a_{2}a_{1}a_{2}a_{1}\underset{(\overline{\mathrm{B4}}_{2})}{\underline{a_{2}f}}a_{1}a_{2}
=\displaystyle= a2ea1ea2¯(B4¯1)a1a2a1fa2a1a2¯(A2)\displaystyle a_{2}ea_{1}\underset{(\overline{\mathrm{B4}}_{1})}{\underline{ea_{2}}}a_{1}a_{2}a_{1}f\underset{(\mathrm{A2})}{\underline{a_{2}a_{1}a_{2}}}
=\displaystyle= a2ea1a2ea1a2a1fa1¯(A2¯3)a2a1\displaystyle a_{2}ea_{1}a_{2}ea_{1}a_{2}\underset{(\overline{\mathrm{A2}}_{3})}{\underline{a_{1}fa_{1}}}a_{2}a_{1}
=\displaystyle= a2ea1a2ea1a2fa1fa2¯(B4¯2)a1\displaystyle a_{2}ea_{1}a_{2}ea_{1}a_{2}fa_{1}\underset{(\overline{\mathrm{B4}}_{2})}{\underline{fa_{2}}}a_{1}
=\displaystyle= (a2ea1)2(a2fa1)2.\displaystyle(a_{2}ea_{1})^{2}(a_{2}fa_{1})^{2}.

Let AA, BB, CC, DD and EE be simple closed curves with arrows as shown in Figure 9. Repeating the relation (3), we have tA=a1(a2ea1)a31(a2ea1)1a11t_{A}=a_{1}(a_{2}ea_{1})a_{3}^{-1}(a_{2}ea_{1})^{-1}a_{1}^{-1}, tB=a1(a2ea1)2a3(a2ea1)2a11t_{B}=a_{1}(a_{2}ea_{1})^{2}a_{3}(a_{2}ea_{1})^{-2}a_{1}^{-1} and tC=a1(a2fa1)1a31(a2fa1)a11t_{C}=a_{1}(a_{2}fa_{1})^{-1}a_{3}^{-1}(a_{2}fa_{1})a_{1}^{-1}. Assuming that (B2¯1)(\overline{\mathrm{B2}}_{1}) holds in XY\langle{X\mid{}Y}\rangle, we calculate

a1{(a2ea1)a31(a2ea1)a3(a2fa1)a31(a2fa1)a3}a11\displaystyle a_{1}\{(a_{2}ea_{1})a_{3}^{-1}(a_{2}ea_{1})a_{3}(a_{2}fa_{1})a_{3}^{-1}(a_{2}fa_{1})a_{3}\}a_{1}^{-1} =\displaystyle= tAtBy2tC(a1a3a11)\displaystyle t_{A}t_{B}y^{2}t_{C}(a_{1}a_{3}a_{1}^{-1})
=(1),(3)\displaystyle\overset{(\ref{bound}),(\ref{t-conj})}{=} tAtBy2tCa3tDtE\displaystyle t_{A}t_{B}y^{2}t_{C}a_{3}t_{D}t_{E}
=(5)\displaystyle\overset{(\ref{lanterns})}{=} 1.\displaystyle 1.

Note that Bα3DEB\cup\alpha_{3}\cup{}D\cup{}E bounds Σ0,4\Sigma_{0,4}. Thus the relation (B1¯)(\overline{\mathrm{B1}}) is satisfied in XY\langle{X\mid{}Y}\rangle, by Section 3.1.2.

Refer to caption
Figure 9.

3.1.2. On the relations (B2¯1)(\overline{\mathrm{B2}}_{1}) and (B2¯2)(\overline{\mathrm{B2}}_{2})

Let AA, BB, CC and DD be simple closed curves with arrows as shown in Figure 10. Repeating the relation (3), we have tA=a2a1ea11a21t_{A}=a_{2}a_{1}ea_{1}^{-1}a_{2}^{-1} and tB=a2a1f1a11a21t_{B}=a_{2}a_{1}f^{-1}a_{1}^{-1}a_{2}^{-1}. In addition, by the relations (1) and (4)(4)^{\prime} we have tC=(fa1a2)4t_{C}=(fa_{1}a_{2})^{4}. Hence we calculate

y2\displaystyle y^{2} =(5)\displaystyle\overset{(\ref{lanterns})}{=} tAtBtCtDa1a11\displaystyle t_{A}t_{B}t_{C}t_{D}a_{1}a_{1}^{-1}
=(1)\displaystyle\overset{(\ref{bound})}{=} (a2a1ea11a21)(a2a1f1a11a21)(fa1a2)4\displaystyle(a_{2}a_{1}ea_{1}^{-1}a_{2}^{-1})(a_{2}a_{1}f^{-1}a_{1}^{-1}a_{2}^{-1})(fa_{1}a_{2})^{4}
=\displaystyle= a2a1ef1a11a21f¯(B4¯2)a1a2fa1a2f¯(B4¯2)a1a2fa1a2\displaystyle a_{2}a_{1}ef^{-1}a_{1}^{-1}\underset{(\overline{\mathrm{B4}}_{2})}{\underline{a_{2}^{-1}f}}a_{1}a_{2}fa_{1}\underset{(\overline{\mathrm{B4}}_{2})}{\underline{a_{2}f}}a_{1}a_{2}fa_{1}a_{2}
=\displaystyle= a2a1ef1a11f¯(A2¯3)a21a1a2fa1f¯(A2¯3)a2a1a2fa1a2\displaystyle a_{2}a_{1}e\underset{(\overline{\mathrm{A2}}_{3})}{\underline{f^{-1}a_{1}^{-1}f}}a_{2}^{-1}a_{1}a_{2}\underset{(\overline{\mathrm{A2}}_{3})}{\underline{fa_{1}f}}a_{2}a_{1}a_{2}fa_{1}a_{2}
=\displaystyle= a2a1ea1f1a11a21a1a2a1f¯(A2),(B4¯2)a1a2a1a2fa1a2\displaystyle a_{2}a_{1}ea_{1}\underset{(\mathrm{A2}),(\overline{\mathrm{B4}}_{2})}{\underline{f^{-1}a_{1}^{-1}a_{2}^{-1}a_{1}a_{2}a_{1}f}}a_{1}a_{2}a_{1}a_{2}fa_{1}a_{2}
=\displaystyle= a2a1ea1a2a1a2a1a2fa1a2.\displaystyle a_{2}a_{1}ea_{1}a_{2}a_{1}a_{2}a_{1}a_{2}fa_{1}a_{2}.

Note that CDα1C\cup{}D\cup\alpha_{1} bounds Σ0,4\Sigma_{0,4}. Thus the relation (B2¯1)(\overline{\mathrm{B2}}_{1}) is satisfied in XY\langle{X\mid{}Y}\rangle.

Refer to caption
Figure 10.

Moreover, similar to the transformation of the relation (B2¯1)(\overline{\mathrm{B2}}_{1}) of Section 3.1.1, the relation (B2¯2)(\overline{\mathrm{B2}}_{2}) can be rewritten as

(a2ea1)2(a2fa1)4(a2ea1)2=1.(a_{2}ea_{1})^{2}(a_{2}fa_{1})^{4}(a_{2}ea_{1})^{2}=1.

By the relations (1) and (4)(4)^{\prime}, we have tC=(a2fa1)4t_{C}=(a_{2}fa_{1})^{4} and tC1=(a2ea1)4t_{C}^{-1}=(a_{2}ea_{1})^{4}. Hence we calculate

(a2ea1)2(a2fa1)4(a2ea1)2\displaystyle(a_{2}ea_{1})^{2}(a_{2}fa_{1})^{4}(a_{2}ea_{1})^{2} =\displaystyle= (a2ea1)2tC(a2ea1)2\displaystyle(a_{2}ea_{1})^{2}t_{C}(a_{2}ea_{1})^{2}
=(3)\displaystyle\overset{(\ref{t-conj})}{=} (a2ea1)4tC\displaystyle(a_{2}ea_{1})^{4}t_{C}
=\displaystyle= tC1tC\displaystyle t_{C}^{-1}t_{C}
=\displaystyle= 1.\displaystyle 1.

Thus the relation (B2¯2)(\overline{\mathrm{B2}}_{2}) is satisfied in XY\langle{X\mid{}Y}\rangle.

3.1.3. On the relations (B6¯1)(\overline{\mathrm{B6}}_{1}) and (B6¯2)(\overline{\mathrm{B6}}_{2})

Let AA, BB, CC, DD and EE be simple closed curves with arrows as shown in Figure 11. Repeating the relation (3), we have tA=a1a2a3f1a31a21a11t_{A}=a_{1}a_{2}a_{3}f^{-1}a_{3}^{-1}a_{2}^{-1}a_{1}^{-1} and tB=a21a31e1a3a2t_{B}=a_{2}^{-1}a_{3}^{-1}e^{-1}a_{3}a_{2}. Then we calculate

bcy2\displaystyle bcy^{2} =(5)\displaystyle\overset{(\ref{lanterns})}{=} tCa1a11a3\displaystyle t_{C}a_{1}a_{1}^{-1}a_{3}
=\displaystyle= tCa3,\displaystyle t_{C}a_{3},
tAtBy2\displaystyle t_{A}t_{B}y^{2} =(5)\displaystyle\overset{(\ref{lanterns})}{=} tCtDtEa3\displaystyle t_{C}t_{D}t_{E}a_{3}
=(1)\displaystyle\overset{(\ref{bound})}{=} tCa3,\displaystyle t_{C}a_{3},

and hence bc=tAtBbc=t_{A}t_{B}. Note that Cα1α3C\cup\alpha_{1}\cup\alpha_{3} and CDEα3C\cup{}D\cup{}E\cup\alpha_{3} bound Σ0,4\Sigma_{0,4}. Thus the relation (B6¯1)(\overline{\mathrm{B6}}_{1}) is satisfied in XY\langle{X\mid{}Y}\rangle.

Refer to caption
Figure 11.

In addition, repeating the relation (3), we have y2by2=ty(γ)y^{2}by^{-2}=t_{y(\gamma)}, a11e1a3a2a31ea1=ty(A)a_{1}^{-1}e^{-1}a_{3}a_{2}a_{3}^{-1}ea_{1}=t_{y(A)} and ea31y2a2y2a3e1=ty(B)ea_{3}^{-1}y^{2}a_{2}y^{-2}a_{3}e^{-1}=t_{y(B)}. Then we calculate

ty(β)ty(γ)y2\displaystyle t_{y(\beta)}t_{y(\gamma)}y^{2} =(5)\displaystyle\overset{(\ref{lanterns})}{=} tCa1a11a3\displaystyle t_{C}a_{1}a_{1}^{-1}a_{3}
=\displaystyle= tCa3,\displaystyle t_{C}a_{3},
ty(A)ty(B)y2\displaystyle t_{y(A)}t_{y(B)}y^{2} =(5)\displaystyle\overset{(\ref{lanterns})}{=} tCty(D)ty(E)a3\displaystyle t_{C}t_{y(D)}t_{y(E)}a_{3}
=(1)\displaystyle\overset{(\ref{bound})}{=} tCa3,\displaystyle t_{C}a_{3},

and hence ty(β)ty(γ)=ty(A)ty(B)t_{y(\beta)}t_{y(\gamma)}=t_{y(A)}t_{y(B)}. Note that Cy(D)y(E)α3=y(CDEα3)C\cup{}y(D)\cup{}y(E)\cup\alpha_{3}=y(C\cup{}D\cup{}E\cup\alpha_{3}) bounds Σ0,4\Sigma_{0,4}. Thus, since c=ty(β)c=t_{y(\beta)}, the relation (B6¯2)(\overline{\mathrm{B6}}_{2}) is satisfied in XY\langle{X\mid{}Y}\rangle.

3.1.4. On the relations (B8¯1)(\overline{\mathrm{B8}}_{1}) and (B8¯2)(\overline{\mathrm{B8}}_{2})

Let AA, BB, CC and DD be simple closed curves with arrows as shown in Figure 12. Repeating the relation (3), we have tA=(a1ea31a41)c(a4a3e1a11)t_{A}=(a_{1}ea_{3}^{-1}a_{4}^{-1})c(a_{4}a_{3}e^{-1}a_{1}^{-1}), tB=(a11a21a31a41)b1(a4a3a2a1)t_{B}=(a_{1}^{-1}a_{2}^{-1}a_{3}^{-1}a_{4}^{-1})b^{-1}(a_{4}a_{3}a_{2}a_{1}) and tC=(a31a21e1a3)a4(a31ea2a3)t_{C}=(a_{3}^{-1}a_{2}^{-1}e^{-1}a_{3})a_{4}(a_{3}^{-1}ea_{2}a_{3}). By the relation (5) we see

tAtDa2a4=tCe1tB1,t_{A}t_{D}a_{2}a_{4}=t_{C}e^{-1}t_{B}^{-1},

and by the relations (1) and (3) we see

tAtB=a41tCa21e1.t_{A}t_{B}=a_{4}^{-1}t_{C}a_{2}^{-1}e^{-1}.

Note that ADα2α4A\cup{}D\cup\alpha_{2}\cup\alpha_{4} bounds Σ0,4\Sigma_{0,4}. Thus the relation (B8¯1)(\overline{\mathrm{B8}}_{1}) is satisfied in XY\langle{X\mid{}Y}\rangle.

Refer to caption
Figure 12.

In addition, repeating the relation (3), we have (a11a21a31a41)b(a4a3a2a1)=ty1(A)(a_{1}^{-1}a_{2}^{-1}a_{3}^{-1}a_{4}^{-1})b(a_{4}a_{3}a_{2}a_{1})=t_{y^{-1}(A)}, (a1fa31a41)y2c1y2(a4a3f1a11)=ty1(B)(a_{1}fa_{3}^{-1}a_{4}^{-1})y^{-2}c^{-1}y^{2}(a_{4}a_{3}f^{-1}a_{1}^{-1})=t_{y^{-1}(B)} and (a31fa2a3)a4(a31a21f1a3)=ty1(C)(a_{3}^{-1}fa_{2}a_{3})a_{4}(a_{3}^{-1}a_{2}^{-1}f^{-1}a_{3})=t_{y^{-1}(C)}. By the relation (5) we see

ty1(A)ty1(D)f1a4=ty1(C)a2ty1(B)1,t_{y^{-1}(A)}t_{y^{-1}(D)}f^{-1}a_{4}=t_{y^{-1}(C)}a_{2}t_{y^{-1}(B)}^{-1},

and by the relations (1) and (3) we see

ty1(A)ty1(B)=a41ty1(C)fa2.t_{y^{-1}(A)}t_{y^{-1}(B)}=a_{4}^{-1}t_{y^{-1}(C)}fa_{2}.

Note that y1(A)y1(D)ζα4=y1(ADα2α4)y^{-1}(A)\cup{}y^{-1}(D)\cup\zeta\cup\alpha_{4}=y^{-1}(A\cup{}D\cup\alpha_{2}\cup\alpha_{4}) bounds Σ0,4\Sigma_{0,4}. Thus the relation (B8¯2)(\overline{\mathrm{B8}}_{2}) is satisfied in XY\langle{X\mid{}Y}\rangle.

3.1.5. On the relation (C1a¯)(\overline{\mathrm{C1a}})

We calculate

ϱ\displaystyle\varrho =(C1a)\displaystyle\overset{(\mathrm{C1a})}{=} (a1a2ag1)g\displaystyle(a_{1}a_{2}\cdots{}a_{g-1})^{g}
=\displaystyle= a1a2ag1(a1a2ag1)g2a1¯(A1),(A2)a2ag1\displaystyle a_{1}a_{2}\cdots{}a_{g-1}\underset{(\mathrm{A1}),(\mathrm{A2})}{\underline{(a_{1}a_{2}\cdots{}a_{g-1})^{g-2}a_{1}}}a_{2}\cdots{}a_{g-1}
=\displaystyle= a1a2ag1ag1(a1a2ag1)g2a2ag1\displaystyle a_{1}a_{2}\cdots{}a_{g-1}a_{g-1}(a_{1}a_{2}\cdots{}a_{g-1})^{g-2}a_{2}\cdots{}a_{g-1}
=\displaystyle= a1a2ag1ag1(a1a2ag1)g3a1¯(A1),(A2)(a2ag1)2\displaystyle a_{1}a_{2}\cdots{}a_{g-1}a_{g-1}\underset{(\mathrm{A1}),(\mathrm{A2})}{\underline{(a_{1}a_{2}\cdots{}a_{g-1})^{g-3}a_{1}}}(a_{2}\cdots{}a_{g-1})^{2}
=\displaystyle= a1a2ag1ag1ag2(a1a2ag1)g3(a2ag1)2\displaystyle a_{1}a_{2}\cdots{}a_{g-1}a_{g-1}a_{g-2}(a_{1}a_{2}\cdots{}a_{g-1})^{g-3}(a_{2}\cdots{}a_{g-1})^{2}
\displaystyle\vdots
=\displaystyle= a1a2a3ag1ag1a3a2a1(a2a3ag1)g1\displaystyle a_{1}a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2}a_{1}(a_{2}a_{3}\cdots{}a_{g-1})^{g-1}
=(A1),(A2)\displaystyle\overset{(\mathrm{A1}),(\mathrm{A2})}{=} (a2a3ag1)g1a1a2a3ag1ag1a3a2a1.\displaystyle(a_{2}a_{3}\cdots{}a_{g-1})^{g-1}a_{1}a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2}a_{1}.

Let ϱ=(a11e1a3ag1)g\varrho^{\prime}=(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g}. Similarly, we calculate

ϱ\displaystyle\varrho^{\prime} =\displaystyle= (a11e1a3ag1)g\displaystyle(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g}
=\displaystyle= a11e1a3ag1(a11e1a3ag1)g2a11¯(A1),(A2),(A1¯1),(A2¯1)(A2¯2)e1a3ag1\displaystyle a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1}\underset{(\mathrm{A1}),(\mathrm{A2}),(\overline{\mathrm{A1}}_{1}),(\overline{\mathrm{A2}}_{1})(\overline{\mathrm{A2}}_{2})}{\underline{(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g-2}a_{1}^{-1}}}e^{-1}a_{3}\cdots{}a_{g-1}
=\displaystyle= a11e1a3ag1ag1(a11e1a3ag1)g2e1a3ag1\displaystyle a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1}a_{g-1}(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g-2}e^{-1}a_{3}\cdots{}a_{g-1}
=\displaystyle= a11e1a3ag1ag1(a11e1a3ag1)g3a11¯(A1),(A2),(A1¯1),(A2¯1)(A2¯2)(e1a3ag1)2\displaystyle a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1}a_{g-1}\underset{(\mathrm{A1}),(\mathrm{A2}),(\overline{\mathrm{A1}}_{1}),(\overline{\mathrm{A2}}_{1})(\overline{\mathrm{A2}}_{2})}{\underline{(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g-3}a_{1}^{-1}}}(e^{-1}a_{3}\cdots{}a_{g-1})^{2}
=\displaystyle= a11e1a3ag1ag1ag2(a11e1a3ag1)g3(e1a3ag1)2\displaystyle a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1}a_{g-1}a_{g-2}(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1})^{g-3}(e^{-1}a_{3}\cdots{}a_{g-1})^{2}
\displaystyle\vdots
=\displaystyle= a11e1a3ag1ag1a3e1a11(e1a3ag1)g1\displaystyle a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}e^{-1}a_{1}^{-1}(e^{-1}a_{3}\cdots{}a_{g-1})^{g-1}
=(A1¯1),(A2¯1)(A2¯2)(A1),(A2)\displaystyle\underset{(\overline{\mathrm{A1}}_{1}),(\overline{\mathrm{A2}}_{1})(\overline{\mathrm{A2}}_{2})}{\overset{(\mathrm{A1}),(\mathrm{A2})}{=}} (a2a3ag1)g1a1a2a3ag1ag1a3a2a1.\displaystyle(a_{2}a_{3}\cdots{}a_{g-1})^{g-1}a_{1}a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2}a_{1}.

We show that ϱ=y2ϱ\varrho^{\prime}=y^{2}\varrho. Since a regular neighborhood of α1εα3αg1\alpha_{1}\cup\varepsilon\cup\alpha_{3}\cup\cdots\cup\alpha_{g-1} is diffeomorphic to Σg12,1\Sigma_{\frac{g-1}{2},1}, by the relations (1) and (4)(4)^{\prime}, we have (ϱ)2=1(\varrho^{\prime})^{2}=1. We calculate

ϱϱ1\displaystyle\varrho^{\prime}\varrho^{-1} =\displaystyle= (ϱ)1ϱ1\displaystyle(\varrho^{\prime})^{-1}\varrho^{-1}
=\displaystyle= a1ea31ag11ag11a31ea1(e1a3ag1)1gϱ1¯(C2),(C2¯)\displaystyle a_{1}ea_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}ea_{1}\underset{(\mathrm{C2}),(\overline{\mathrm{C2}})}{\underline{(e^{-1}a_{3}\cdots{}a_{g-1})^{1-g}\varrho^{-1}}}
=\displaystyle= a1ea31ag11ag11a31ea1ϱ1¯(f1a3ag1)1g\displaystyle a_{1}ea_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}ea_{1}\underline{\varrho^{-1}}(f^{-1}a_{3}\cdots{}a_{g-1})^{1-g}
=\displaystyle= a1ea31ag11ag11a31ea1¯\displaystyle a_{1}ea_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}\underline{ea_{1}}
a11a21¯a31ag11ag11a31a21a11(a2a3ag1)1g\displaystyle\underline{a_{1}^{-1}a_{2}^{-1}}a_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}a_{2}^{-1}a_{1}^{-1}(a_{2}a_{3}\cdots{}a_{g-1})^{1-g}
(f1a3ag1)1g\displaystyle(f^{-1}a_{3}\cdots{}a_{g-1})^{1-g}
=(B4¯1)\displaystyle\overset{(\overline{\mathrm{B4}}_{1})}{=} a1{ea2(a21a31ag11ag11a31a21)}2a11\displaystyle a_{1}\{ea_{2}(a_{2}^{-1}a_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}a_{2}^{-1})\}^{2}a_{1}^{-1}
(a2a3ag1)1g(f1a3ag1)1g.\displaystyle(a_{2}a_{3}\cdots{}a_{g-1})^{1-g}(f^{-1}a_{3}\cdots{}a_{g-1})^{1-g}.

It suffices to show the equality

y2(f1a3ag1)g1(a2a3ag1)g1\displaystyle y^{2}(f^{-1}a_{3}\cdots{}a_{g-1})^{g-1}(a_{2}a_{3}\cdots{}a_{g-1})^{g-1}
=a1{ea2(a21a31ag11ag11a31a21)}2a11.\displaystyle=a_{1}\{ea_{2}(a_{2}^{-1}a_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}a_{2}^{-1})\}^{2}a_{1}^{-1}.

Let γ1\gamma_{1}, γ2\gamma_{2} and γ3\gamma_{3} be oriented simple loops based at \ast which is a point obtained by the blowdown with respect to the first crosscap in Ng,0N_{g,0} when gg is odd, as shown in Figure 13 (a), and AA and BB simple closed curves with arrows as shown in Figure 13 (b). By the relations (1), (4)(4)^{\prime} and Remark 2.3, we have y2=𝒫(α2)y^{2}=\mathcal{P}_{\otimes}(\alpha^{2}), (a2a3ag1)g1=𝒫(γ1)(a_{2}a_{3}\cdots{}a_{g-1})^{g-1}=\mathcal{P}_{\otimes}(\gamma_{1}), (f1a3ag1)g1=𝒫(γ2)(f^{-1}a_{3}\cdots{}a_{g-1})^{g-1}=\mathcal{P}_{\otimes}(\gamma_{2}) and tAtB=𝒫(γ3)t_{A}t_{B}=\mathcal{P}_{\otimes}(\gamma_{3}). Hence by Remark 2.4, we calculate

y2(f1a3ag1)g1(a2a3ag1)g1\displaystyle y^{2}(f^{-1}a_{3}\cdots{}a_{g-1})^{g-1}(a_{2}a_{3}\cdots{}a_{g-1})^{g-1} =\displaystyle= 𝒫(α2)𝒫(γ2)𝒫(γ1)\displaystyle\mathcal{P}_{\otimes}(\alpha^{2})\mathcal{P}_{\otimes}(\gamma_{2})\mathcal{P}_{\otimes}(\gamma_{1})
=\displaystyle= 𝒫(α2)𝒫(γ1γ2)\displaystyle\mathcal{P}_{\otimes}(\alpha^{2})\mathcal{P}_{\otimes}(\gamma_{1}\gamma_{2})
=\displaystyle= 𝒫(γ1γ2α2)\displaystyle\mathcal{P}_{\otimes}(\gamma_{1}\gamma_{2}\alpha^{2})
=\displaystyle= 𝒫(γ3)\displaystyle\mathcal{P}_{\otimes}(\gamma_{3})
=\displaystyle= tAtB\displaystyle t_{A}t_{B}
=(1)\displaystyle\overset{(\ref{bound})}{=} tA.\displaystyle t_{A}.

On the other hand, we see

a1{ea2(a21a31ag11ag11a31a21)}2a11\displaystyle a_{1}\{ea_{2}(a_{2}^{-1}a_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}a_{2}^{-1})\}^{2}a_{1}^{-1}
=\displaystyle= a1(ea2)a11a1(a2a3ag1ag1a3a2)1ea2(a2a3ag1ag1a3a2)a11\displaystyle a_{1}(ea_{2})a_{1}^{-1}\cdot{}a_{1}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})^{-1}ea_{2}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})a_{1}^{-1}
a1(a2a3ag1ag1a3a2)2a11.\displaystyle a_{1}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})^{-2}a_{1}^{-1}.

In addition, by the calculation similar to the beginning of Section 3.1.5, we see

(a2a3ag1)2(g1)\displaystyle(a_{2}a_{3}\cdots{}a_{g-1})^{2(g-1)} =\displaystyle= {(a2a3ag1)g1}2\displaystyle\left\{(a_{2}a_{3}\cdots{}a_{g-1})^{g-1}\right\}^{2}
=(A1),(A2)\displaystyle\overset{(\mathrm{A1}),(\mathrm{A2})}{=} {(a3ag1)g2(a2a3ag1ag1a3a2)}2\displaystyle\left\{(a_{3}\cdots{}a_{g-1})^{g-2}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})\right\}^{2}
=(A1),(A2)\displaystyle\overset{(\mathrm{A1}),(\mathrm{A2})}{=} (a3ag1)2(g2)(a2a3ag1ag1a3a2)2,\displaystyle(a_{3}\cdots{}a_{g-1})^{2(g-2)}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})^{2},

and hence

a1(a2a3ag1ag1a3a2)2a11\displaystyle a_{1}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})^{-2}a_{1}^{-1}
=\displaystyle= a1(a2a3ag1)2(g1)(a3ag1)2(g2)a11\displaystyle a_{1}(a_{2}a_{3}\cdots{}a_{g-1})^{-2(g-1)}(a_{3}\cdots{}a_{g-1})^{2(g-2)}a_{1}^{-1}
=(A1)\displaystyle\overset{(\mathrm{A1})}{=} a1(a2a3ag1)2(g1)a11(a3ag1)2(g2).\displaystyle a_{1}(a_{2}a_{3}\cdots{}a_{g-1})^{-2(g-1)}a_{1}^{-1}(a_{3}\cdots{}a_{g-1})^{2(g-2)}.

Let γ4\gamma_{4}, γ5\gamma_{5}, γ6\gamma_{6} and γ7\gamma_{7} be oriented simple loops based at \ast which is a point obtained by the blowdown with respect to the second crosscap, as shown in Figure 13 (a), γ5,1\gamma_{5,1}, γ5,2\gamma_{5,2} and γ5,3\gamma_{5,3} oriented simple loops based at \ast which is a point obtained by the blowdown with respect to the first crosscap, as shown in Figure 13 (b), and CC a simple closed curve with an arrow as shown in Figure 13 (c). By the relations (3), (4)(4)^{\prime} and Remark 2.3, we have a1(a2a3ag1)(g1)a11=𝒫(γ4)a_{1}(a_{2}a_{3}\cdots{}a_{g-1})^{-(g-1)}a_{1}^{-1}=\mathcal{P}_{\otimes}(\gamma_{4}),

a1(a2a3ag1ag1a3a2)1ea2(a2a3ag1ag1a3a2)a11\displaystyle a_{1}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})^{-1}ea_{2}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})a_{1}^{-1}
=\displaystyle= a1(a2a3ag1ag1a3a2)1𝒫(γ5,1)(a2a3ag1ag1a3a2)a11\displaystyle a_{1}(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})^{-1}\mathcal{P}_{\otimes}(\gamma_{5,1})(a_{2}a_{3}\cdots{}a_{g-1}a_{g-1}\cdots{}a_{3}a_{2})a_{1}^{-1}
=\displaystyle= a1(ag1a3a2)1𝒫(γ5,2)(ag1a3a2)a11\displaystyle a_{1}(a_{g-1}\cdots{}a_{3}a_{2})^{-1}\mathcal{P}_{\otimes}(\gamma_{5,2})(a_{g-1}\cdots{}a_{3}a_{2})a_{1}^{-1}
=\displaystyle= a1𝒫(γ5,3)a11\displaystyle a_{1}\mathcal{P}_{\otimes}(\gamma_{5,3})a_{1}^{-1}
=\displaystyle= 𝒫(γ5),\displaystyle\mathcal{P}_{\otimes}(\gamma_{5}),

a1(ea2)a11=𝒫(γ6)a_{1}(ea_{2})a_{1}^{-1}=\mathcal{P}_{\otimes}(\gamma_{6}), tAtC1=𝒫(γ7)t_{A}t_{C}^{-1}=\mathcal{P}_{\otimes}(\gamma_{7}) and (a3ag1)2(g2)=tC(a_{3}\cdots{}a_{g-1})^{2(g-2)}=t_{C}. Hence by Remark 2.4, we calculate

a1{ea2(a21a31ag11ag11a31a21)}2a11\displaystyle a_{1}\{ea_{2}(a_{2}^{-1}a_{3}^{-1}\cdots{}a_{g-1}^{-1}a_{g-1}^{-1}\cdots{}a_{3}^{-1}a_{2}^{-1})\}^{2}a_{1}^{-1} =\displaystyle= 𝒫(γ6)𝒫(γ5)𝒫(γ4)2tC\displaystyle\mathcal{P}_{\otimes}(\gamma_{6})\mathcal{P}_{\otimes}(\gamma_{5})\mathcal{P}_{\otimes}(\gamma_{4})^{2}t_{C}
=\displaystyle= 𝒫(γ6)𝒫(γ4γ5)𝒫(γ4)tC\displaystyle\mathcal{P}_{\otimes}(\gamma_{6})\mathcal{P}_{\otimes}(\gamma_{4}\gamma_{5})\mathcal{P}_{\otimes}(\gamma_{4})t_{C}
=\displaystyle= 𝒫(γ4γ5γ6)𝒫(γ4)tC\displaystyle\mathcal{P}_{\otimes}(\gamma_{4}\gamma_{5}\gamma_{6})\mathcal{P}_{\otimes}(\gamma_{4})t_{C}
=\displaystyle= 𝒫(γ42γ5γ6)tC\displaystyle\mathcal{P}_{\otimes}(\gamma_{4}^{2}\gamma_{5}\gamma_{6})t_{C}
=\displaystyle= 𝒫(γ7)tC\displaystyle\mathcal{P}_{\otimes}(\gamma_{7})t_{C}
=\displaystyle= tA.\displaystyle t_{A}.

Thus the relation (C1a¯)(\overline{\mathrm{C1a}}) is satisfied in XY\langle{X\mid{}Y}\rangle.

Refer to caption
Refer to caption
Refer to caption
Figure 13.

3.1.6. On the relation (C4a¯)(\overline{\mathrm{C4a}})

For 1ig12\displaystyle 1\leq{i}\leq{\frac{g-1}{2}}, let γi\gamma_{i} be an oriented simple loop based at \ast which is a point obtained by the blowdown with respect to the first crosscap, as shown in Figure 14, and Φ=e1a3ag1\Phi=e^{-1}a_{3}\cdots{}a_{g-1}.

We now prove the following lemma.

Lemma 3.2.

In XY\langle{X\mid{}Y}\rangle, we have 𝒫(γ1)Φ2𝒫(γi)=𝒫(γi+1)Φ2\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2}\mathcal{P}_{\otimes}(\gamma_{i})=\mathcal{P}_{\otimes}(\gamma_{i+1})\Phi^{2} for 1ig32\displaystyle 1\leq{i}\leq{\frac{g-3}{2}}.

Proof.

Let δi\delta_{i} be an oriented simple loop based at \ast which is a point obtained by the blowdown with respect to the first crosscap, as shown in Figure 14. By Remark 2.3, repeating the relation (3), we have Φ2𝒫(γi)Φ2=𝒫(δi)\Phi^{2}\mathcal{P}_{\otimes}(\gamma_{i})\Phi^{-2}=\mathcal{P}_{\otimes}(\delta_{i}). In addition, by Remarks 2.3 and 2.4, we have 𝒫(γ1)𝒫(δi)=𝒫(γi+1)\mathcal{P}_{\otimes}(\gamma_{1})\mathcal{P}_{\otimes}(\delta_{i})=\mathcal{P}_{\otimes}(\gamma_{i+1}), and hence the claim holds. ∎

Refer to caption
Figure 14.

By the relation (4)(4)^{\prime} and Remark 2.3, we have a2e=𝒫(γ1)a_{2}e=\mathcal{P}_{\otimes}(\gamma_{1}) and 𝒫(γg12)=Φ1g\mathcal{P}_{\otimes}(\gamma_{\frac{g-1}{2}})=\Phi^{1-g}. Hence by Lemma 3.2, we calculate

(a2a3ag1e1a3ag1)g12\displaystyle(a_{2}a_{3}\cdots{}a_{g-1}e^{-1}a_{3}\cdots{}a_{g-1})^{\frac{g-1}{2}} =\displaystyle= (a2e(e1a3ag1)2)g12\displaystyle(a_{2}e(e^{-1}a_{3}\cdots{}a_{g-1})^{2})^{\frac{g-1}{2}}
=\displaystyle= (𝒫(γ1)Φ2)g12\displaystyle(\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2})^{\frac{g-1}{2}}
=\displaystyle= (𝒫(γ1)Φ2)g52𝒫(γ2)Φ4\displaystyle(\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2})^{\frac{g-5}{2}}\mathcal{P}_{\otimes}(\gamma_{2})\Phi^{4}
=\displaystyle= (𝒫(γ1)Φ2)g72𝒫(γ3)Φ6\displaystyle(\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2})^{\frac{g-7}{2}}\mathcal{P}_{\otimes}(\gamma_{3})\Phi^{6}
\displaystyle\vdots
=\displaystyle= 𝒫(γg12)Φg1\displaystyle\mathcal{P}_{\otimes}(\gamma_{\frac{g-1}{2}})\Phi^{g-1}
=\displaystyle= 1.\displaystyle 1.

Thus the relation (C4a¯)(\overline{\mathrm{C4a}}) is satisfied in XY\langle{X\mid{}Y}\rangle.

3.1.7. On the relation (C3¯)(\overline{\mathrm{C3}})

In (Ng,0)\mathcal{M}(N_{g,0}), ϱ¯\bar{\varrho} is defined as ϱ¯=yϱ\bar{\varrho}=y\varrho and we have the relation ϱ=(y1a2a3ag1ya2a3ag1)g22y1a2a3ag1\varrho=(y^{-1}a_{2}a_{3}\cdots{}a_{g-1}ya_{2}a_{3}\cdots{}a_{g-1})^{\frac{g-2}{2}}y^{-1}a_{2}a_{3}\cdots{}a_{g-1} (see Theorem 2.2 in [26]). In addition, note that Lemma 3.2 holds even if gg is even. Hence we calculate

ϱ¯\displaystyle\bar{\varrho} =\displaystyle= y(y1a2a3ag1ya2a3ag1)g22y1a2a3ag1\displaystyle y(y^{-1}a_{2}a_{3}\cdots{}a_{g-1}ya_{2}a_{3}\cdots{}a_{g-1})^{\frac{g-2}{2}}y^{-1}a_{2}a_{3}\cdots{}a_{g-1}
=\displaystyle= (a2a3ag1ya2a3ag1y1)g22a2a3ag1\displaystyle(a_{2}a_{3}\cdots{}a_{g-1}ya_{2}a_{3}\cdots{}a_{g-1}y^{-1})^{\frac{g-2}{2}}a_{2}a_{3}\cdots{}a_{g-1}
=\displaystyle= (a2a3ag1e1a3ag1)g22a2a3ag1\displaystyle(a_{2}a_{3}\cdots{}a_{g-1}e^{-1}a_{3}\cdots{}a_{g-1})^{\frac{g-2}{2}}a_{2}a_{3}\cdots{}a_{g-1}
=\displaystyle= (𝒫(γ1)Φ2)g22𝒫(γ1)Φ\displaystyle(\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2})^{\frac{g-2}{2}}\mathcal{P}_{\otimes}(\gamma_{1})\Phi
=\displaystyle= (𝒫(γ1)Φ2)g42𝒫(γ2)Φ3\displaystyle(\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2})^{\frac{g-4}{2}}\mathcal{P}_{\otimes}(\gamma_{2})\Phi^{3}
=\displaystyle= (𝒫(γ1)Φ2)g62𝒫(γ3)Φ5\displaystyle(\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2})^{\frac{g-6}{2}}\mathcal{P}_{\otimes}(\gamma_{3})\Phi^{5}
\displaystyle\vdots
=\displaystyle= 𝒫(γ1)Φ2𝒫(γg22)Φg3\displaystyle\mathcal{P}_{\otimes}(\gamma_{1})\Phi^{2}\mathcal{P}_{\otimes}(\gamma_{\frac{g-2}{2}})\Phi^{g-3}
=\displaystyle= 𝒫(γ1)(Φ2𝒫(γg22)Φ2)Φg1,\displaystyle\mathcal{P}_{\otimes}(\gamma_{1})(\Phi^{2}\mathcal{P}_{\otimes}(\gamma_{\frac{g-2}{2}})\Phi^{-2})\Phi^{g-1},

where γi\gamma_{i} is defined in Section 3.1.6. Let δ1\delta_{1} and δ2\delta_{2} be oriented simple loops based at \ast which is a point obtained by the blowdown with respect to the first crosscap, as shown in Figure 15 (a). Repeating the relation (3), we have Φ2𝒫(γg22)Φ2=𝒫(δ1)\Phi^{2}\mathcal{P}_{\otimes}(\gamma_{\frac{g-2}{2}})\Phi^{-2}=\mathcal{P}_{\otimes}(\delta_{1}). In addition, by Remark 2.4, we have 𝒫(γ1)𝒫(δ1)=𝒫(δ2)\mathcal{P}_{\otimes}(\gamma_{1})\mathcal{P}_{\otimes}(\delta_{1})=\mathcal{P}_{\otimes}(\delta_{2}). Let δ3\delta_{3} and δ4\delta_{4} be oriented simple loops based at \ast which is a point obtained by the blowdown with respect to the first crosscap, as shown in Figure 15 (a). Repeating the relation (3), we have Φg1𝒫(δ2)Φ1g=𝒫(δ3)\Phi^{g-1}\mathcal{P}_{\otimes}(\delta_{2})\Phi^{1-g}=\mathcal{P}_{\otimes}(\delta_{3}). In addition, by Remark 2.4, we have 𝒫(δ2)𝒫(δ3)=𝒫(δ4)\mathcal{P}_{\otimes}(\delta_{2})\mathcal{P}_{\otimes}(\delta_{3})=\mathcal{P}_{\otimes}(\delta_{4}). Let AA and BB be simple closed curves with arrows as shown in Figure 15 (b). By the relation (4)(4)^{\prime} and Remark 2.3, we have Φ2g2=tA\Phi^{2g-2}=t_{A} and 𝒫(δ4)=tBtA1\mathcal{P}_{\otimes}(\delta_{4})=t_{B}t_{A}^{-1}. Hence we calculate

ϱ¯2\displaystyle\bar{\varrho}^{2} =\displaystyle= (𝒫(δ2)Φg1)2.\displaystyle(\mathcal{P}_{\otimes}(\delta_{2})\Phi^{g-1})^{2}.
=\displaystyle= 𝒫(δ2)𝒫(δ3)Φ2g2\displaystyle\mathcal{P}_{\otimes}(\delta_{2})\mathcal{P}_{\otimes}(\delta_{3})\Phi^{2g-2}
=\displaystyle= 𝒫(δ4)tA\displaystyle\mathcal{P}_{\otimes}(\delta_{4})t_{A}
=\displaystyle= tB\displaystyle t_{B}
=(1)\displaystyle\overset{(\ref{bound})}{=} 1.\displaystyle 1.

Thus the relation (C3¯)(\overline{\mathrm{C3}}) is satisfied in XY\langle{X\mid{}Y}\rangle.

Refer to caption
Refer to caption
Figure 15.

3.1.8. On the relation (C4¯)(\overline{\mathrm{C4}})

Note that the relation a2ϱ¯=ϱ¯e1a_{2}\bar{\varrho}=\bar{\varrho}e^{-1} is obtained from the relations (C2¯3)(\overline{\mathrm{C2}}_{3}) and (C3¯)(\overline{\mathrm{C3}}). We call this relation (C2¯4)(\overline{\mathrm{C2}}_{4}). By the equality ϱ¯=(a2a3ag1e1a3ag1)g22a2a3ag1\bar{\varrho}=(a_{2}a_{3}\cdots{}a_{g-1}e^{-1}a_{3}\cdots{}a_{g-1})^{\frac{g-2}{2}}a_{2}a_{3}\cdots{}a_{g-1} which appeared in Section 3.1.7, we calculate

(ϱ¯a2a3ag1)g1\displaystyle(\bar{\varrho}a_{2}a_{3}\cdots{}a_{g-1})^{g-1} =\displaystyle= ϱ¯a2a3ag1ϱ¯¯(C2¯2),(C2¯4)a2a3ag1(ϱ¯a2a3ag1)g3\displaystyle\bar{\varrho}\underset{(\overline{\mathrm{C2}}_{2}),(\overline{\mathrm{C2}}_{4})}{\underline{a_{2}a_{3}\cdots{}a_{g-1}\bar{\varrho}}}a_{2}a_{3}\cdots{}a_{g-1}(\bar{\varrho}a_{2}a_{3}\cdots{}a_{g-1})^{g-3}
=\displaystyle= ϱ¯2e1a3ag1a2a3ag1ϱ¯¯(C2¯2),(C2¯3),(C2¯4)a2a3ag1(ϱ¯a2a3ag1)g4\displaystyle\bar{\varrho}^{2}\underset{(\overline{\mathrm{C2}}_{2}),(\overline{\mathrm{C2}}_{3}),(\overline{\mathrm{C2}}_{4})}{\underline{e^{-1}a_{3}\cdots{}a_{g-1}a_{2}a_{3}\cdots{}a_{g-1}\bar{\varrho}}}a_{2}a_{3}\cdots{}a_{g-1}(\bar{\varrho}a_{2}a_{3}\cdots{}a_{g-1})^{g-4}
=\displaystyle= ϱ¯3a2a3ag1e1a3ag1a2a3ag1(ϱ¯a2a3ag1)g4\displaystyle\bar{\varrho}^{3}a_{2}a_{3}\cdots{}a_{g-1}e^{-1}a_{3}\cdots{}a_{g-1}a_{2}a_{3}\cdots{}a_{g-1}(\bar{\varrho}a_{2}a_{3}\cdots{}a_{g-1})^{g-4}
\displaystyle\vdots
=\displaystyle= ϱ¯g1(a2a3ag1e1a3ag1)g22a2a3ag1\displaystyle\bar{\varrho}^{g-1}(a_{2}a_{3}\cdots{}a_{g-1}e^{-1}a_{3}\cdots{}a_{g-1})^{\frac{g-2}{2}}a_{2}a_{3}\cdots{}a_{g-1}
=\displaystyle= ϱ¯g\displaystyle\bar{\varrho}^{g}
=(C3¯)\displaystyle\overset{(\overline{\mathrm{C3}})}{=} 1.\displaystyle 1.

Thus the relation (C4¯)(\overline{\mathrm{C4}}) is satisfied in XY\langle{X\mid{}Y}\rangle.

Therefore well-definedness of ψ\psi follows.

3.2. Surjectivity of ψ\psi

For any simple closed curve dd of Ng,nN_{g,n}, we denote the complement of the interior of a regular neighborhood of dd by Ng,ndN_{g,n}\setminus{d}. If Ng,ndN_{g,n}\setminus{d} and Ng,ddN_{g,d}\setminus{d^{\prime}} are diffeomorphic, there exists x(Ng,n)x\in\mathcal{M}(N_{g,n}) with x(d)=dx(d^{\prime})=d for n1n\leq 1. In particular, if dd and dd^{\prime} are two sided, we have xtd;θx1=td;θxt_{d^{\prime};\theta^{\prime}}x^{-1}=t_{d;\theta} for some orientation θ\theta and θ\theta^{\prime}.

For any two sided simple closed curve dd of Ng,nN_{g,n} where n1n\leq 1, Ng,ndN_{g,n}\setminus{d} is diffeomorphic to either one of

  • Ng2,n+2N_{g-2,n+2}, where g3g\geq 3,

  • Σg22,n+2\Sigma_{\frac{g-2}{2},n+2}, where gg is even,

  • Σh,1Ng2h,n+1\Sigma_{h,1}\sqcup{}N_{g-2h,n+1}, where 0h<g2\displaystyle 0\leq{h}<\frac{g}{2},

  • Ni,1Ngi,n+1N_{i,1}\sqcup{}N_{g-i,n+1}, where 1ig11\leq{i}\leq{}g-1 and g2g\geq 2,

  • Ni,1Σgi2,n+1N_{i,1}\sqcup\Sigma_{\frac{g-i}{2},n+1}, where 1ig1\leq{i}\leq{}g and gig-i is even.

For each case, we would like to find a word ww on X0X_{0} such that ψ(w)=td;θ\psi(w)=t_{d;\theta}, where X0X_{0} is defined at the beginning of Section 3.

3.2.1. The case where Ng,ndN_{g,n}\setminus{d} is diffeomorphic to Ng2,n+2N_{g-2,n+2} or Σg22,n+2\Sigma_{\frac{g-2}{2},n+2}

Since Ng,ndN_{g,n}\setminus{d} is diffeomorphic to either Ng,nα1N_{g,n}\setminus{\alpha_{1}} or Ng,nβg22N_{g,n}\setminus{\beta_{\frac{g-2}{2}}}, there exists x(Ng,n)x\in\mathcal{M}(N_{g,n}) such that xtd;θx1=td;θxt_{d^{\prime};\theta^{\prime}}x^{-1}=t_{d;\theta}, where td;θ=a1t_{d^{\prime};\theta^{\prime}}=a_{1} or bg22b_{\frac{g-2}{2}} respectively. If x𝒯(Ng,n)x\in\mathcal{T}(N_{g,n}), there exists a word x=x1x2xsx=x_{1}x_{2}\cdots{}x_{s} on X0X_{0}. Then we obtain ψ((x1x2xs)td;θ(x1x2xs)1)=td;θ\psi((x_{1}x_{2}\cdots{}x_{s})t_{d^{\prime};\theta^{\prime}}(x_{1}x_{2}\cdots{}x_{s})^{-1})=t_{d;\theta}, repeating the relation (3). If x𝒯(Ng,n)x\notin\mathcal{T}(N_{g,n}), since xy1𝒯(Ng,n)xy^{-1}\in\mathcal{T}(N_{g,n}) by the sequence (3) in Section 2.2, there exists a word xy1=x1x2xsxy^{-1}=x_{1}x_{2}\cdots{}x_{s} on X0X_{0}. Since ytd;θy1=a11yt_{d^{\prime};\theta^{\prime}}y^{-1}=a_{1}^{-1} or b¯g22\bar{b}_{\frac{g-2}{2}}, we obtain ψ((x1x2xs)(ytd;θy1)(x1x2xs)1)=td;θ\psi((x_{1}x_{2}\cdots{}x_{s})(yt_{d^{\prime};\theta^{\prime}}y^{-1})(x_{1}x_{2}\cdots{}x_{s})^{-1})=t_{d;\theta}, repeating the relation (3).

3.2.2. The case where Ng,ndN_{g,n}\setminus{d} is diffeomorphic to Σh,1Ng2h,n+1\Sigma_{h,1}\sqcup{}N_{g-2h,n+1}

When h=0h=0, since td;θ=1t_{d;\theta}=1 by the relation (1), we obtain ψ(1)=td;θ\psi(1)=t_{d;\theta}. When h1h\geq 1, there exists x(Ng,n)x\in\mathcal{M}(N_{g,n}) such that xtd;θx1=td;θxt_{d^{\prime};\theta^{\prime}}x^{-1}=t_{d;\theta}, where dd^{\prime} is the boundary curve of a regular neighborhood of α1α2α2h\alpha_{1}\cup\alpha_{2}\cup\cdots\cup\alpha_{2h} which is diffeomorphic to Σh,1\Sigma_{h,1}, as shown in Figure 16. Note that we have td;θϵ=(a1a2a2h)4h+2t_{d^{\prime};\theta^{\prime}}^{\epsilon}=(a_{1}a_{2}\cdots{}a_{2h})^{4h+2} for some ϵ=±1\epsilon=\pm 1, by the relation (4)(4)^{\prime}. If x𝒯(Ng,n)x\in\mathcal{T}(N_{g,n}), there exists a word x=x1x2xsx=x_{1}x_{2}\cdots{}x_{s} on X0X_{0}. Then we obtain ψ((x1x2xs)(a1a2a2h)ϵ(4h+2)(x1x2xs)1)=td;θ\psi((x_{1}x_{2}\cdots{}x_{s})(a_{1}a_{2}\cdots{}a_{2h})^{\epsilon(4h+2)}(x_{1}x_{2}\cdots{}x_{s})^{-1})=t_{d;\theta}, repeating the relations (3) and (4)(4)^{\prime}. If x𝒯(Ng,n)x\notin\mathcal{T}(N_{g,n}), there exists a word xy1=x1x2xsxy^{-1}=x_{1}x_{2}\cdots{}x_{s} on X0X_{0}. Since ya1y1=a11ya_{1}y^{-1}=a_{1}^{-1}, ya2y1=e1ya_{2}y^{-1}=e^{-1} and yaiy1=aiya_{i}y^{-1}=a_{i} for i3i\geq 3, we obtain ψ((x1x2xs)(a11e1a3a2h)ϵ(4h+2)(x1x2xs)1)=td;θ\psi((x_{1}x_{2}\cdots{}x_{s})(a_{1}^{-1}e^{-1}a_{3}\cdots{}a_{2h})^{\epsilon(4h+2)}(x_{1}x_{2}\cdots{}x_{s})^{-1})=t_{d;\theta}, repeating the relations (3) and (4)(4)^{\prime}.

Refer to caption
Figure 16.

3.2.3. The case where Ng,ndN_{g,n}\setminus{d} is diffeomorphic to Ni,1Ngi,n+1N_{i,1}\sqcup{}N_{g-i,n+1}

When i=1i=1, since td;θ=1t_{d;\theta}=1 by the relation (1), we obtain ψ(1)=td;θ\psi(1)=t_{d;\theta}. When i=2i=2, there exists x(Ng,n)x\in\mathcal{M}(N_{g,n}) such that xy2x1=xtδx1=td;θxy^{2}x^{-1}=xt_{\delta}{}x^{-1}=t_{d;\theta}, where δ\delta appeared in Figure 8. If x𝒯(Ng,n)x\in\mathcal{T}(N_{g,n}), there exists a word x=x1x2xsx=x_{1}x_{2}\cdots{}x_{s} on X0X_{0}. Then we obtain ψ((x1x2xs)y2(x1x2xs)1)=td;θ\psi((x_{1}x_{2}\cdots{}x_{s})y^{2}(x_{1}x_{2}\cdots{}x_{s})^{-1})=t_{d;\theta}, repeating the relation (3). If x𝒯(Ng,n)x\notin\mathcal{T}(N_{g,n}), there exists a word xy1=x1x2xsxy^{-1}=x_{1}x_{2}\cdots{}x_{s} on X0X_{0}. Since yy2y1=y2y\cdot{}y^{2}\cdot{}y^{-1}=y^{2}, we have ψ((x1x2xs)y2(x1x2xs)1)=td;θ\psi((x_{1}x_{2}\cdots{}x_{s})y^{2}(x_{1}x_{2}\cdots{}x_{s})^{-1})=t_{d;\theta}, repeating the relation (3). When i3i\geq 3, we take simple closed curves d1,,d6d_{1},\dots,d_{6} as shown in Figure 17. By induction on ii, we can suppose that tdj;θjt_{d_{j};\theta_{j}} is described as a word on X0X_{0} for 1j61\leq{j}\leq 6. Hence we obtain ψ(td1;θ1ϵ1td2;θ2ϵ2td3;θ3ϵ3td4;θ4ϵ4td5;θ5ϵ5td6;θ6ϵ6)=td;θ\psi(t_{d_{1};\theta_{1}}^{\epsilon_{1}}t_{d_{2};\theta_{2}}^{\epsilon_{2}}t_{d_{3};\theta_{3}}^{\epsilon_{3}}t_{d_{4};\theta_{4}}^{\epsilon_{4}}t_{d_{5};\theta_{5}}^{\epsilon_{5}}t_{d_{6};\theta_{6}}^{\epsilon_{6}})=t_{d;\theta} for some ϵj=±1\epsilon_{j}=\pm 1, by the relations (2) and (5).

Refer to caption
Figure 17.

3.2.4. The case where Ng,ndN_{g,n}\setminus{d} is diffeomorphic to Ni,1Σgi2,n+1N_{i,1}\sqcup\Sigma_{\frac{g-i}{2},n+1}

When i=1i=1, since td;θ=1t_{d;\theta}=1 by the relation (1), we obtain ψ(1)=td;θ\psi(1)=t_{d;\theta}. When i=2i=2, dd is described as shown in Figure 18, for some model of Ng,nN_{g,n}. Let γ1\gamma_{1} and γ2\gamma_{2} be oriented loops of Ng1,nN_{g-1,n} based at \ast as shown in Figure 18, where \ast is the point obtained by the blowdown with respect to the crosscap MM in Figure 18. Note that γ1\gamma_{1} and γ2\gamma_{2} are two sided since gg is even when i=2i=2. In addition, when 𝒫(γj)=tcjtcj1\mathcal{P}_{\otimes}(\gamma_{j})=t_{c_{j}}t_{c^{\prime}_{j}}^{-1}, Ng,n(cjcj)N_{g,n}\setminus(c_{j}\cup{}c^{\prime}_{j}) is diffeomorphic to Ng,n(α2ζ)N_{g,n}\setminus(\alpha_{2}\cup\zeta) for j=1j=1 and 22, where α2\alpha_{2} and ζ\zeta appeared in Figure 8. Hence by Remark 2.3, we can describe 𝒫(γj)=xj(a2f)xj1\mathcal{P}_{\otimes}(\gamma_{j})=x_{j}(a_{2}f)x_{j}^{-1} for some xj(Ng,n)x_{j}\in\mathcal{M}(N_{g,n}), for j=1j=1 and 22. If xj𝒯(Ng,n)x_{j}\in\mathcal{T}(N_{g,n}), xjx_{j} is represented by a word on X0X_{0}, and hence 𝒫(γj)\mathcal{P}_{\otimes}(\gamma_{j}) is a word on X0X_{0}. If xj𝒯(Ng,n)x_{j}\notin\mathcal{T}(N_{g,n}), xjy1x_{j}y^{-1} is represented by a word on X0X_{0}. Since y(a2f)y1=e1a21y(a_{2}f)y^{-1}=e^{-1}a_{2}^{-1}, 𝒫(γj)\mathcal{P}_{\otimes}(\gamma_{j}) is a word on X0X_{0}. We denote the word presentation of 𝒫(γj)\mathcal{P}_{\otimes}(\gamma_{j}) on X0X_{0} by wjw_{j}. Then we calculate ψ(w2w1)=w2w1=(3)𝒫(γ2)𝒫(γ1)=𝒫(γ1γ2)=(1)td;θ\psi(w_{2}w_{1})=w_{2}w_{1}\overset{(\ref{t-conj})}{=}\mathcal{P}_{\otimes}(\gamma_{2})\mathcal{P}_{\otimes}(\gamma_{1})=\mathcal{P}_{\otimes}(\gamma_{1}\gamma_{2})\overset{(\ref{bound})}{=}t_{d;\theta}, by Remarks 2.3 and 2.4. When i3i\geq 3, we take simple closed curves d1,,d6d_{1},\dots,d_{6} as shown in Figure 17, similar to Section 3.2.3. Since Ng,ndjN_{g,n}\setminus{d_{j}} is diffeomorphic to either N1,1Ng1,n+1N_{1,1}\sqcup{}N_{g-1,n+1}, N2,1Ng2,n+1N_{2,1}\sqcup{}N_{g-2,n+1}, Ni1,1Ngi+1,n+1N_{i-1,1}\sqcup{}N_{g-i+1,n+1} or Ni2,1Ngi+2,n+1N_{i-2,1}\sqcup{}N_{g-i+2,n+1}, by Section 3.2.3, tdj;θjt_{d_{j};\theta_{j}} is described as a word on X0X_{0} for 1j61\leq{j}\leq 6. Hence we obtain ψ(td1;θ1ϵ1td2;θ2ϵ2td3;θ3ϵ3td4;θ4ϵ4td5;θ5ϵ5td6;θ6ϵ6)=td;θ\psi(t_{d_{1};\theta_{1}}^{\epsilon_{1}}t_{d_{2};\theta_{2}}^{\epsilon_{2}}t_{d_{3};\theta_{3}}^{\epsilon_{3}}t_{d_{4};\theta_{4}}^{\epsilon_{4}}t_{d_{5};\theta_{5}}^{\epsilon_{5}}t_{d_{6};\theta_{6}}^{\epsilon_{6}})=t_{d;\theta} for some ϵj=±1\epsilon_{j}=\pm 1, by the relations (2) and (5).

Refer to caption
Figure 18.

Therefore surjectivity of ψ\psi follows, and hence the proof of Proposition 3.1 is completed.

4. Proof of Theorem 1.1 for g1g\geq 1 and n2n\geq 2

Recall the capping map 𝒞\mathcal{C}, the point pushing map 𝒫\mathcal{P}_{\ast} and the forgetful map \mathcal{F} defined in Section 2.1. Let 𝒯+(Ng,n1,)=𝒞(𝒯(Ng,n))\mathcal{T}^{+}(N_{g,n-1},\ast)=\mathcal{C}(\mathcal{T}(N_{g,n})), then for g1g\geq 1 and n2n\geq 2, from the sequences (1) and (2) in Section 2.1, we have the short exact sequences

(4) 1𝒯(Ng,n)𝒞𝒯+(Ng,n1,)1,\displaystyle 1\to\mathbb{Z}\to\mathcal{T}(N_{g,n})\overset{\mathcal{C}}{\to}\mathcal{T}^{+}(N_{g,n-1},\ast)\to 1,
(5) 1π1+(Ng,n1,)𝒫𝒯+(Ng,n1,)𝒯(Ng,n1)1.\displaystyle 1\to\pi_{1}^{+}(N_{g,n-1},\ast)\overset{\mathcal{P}_{\ast}}{\to}\mathcal{T}^{+}(N_{g,n-1},\ast)\overset{\mathcal{F}}{\to}\mathcal{T}(N_{g,n-1})\to 1.

Using these sequences and the presentation of 𝒯(Ng,1)\mathcal{T}(N_{g,1}) given in Section 3, we give the presentation of 𝒯(Ng,n)\mathcal{T}(N_{g,n}) for g1g\geq 1 and n2n\geq 2, by induction on nn.

Note that, in general, as basics on combinatorial group theory, from presented groups G1R1\langle{G_{1}\mid{R_{1}}}\rangle and G3R3\langle{G_{3}\mid{R_{3}}}\rangle with a short exact sequence 1G1R1𝑖G𝑝G3R311\to\langle{G_{1}\mid{R_{1}}}\rangle\overset{i}{\to}{G}\overset{p}{\to}\langle{G_{3}\mid{R_{3}}}\rangle\to 1, we can give a presentation for GG as follows. Let g~\tilde{g} be any lift of gG3g\in{G_{3}} with respect to pp, and r~\tilde{r} a word obtained from rR3r\in{R_{3}} by replacing each gG3g\in{G_{3}} to g~\tilde{g}. For xkerpx\in\ker{p}, denote by wxw_{x} a word on i(G1)i(G_{1}) corresponding to xx. For rR1r\in{R_{1}}, denote by r¯\bar{r} a word on i(G1)i(G_{1}) obtained from rr by replacing hG1h\in{G_{1}} to i(h)i(h). Let G2={i(h),g~hG1,gG3}G_{2}=\{i(h),\tilde{g}\mid{h\in{G_{1}},g\in{G_{3}}}\} and R2={r1¯,r3~wr3~1,g~i(h)g~1wg~i(h)g~11r1R1,r3R3,hG1,gG3}R_{2}=\{\overline{r_{1}},\widetilde{r_{3}}w_{\widetilde{r_{3}}}^{-1},\tilde{g}i(h)\tilde{g}^{-1}w_{\tilde{g}i(h)\tilde{g}^{-1}}^{-1}\mid{r_{1}\in{R_{1}},r_{3}\in{R_{3}},h\in{G_{1}},g\in{G_{3}}}\}. Then we have a presentation G=G2R2G=\langle{G_{2}\mid{R_{2}}}\rangle. For details, for instance see Proposition 1 in 10.2 in [11].

We now show the following proposition, using the sequence (5). Remember the infinite presentation XY\langle{X\mid{Y}}\rangle for the group presented in Theorem 1.1.

Proposition 4.1.

For g1g\geq 1 and n2n\geq 2, suppose 𝒯(Ng,n1)=XY\mathcal{T}(N_{g,n-1})=\langle{X\mid{Y}}\rangle, then 𝒯+(Ng,n1,)\mathcal{T}^{+}(N_{g,n-1},\ast) admits a presentation with a generating set

X~={tc~,θ~|c~is a two sided simple closed curve ofNg,n1{}which does not bound a disk neighborhood of,andθ~is an orientation of a regular neighborhood ofc~.}.\widetilde{X}=\left\{t_{\tilde{c},\tilde{\theta}}\left|\begin{array}[]{l}\tilde{c}~{}\textrm{is a two sided simple closed curve of}~{}N_{g,n-1}\setminus\{\ast\}\\ \textrm{which does not bound a disk neighborhood of}~{}\ast,~{}\textrm{and}\\ \tilde{\theta}~{}\textrm{is an orientation of a regular neighborhood of}~{}\tilde{c}.\end{array}\right.\right\}.

The defining relations are

  1. (1~)(\tilde{1})

    tc~,θ~=1t_{\tilde{c},\tilde{\theta}}=1 if c~\tilde{c} bounds a disk or a Möbius band,

  2. (2~)(\tilde{2})

    tc~;c~1=tc~;+c~t_{\tilde{c};-_{\tilde{c}}}^{-1}=t_{\tilde{c};+_{\tilde{c}}},

  3. (3~)(\tilde{3})

    all the conjugation relations ftc~,θ~f1=tf(c~);f(θ~)ft_{\tilde{c},\tilde{\theta}}f^{-1}=t_{f(\tilde{c});f_{\ast}(\tilde{\theta})} for fX~f\in\widetilde{X},

  4. (4~)(\tilde{4})

    all the 22-chain relations,

  5. (5~)(\tilde{5})

    all the lantern relations,

  6. (5~)(\tilde{5})^{\prime}

    all the extended lantern relations defined in Remark 2.2.

Proof.

π1+(Ng,n1,)\pi_{1}^{+}(N_{g,n-1},\ast) is a finite rank free group for n2n\geq 2. However, we consider an infinite presentation for π1+(Ng,n1,)\pi_{1}^{+}(N_{g,n-1},\ast). Let π\pi be the group generated by symbols SαS_{\alpha} for a non-trivial simple loop απ1+(Ng,n1,)\alpha\in\pi_{1}^{+}(N_{g,n-1},\ast), and with the defining relations

  • Sα1=Sα1S_{\alpha^{-1}}=S_{\alpha}^{-1},

  • SαSβ=SγS_{\alpha}S_{\beta}=S_{\gamma} if αβ=γ\alpha\beta=\gamma,

  • SαSβSα1=SγS_{\alpha}S_{\beta}S_{\alpha}^{-1}=S_{\gamma} if αβα1=γ\alpha\beta\alpha^{-1}=\gamma.

Then we have that π\pi is isomorphic to π1+(Ng,n1,)\pi_{1}^{+}(N_{g,n-1},\ast) (see Theorem 1.2 in [13]). So π\pi gives an infinite presentation for π1+(Ng,n1,)\pi_{1}^{+}(N_{g,n-1},\ast). By this we identify π1+(Ng,n1,)\pi_{1}^{+}(N_{g,n-1},\ast) with π\pi.

We take a simple path PP of Ng,n1N_{g,n-1} between \ast and the (n1)(n-1)-st boundary component. For a simple closed curve cc of Ng,n1N_{g,n-1}, let c^\hat{c} be a simple closed curve of Ng,n1{}N_{g,n-1}\setminus\{\ast\} corresponding to cc which does not intersect PP, as shown in Figure 19.

Refer to caption
Figure 19. Examples of simple closed curves of Ng,n1{}N_{g,n-1}\setminus\{\ast\} corresponding to simple closed curves of Ng,n1N_{g,n-1} which do not intersect the path PP.

By the sequence (5), 𝒯+(Ng,n1,)\mathcal{T}^{+}(N_{g,n-1},\ast) is generated by

  • tc^;θ^t_{\hat{c};\hat{\theta}} for any simple closed curve cc of Ng,n1N_{g,n-1} and

  • 𝒫(Sα)\mathcal{P}_{\ast}(S_{\alpha}) for any generator SαS_{\alpha} of π\pi.

The defining relations are

  • r~=wr~\tilde{r}=w_{\tilde{r}} for the lift of any rYr\in{}Y with respect to \mathcal{F},

    • 𝒫(Sα1)=𝒫(Sα)1\mathcal{P}_{\ast}(S_{\alpha^{-1}})=\mathcal{P}_{\ast}(S_{\alpha})^{-1},

    • 𝒫(Sβ)𝒫(Sα)=𝒫(Sγ)\mathcal{P}_{\ast}(S_{\beta})\mathcal{P}_{\ast}(S_{\alpha})=\mathcal{P}_{\ast}(S_{\gamma}) for any SαS_{\alpha}, SβS_{\beta} and SγS_{\gamma} satisfying αβ=γ\alpha\beta=\gamma,

    • 𝒫(Sα)1𝒫(Sβ)𝒫(Sα)=𝒫(Sγ)\mathcal{P}_{\ast}(S_{\alpha})^{-1}\mathcal{P}_{\ast}(S_{\beta})\mathcal{P}_{\ast}(S_{\alpha})=\mathcal{P}_{\ast}(S_{\gamma}) for any SαS_{\alpha}, SβS_{\beta} and SγS_{\gamma} satisfying αβα1=γ\alpha\beta\alpha^{-1}=\gamma and

  • tc^;θ^𝒫(Sα)tc^;θ^1=wtc^;θ^𝒫(Sα)tc^;θ^1t_{\hat{c};\hat{\theta}}\mathcal{P}_{\ast}(S_{\alpha})t_{\hat{c};\hat{\theta}}^{-1}=w_{t_{\hat{c};\hat{\theta}}\mathcal{P}_{\ast}(S_{\alpha})t_{\hat{c};\hat{\theta}}^{-1}} for any tc^;θ^t_{\hat{c};\hat{\theta}} and SαS_{\alpha},

where wxw_{x} is a word corresponding to xx on {𝒫(Sα)}\{\mathcal{P}_{\ast}(S_{\alpha})\}. We denote this presentation by X~0Y~0\langle{\widetilde{X}_{0}\mid\widetilde{Y}_{0}}\rangle. In addition, let X~Y~\langle{\widetilde{X}\mid\widetilde{Y}}\rangle be the infinitely presented group with the presentation given in Proposition 4.1. We show that X~Y~\langle{\widetilde{X}\mid\widetilde{Y}}\rangle is isomorphic to X~0Y~0=𝒯+(Ng,n1,)\langle{\widetilde{X}_{0}\mid\widetilde{Y}_{0}}\rangle=\mathcal{T}^{+}(N_{g,n-1},\ast).

Denote by F(X~0)F(\widetilde{X}_{0}) the free group freely generated by X~0\widetilde{X}_{0}. Let p~:F(X~0)X~0Y~0\widetilde{p}:F(\widetilde{X}_{0})\to\langle{\widetilde{X}_{0}\mid\widetilde{Y}_{0}}\rangle be the natural projection and η~:F(X~0)X~Y~\widetilde{\eta}:F(\widetilde{X}_{0})\to\langle{\widetilde{X}\mid\widetilde{Y}}\rangle the homomorphism defined as η~(x)=x\widetilde{\eta}(x)=x for any xX~0x\in\widetilde{X}_{0}. Note that, by Remark 2.1, 𝒫(Sα)\mathcal{P}_{\ast}(S_{\alpha}) is also in X~Y~\langle{\widetilde{X}\mid\widetilde{Y}}\rangle for any generator SαS_{\alpha} of π\pi. Hence η~\widetilde{\eta} is well-defined. We consider a correspondence

ψ~:X~0Y~0X~Y~\widetilde{\psi}:\langle{\widetilde{X}_{0}\mid\widetilde{Y}_{0}}\rangle\to\langle{\widetilde{X}\mid\widetilde{Y}}\rangle

satisfying ψ~p~=η~\widetilde{\psi}\circ\widetilde{p}=\widetilde{\eta}. We show that ψ~\widetilde{\psi} is an isomorphism.

First we show well-definedness of ψ~\widetilde{\psi}. In the first relation of Y~0\widetilde{Y}_{0}, from the definition of c^\hat{c}, it is clear that wr~=1w_{\tilde{r}}=1 for any rYr\in{}Y. Hence the first relation of Y~0\widetilde{Y}_{0} is either one of the relations (1~)(\tilde{1})-(5~)(\tilde{5}). In the second relation of Y~0\widetilde{Y}_{0}, the relation 𝒫(Sα1)=𝒫(Sα)1\mathcal{P}_{\ast}(S_{\alpha^{-1}})=\mathcal{P}_{\ast}(S_{\alpha})^{-1} is trivial. The relation 𝒫(Sβ)𝒫(Sα)=𝒫(Sγ)\mathcal{P}_{\ast}(S_{\beta})\mathcal{P}_{\ast}(S_{\alpha})=\mathcal{P}_{\ast}(S_{\gamma}) is obtained from the relations (3~)(\tilde{3}) or (5~)(\tilde{5})^{\prime} by Remark 2.2. The relation 𝒫(Sα)1𝒫(Sβ)𝒫(Sα)=𝒫(Sγ)\mathcal{P}_{\ast}(S_{\alpha})^{-1}\mathcal{P}_{\ast}(S_{\beta})\mathcal{P}_{\ast}(S_{\alpha})=\mathcal{P}_{\ast}(S_{\gamma}) is the relation (3~)(\tilde{3}). In the third relation of Y~0\widetilde{Y}_{0}, we have wtc~;θ~𝒫(Sα)tc~;θ~1=𝒫(Stc~;θ~(α))w_{t_{\tilde{c};\tilde{\theta}}\mathcal{P}_{\ast}(S_{\alpha})t_{\tilde{c};\tilde{\theta}}^{-1}}=\mathcal{P}_{\ast}(S_{t_{\tilde{c};\tilde{\theta}}(\alpha)}), and hence it is the relation (3~)(\tilde{3}). Therefore, any relation of Y~0\widetilde{Y}_{0} is satisfied in X~Y~\langle{\widetilde{X}\mid\widetilde{Y}}\rangle. So ψ~\widetilde{\psi} is well-defined as a homomorphism.

Next we show bijectivity of ψ~\widetilde{\psi}. Let φ~:X~Y~X~0Y~0=𝒯+(Ng,n1,)\widetilde{\varphi}:\langle{\widetilde{X}\mid\widetilde{Y}}\rangle\to\langle{\widetilde{X}_{0}\mid\widetilde{Y}_{0}}\rangle=\mathcal{T}^{+}(N_{g,n-1},\ast) be the homomorphism defined as φ~(x)=x\widetilde{\varphi}(x)=x for any xX~x\in\widetilde{X}. Since φ~(Y~)=1\widetilde{\varphi}(\widetilde{Y})=1 in 𝒯+(Ng,n1,)\mathcal{T}^{+}(N_{g,n-1},\ast) clearly, φ~\widetilde{\varphi} is well-defined. By the definitions of ψ~\widetilde{\psi} and φ~\widetilde{\varphi}, it is clear that φ~ψ~\widetilde{\varphi}\circ\widetilde{\psi} is the identity map, and hence ψ~\widetilde{\psi} is injective. For any tc~,θ~X~t_{\tilde{c},\tilde{\theta}}\in\widetilde{X}, if c~\tilde{c} intersects PP transversally at l1l\geq 1 points, there exist tc~,θ~X~t_{\tilde{c}^{\prime},\tilde{\theta}^{\prime}}\in\widetilde{X} and SαS_{\alpha} such that tc~,θ~=ψ~(𝒫(Sα))tc~,θ~t_{\tilde{c},\tilde{\theta}}=\widetilde{\psi}(\mathcal{P}_{\ast}(S_{\alpha}))t_{\tilde{c}^{\prime},\tilde{\theta}^{\prime}}, as shown in Figure 20, by Remark 2.3. We notice that c~\tilde{c}^{\prime} intersects PP transversally at l1l-1 points. By induction on the intersection number ll of c~\tilde{c} and PP, we see that tc~,θ~t_{\tilde{c},\tilde{\theta}} is a product of some ψ~(tc^;θ^)\widetilde{\psi}(t_{\hat{c};\hat{\theta}}) and some ψ~(𝒫(Sα))\widetilde{\psi}(\mathcal{P}_{\ast}(S_{\alpha}))’s. Hence ψ~\widetilde{\psi} is surjective, and so bijective.

Refer to caption
Figure 20. c~\tilde{c} and c~\tilde{c}^{\prime} are the boundary curves of a regular neighborhood of α\alpha.

Therefore ψ~\widetilde{\psi} is the isomorphism. Thus the claim is obtained. ∎

Finally we complete the proof of Theorem 1.1.

Proof of Theorem 1.1.

From the case where n=1n=1 of Theorem 1.1, Proposition 4.1 and the sequence (4), by induction on nn, 𝒯(Ng,n)\mathcal{T}(N_{g,n}) is generated by

  • the natural lift tc;θt_{c;\theta} of tc~;θ~t_{\tilde{c};\tilde{\theta}} with respect to 𝒞\mathcal{C} and

  • the Dehn twist tn;θnt_{\partial_{n};\theta_{n}} about the nn-th boundary curve n\partial_{n},

that is, XX generate 𝒯(Ng,n)\mathcal{T}(N_{g,n}). In addition 𝒯(Ng,n)\mathcal{T}(N_{g,n}) has the relations

  • r~=wr~\tilde{r}=w_{\tilde{r}} for the lift of any relator rr of 𝒯+(Ng,n1,)\mathcal{T}^{+}(N_{g,n-1},\ast) with respect to 𝒞\mathcal{C} and

  • tc;θtn;θntc;θ1=wtc;θtn;θntc;θ1t_{c;\theta}t_{\partial_{n};\theta_{n}}t_{c;\theta}^{-1}=w_{t_{c;\theta}t_{\partial_{n};\theta_{n}}t_{c;\theta}^{-1}},

where wx=tn;θnmw_{x}=t_{\partial_{n};\theta_{n}}^{m} for some integer mm corresponding to xx. In the first relation, it is clear that wr~=1w_{\tilde{r}}=1 if rr is a relator corresponding to the relations (1~)(\tilde{1})-(5~)(\tilde{5}) of Proposition 4.1. On the other hand, if rr is a relator corresponding to the relation (5~)(\tilde{5})^{\prime} of Proposition 4.1, then wr~=tn;θnϵw_{\tilde{r}}=t_{\partial_{n};\theta_{n}}^{\epsilon} for some ϵ=±1\epsilon=\pm 1. Hence the first relation is either one of the relation (1)-(5). In the second relation, it is clear that wtc;θtn;θntc;θ1=tn;θnw_{t_{c;\theta}t_{\partial_{n};\theta_{n}}t_{c;\theta}^{-1}}=t_{\partial_{n};\theta_{n}}, and hence it is the relation (3).

Thus we complete the proof. ∎

Acknowledgement

The authors would like to express their gratitude to Susumu Hirose for his useful comments.

References

  • [1] J. S. Birman, Mapping class groups and their relationship to braid groups, Comm. Pure Appl. Math. 22 (1969), 213–238.
  • [2] J. S. Birman and D. R. J. Chillingworth, On the homeotopy group of a non-orientable surface, Proc. Cambridge Philos. Soc. 71 (1972), 437–448. Erratum: Math. Proc. Cambridge Philos. Soc. 136 (2004), no. 2, 441.
  • [3] M. Dehn, Die Gruppe der Abbildungsklassen, Das arithmetische Feld auf Flächen, Acta Math. 69 (1938), no. 1, 135–206.
  • [4] M. Dehn, Papers on group theory and topology, Springer-Verlag, New Tork, 1987.
  • [5] D. B. A. Epstein, Curves on 22-manifolds and isotopies, Acta Math. 115 (1966), 83–107.
  • [6] B. Farb and D. Margalit, A primer on mapping class groups, Princeton Mathematical Series, 49.
  • [7] S. Gervais, Presentation and central extensions of mapping class groups, Trans. Amer. Math. Soc. 348 (1996), no. 8, 3097–3132.
  • [8] S. Gervais, A finite presentation of the mapping class group of a punctured surface, Topology 40 (2001), no. 4, 703–725.
  • [9] J. Harer, The second homology group of the mapping class group of an orientable surface, Invent. Math. 72 (1983), no. 2, 221–239.
  • [10] A. Hatcher and W. Thurston, A presentation for the mapping class group of a closed orientable surface, Topology 19 (1980), no. 3, 221–237.
  • [11] D. L. Johnson, Presentations of groups, Second edition. London Mathematical Society Student Texts, 15. Cambridge University Press, Cambridge, 1997.
  • [12] R. Kobayashi and G. Omori, An infinite presentation for the mapping class group of a non-orientable surface with boundary, Osaka J. Math., 59 (2) (2022), 269–314.
  • [13] R. Kobayashi, Infinite presentations for fundamental groups of surfaces, Hiroshima Math. J., to appear.
  • [14] M. Korkmaz, Mapping class groups of nonorientable surfaces, Geom. Dedicata 89 (2002), 109–133.
  • [15] C. Labruère and L. Paris, Presentations for the punctured mapping class groups in terms of Artin groups, Algebr. Geom. Topol. 1 (2001), 73–114.
  • [16] W. B. R. Lickorish, Homeomorphisms of non-orientable two-manifolds, Proc. Cambridge Philos. Soc. 59 (1963), 307–317.
  • [17] W. B. R. Lickorish, A finite set of generators for the homeotopy group of a 22-manifold, Proc. Cambridge Philos. Soc. 60 (1964), 769–778.
  • [18] W. B. R. Lickorish, On the homeomorphisms of a non-orientable surface, Proc. Cambridge Philos. Soc. 61 (1965), 61–64.
  • [19] F. Luo, A presentation of the mapping class groups, Math. Res. Lett. 4 (1997), no. 5, 735–739.
  • [20] M. Matsumoto, A presentation of mapping class groups in terms of Artin groups and geometric monodromy of singularities, Math. Ann. 316 (2000), no. 3, 401–418.
  • [21] G. Omori, An infinite presentation for the mapping class group nonorientable surface, Algebr. Geom. Topol. 17 (2017), no. 1, 419–437.
  • [22] L. Paris and B. Szepietowski, A presentation for the mapping class group of a nonorientable surface, Bull. Soc. Math. France 143 (2015), no. 3, 503–566.
  • [23] M. Stukow, Dehn twists on nonorientable surfaces, Fund. Math. 189 (2006), no. 2, 117–147.
  • [24] M. Stukow, Commensurability of geometric subgroups of mapping class groups, Geom. Dedicata 143 (2009), 117–142.
  • [25] M. Stukow, A finite presentation for the mapping class group of a nonorientable surface with Dehn twists and one crosscap slide as generators, J. Pure Appl. Algebra 218 (2014), no. 12, 2226–2239.
  • [26] M. Stukow, A finite presentation for the twist subgroup of the mapping class group of a nonorientable surface, Bull. Korean Math. Soc. 53 (2016), no. 2, 601–614.
  • [27] B. Szepietowski, A presentation for the mapping class group of the closed non-orientable surface of genus 4, J. Pure Appl. Algebra 213 (2009), no. 11, 2001–2016.
  • [28] B. Wajnryb, A simple presentation for the mapping class group of an orientable surface, Israel J. Math. 45 (1983), no. 2-3, 157–174.