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

\AtAppendix

Group actions on contractible 22-complexes II

Kevin Iván Piterman  and  Iván Sadofschi Costa Departamento de Matemática - IMAS
FCEyN, Universidad de Buenos Aires. Buenos Aires, Argentina.
[email protected] [email protected]
Abstract.

In this second part we prove that, if GG is one of the groups PSL2(q)\mathrm{PSL}_{2}(q) with q>5q>5 and q5(mod24)q\equiv 5\pmod{24} or q13(mod24)q\equiv 13\pmod{24}, then the fundamental group of every acyclic 22-dimensional, fixed point free and finite GG-complex admits a nontrivial representation in a unitary group 𝐔(m)\mathbf{U}(m). This completes the proof of the following result: every action of a finite group on a finite and contractible 22-complex has a fixed point.

Key words and phrases:
Group actions, contractible 22-complexes, moduli of group representations, mapping degree, finite simple groups
2010 Mathematics Subject Classification:
57S17, 57M20, 57M60, 55M20, 55M25, 20F05, 20E06, 20D05
Researchers of CONICET. Kevin Iván Piterman was partially supported by grants PIP 11220170100357, PICT 2017-2997, and UBACYT 20020160100081BA, and by an Oberwolfach Leibniz Fellowship. Iván Sadofschi Costa was partially supported by grants PICT-2017-2806, PIP 11220170100357CO and UBACyT 20020160100081BA

1. Introduction

In this second part we prove the following:

Theorem C.

Let GG be one of the groups PSL2(q)\mathrm{PSL}_{2}(q) with q>5q>5 and q5(mod24)q\equiv 5\pmod{24} or q13(mod24)q\equiv 13\pmod{24}. Then the fundamental group of every 22-dimensional, fixed point free, finite and acyclic GG-complex admits a nontrivial representation in a unitary group 𝐔(m)\mathbf{U}(m).

This completes the proof of the following result: every action of a finite group GG on a finite and contractible 22-complex XX has a fixed point.

The groups GG considered in [SC21, Theorem B] share a key property: they admit a nontrivial representation ρ0{\rho_{0}} which restricts to an irreducible representation on the Borel subgroup. The moduli k\mathcal{M}_{k} of representations of Γk=X1OS+k(G):G\Gamma_{k}=X_{1}^{OS+k}(G):G constructed in the proof of [SC21, Theorem B] is built from a representation with this property. When q1(mod4)q\equiv 1\pmod{4} no nontrivial representation of PSL2(q)\mathrm{PSL}_{2}(q) restricts to an irreducible representation on the Borel subgroup. To prove Theorem C, we circumvent this difficulty by instead considering the action of G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q) on X1OS+k(G)X_{1}^{OS+k}(G).


Acknowledgements. This work was partially done during a stay of the first author at The Mathematisches Forschungsinstitut Oberwolfach. He is very grateful to the MFO for their hospitality and support.

2. More representation theory

We denote the set of eigenvalues of a square matrix MM by Λ(M)\Lambda(M).

Lemma 2.0.1.

Let GG be a finite group, g1,g2Gg_{1},g_{2}\in G and Hi=giH_{i}=\langle g_{i}\rangle. Let ρ:G𝐔(m)\rho\colon G\to\mathbf{U}(m) be a unitary representation and let ki=#Λ(ρ(gi))k_{i}=\#\Lambda(\rho(g_{i})). Then there are matrices A1,A2𝐔(m)A_{1},A_{2}\in\mathbf{U}(m) such that Aiρ(gi)Ai1A_{i}\rho(g_{i})A_{i}^{-1} is diagonal (for i=1,2i=1,2), A11A2A_{1}^{-1}A_{2} commutes with 𝐂𝐔(m)(ρ(G)){\mathbf{C}}_{\mathbf{U}(m)}(\rho(G)) and

dim((A1𝐂𝐔(m)(ρ(H1))A11)(A2𝐂𝐔(m)(ρ(H2))A21))m2k1k2.\dim\left(\left(A_{1}{\mathbf{C}}_{\mathbf{U}(m)}\left(\rho(H_{1})\right)A_{1}^{-1}\right)\cap\left(A_{2}{\mathbf{C}}_{\mathbf{U}(m)}\left(\rho(H_{2})\right)A_{2}^{-1}\right)\right)\geq\frac{m^{2}}{k_{1}k_{2}}.
Proof.

By [SC21, Theorems 4.1 and 4.2], we can take T𝐔(n)T\in\mathbf{U}(n) and irreducible representations ρj:G𝐔(mj)\rho_{j}\colon G\to\mathbf{U}(m_{j}) with j=1,,kj=1,\ldots,k such that TρT1=ρ1ρkT\rho T^{-1}=\rho_{1}\oplus\ldots\oplus\rho_{k}. Moreover, we can do this so that whenever ρj\rho_{j} and ρj\rho_{j^{\prime}} are isomorphic we have ρj=ρj\rho_{j}=\rho_{j^{\prime}}. For each i=1,2i=1,2 we take matrices Di,1,,Di,kD_{i,1},\ldots,D_{i,k} with Di,j𝐔(mj)D_{i,j}\in\mathbf{U}(m_{j}) such that

Di,jρj(gi)Di,j1D_{i,j}\rho_{j}(g_{i})D_{i,j}^{-1}

is diagonal. We choose the Di,jD_{i,j} so that ρj=ρj\rho_{j}=\rho_{j^{\prime}} implies Di,j=Di,jD_{i,j}=D_{i,j^{\prime}}. Let Di=Di,1Di,kD_{i}=D_{i,1}\oplus\ldots\oplus D_{i,k}. Then, by [SC21, Proposition 4.3 and Remark 4.4], DiD_{i} commutes with 𝐂𝐔(m)(Tρ(G)T1){\mathbf{C}}_{\mathbf{U}(m)}(T\rho(G)T^{-1}) and letting Ai=DiTA_{i}=D_{i}T we have that Aiρ(gi)Ai1A_{i}\rho(g_{i})A_{i}^{-1} is diagonal and A11A2A_{1}^{-1}A_{2} commutes with 𝐂𝐔(m)(ρ(G)){\mathbf{C}}_{\mathbf{U}(m)}(\rho(G)). Now for λ1Λ(ρ(g1))\lambda_{1}\in\Lambda(\rho(g_{1})) and λ2Λ(ρ(g2))\lambda_{2}\in\Lambda(\rho(g_{2})) we define

n(λ1,λ2)=#{j:1jm and (A1ρ(g1)A11)j,j=λ1 and (A2ρ(g2)A21)j,j=λ2}.n(\lambda_{1},\lambda_{2})=\#\{j\mathrel{{\>:\>}}1\leq j\leq m\text{ and }(A_{1}\rho(g_{1})A_{1}^{-1})_{j,j}=\lambda_{1}\text{ and }(A_{2}\rho(g_{2})A_{2}^{-1})_{j,j}=\lambda_{2}\}.

Note that

(A1𝐂𝐔(m)(ρ(H1))A11)(A2𝐂𝐔(m)(ρ(H2))A21)\left(A_{1}{\mathbf{C}}_{\mathbf{U}(m)}\left(\rho(H_{1})\right)A_{1}^{-1}\right)\cap\left(A_{2}{\mathbf{C}}_{\mathbf{U}(m)}\left(\rho(H_{2})\right)A_{2}^{-1}\right)

has a subgroup isomorphic to

λ1Λ(ρ(g1)),λ2Λ(ρ(g2))𝐔(n(λ1,λ2))\prod_{\lambda_{1}\in\Lambda(\rho(g_{1})),\lambda_{2}\in\Lambda(\rho(g_{2}))}\mathbf{U}(n(\lambda_{1},\lambda_{2}))

and therefore has dimension at least λ1Λ(ρ(g1)),λ2Λ(ρ(g2))n(λ1,λ2)2\sum_{\lambda_{1}\in\Lambda(\rho(g_{1})),\lambda_{2}\in\Lambda(\rho(g_{2}))}n(\lambda_{1},\lambda_{2})^{2}. The AM-QM inequality gives

mk1k2=λ1Λ(ρ(g1)),λ2Λ(ρ(g2))n(λ1,λ2)k1k2λ1Λ(ρ(g1)),λ2Λ(ρ(g2))n(λ1,λ2)2k1k2,\frac{m}{k_{1}k_{2}}=\frac{\displaystyle\sum_{\lambda_{1}\in\Lambda(\rho(g_{1})),\lambda_{2}\in\Lambda(\rho(g_{2}))}n(\lambda_{1},\lambda_{2})}{k_{1}k_{2}}\leq\sqrt{\frac{\displaystyle\sum_{\lambda_{1}\in\Lambda(\rho(g_{1})),\lambda_{2}\in\Lambda(\rho(g_{2}))}n(\lambda_{1},\lambda_{2})^{2}}{k_{1}k_{2}}},

and we obtain the desired inequality. ∎

3. The G^\widehat{G}-graph X1OS(G)X_{1}^{OS}(G)

Let G=PSL2(q)G=\mathrm{PSL}_{2}(q) with q5(mod24)q\equiv 5\pmod{24} or q13(mod24)q\equiv 13\pmod{24}. Let G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q), so that 𝐙(G^)={1,1}{\mathbf{Z}}(\widehat{G})=\{1,-1\} and G^/𝐙(G^)=G\widehat{G}/{\mathbf{Z}}(\widehat{G})=G. In what follows we denote z=1𝐙(G^)z=-1\in{\mathbf{Z}}(\widehat{G}).

We consider a construction of X1OS(G)X_{1}^{OS}(G) as in [SC21, Proposition 5.5]. Recall that for any k0k\geq 0, we can also consider the GG-graph X1OS+k(G)X_{1}^{OS+k}(G) obtained from X1OS(G)X_{1}^{OS}(G) by attaching kk free orbits of 11-cells (note that by [SC20, Proposition 3.10] the GG-homotopy type of X1OS+k(G)X_{1}^{OS+k}(G) does not depend on the particular way these free orbits are attached).

We consider the action of G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q) on X1OS+k(G)X_{1}^{OS+k}(G) defined using the projection π:G^G\pi\colon\widehat{G}\to G. The stabilizer of a vertex (resp. edge) for the action of G^\widehat{G} is a central extension, by 𝐙(G^){\mathbf{Z}}(\widehat{G}), of the stabilizer for the action of GG. Then the G^\widehat{G}-orbits are connected as in Figure 1. The group BB denotes the Borel subgroup of G^\widehat{G} and Q8Q_{8} denotes the quaternion group.

B{B}SL2(3){\mathrm{SL}_{2}(3)}2Dq1{2D_{q-1}}2Dq+1{2D_{q+1}}Cq1\scriptstyle{C_{q-1}}C6\scriptstyle{C_{6}}Q8\scriptstyle{Q_{8}}C4\scriptstyle{C_{4}}

G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q), q13(mod24)q\equiv 13\pmod{24}.

B{B}2Dq+1{2D_{q+1}}2Dq1{2D_{q-1}}SL2(3){\mathrm{SL}_{2}(3)}Cq1\scriptstyle{C_{q-1}}C4\scriptstyle{C_{4}}C6\scriptstyle{C_{6}}Q8\scriptstyle{Q_{8}}

G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q), q5(mod24)q\equiv 5\pmod{24}.

Figure 1. The G^\widehat{G}-graph X1OS(G)X_{1}^{OS}(G).

We now apply Brown’s result [SC21, Theorem 3.1]. The choices in each case are the following. Note that in each case the stabilizers are given in Tables 1 and 2.

  • For G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q) with q13(mod24)q\equiv 13\pmod{24} we take V={v0,v1,v2,v3}V=\{v_{0},v_{1},v_{2},v_{3}\}, E={η0,η1,η2,η3,η1,,ηk}E=\{\eta_{0},\eta_{1},\eta_{2},\allowbreak\eta_{3},\allowbreak\eta^{\prime}_{1},\ldots,\allowbreak\eta^{\prime}_{k}\}, T={η0,η1,η2}T=\{\eta_{0},\eta_{1},\eta_{2}\}, with v0η0v1v_{0}\xrightarrow{\eta_{0}}v_{1}, v1η1v2v_{1}\xrightarrow{\eta_{1}}v_{2}, v1η2v3v_{1}\xrightarrow{\eta_{2}}v_{3}, v3η3gη3v0v_{3}\xrightarrow{\eta_{3}}g_{\eta_{3}}v_{0} and v0ηiv0v_{0}\xrightarrow{\eta^{\prime}_{i}}v_{0} for i=1,,ki=1,\ldots,k.

  • For G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q) with q5(mod24)q\equiv 5\pmod{24} we take V={v0,v1,v2,v3}V=\{v_{0},v_{1},v_{2},v_{3}\}, E={η0,η1,η2,η3,η1,,ηk}E=\{\eta_{0},\eta_{1},\eta_{2},\allowbreak\eta_{3},\allowbreak\eta^{\prime}_{1},\ldots,\eta^{\prime}_{k}\}, T={η0,η2,η3}T=\{\eta_{0},\eta_{2},\eta_{3}\}, with v0η0v1v_{0}\xrightarrow{\eta_{0}}v_{1}, v1η2v3v_{1}\xrightarrow{\eta_{2}}v_{3}, v3η3v2v_{3}\xrightarrow{\eta_{3}}v_{2}, v2η1gη1v0v_{2}\xrightarrow{\eta_{1}}g_{\eta_{1}}v_{0} and v0ηiv0v_{0}\xrightarrow{\eta^{\prime}_{i}}v_{0} for i=1,,ki=1,\ldots,k.

G^{\widehat{G}} q{q} G^v0\widehat{G}_{v_{0}} G^v1\widehat{G}_{v_{1}} G^v2\widehat{G}_{v_{2}} G^v3\widehat{G}_{v_{3}}
SL2(q)\mathrm{SL}_{2}(q) qq odd B=𝔽qCq1B=\mathbb{F}_{q}\rtimes C_{q-1} 2Dq12D_{q-1} 2Dq+12D_{q+1} SL2(3)\mathrm{SL}_{2}(3)
Table 1. Stabilizers of vertices for the graph X1OS(G)X_{1}^{OS}(G).
G^\widehat{G} q{q} G^η0\widehat{G}_{\eta_{0}} G^η1\widehat{G}_{\eta_{1}} G^η2\widehat{G}_{\eta_{2}} G^η3\widehat{G}_{\eta_{3}} G^ηi\widehat{G}_{\eta^{\prime}_{i}}
SL2(q)\mathrm{SL}_{2}(q) q{q} odd Cq1C_{q-1} C4C_{4} Q8Q_{8} C6C_{6} 𝐙(G^){\mathbf{Z}}({\widehat{G}})
Table 2. Stabilizers of edges for the graph X1OS(G)X_{1}^{OS}(G).

In what follows Γk\Gamma_{k} is the group obtained by applying Brown’s result to the action of GG on X1OS+k(G)X_{1}^{OS+k}(G) with these choices. The following lemma is an extension of [SC21, Lemma 5.6] for the action of G^\widehat{G} on X1OS(G)X_{1}^{OS}(G).

Lemma 3.0.1.

Let GG be one of the groups in Theorem C. Let EE be a set of representatives of the orbits of edges in X1OS(G)X_{1}^{OS}(G). Let XX be an acyclic 22-complex obtained from X1OS(G)X_{1}^{OS}(G) by attaching a free orbit of 22-cells along the GG-orbit of a closed edge path ξ=(a1e1ε1,,anenεn)\xi=(a_{1}e_{1}^{\varepsilon_{1}},\ldots,a_{n}e_{n}^{\varepsilon_{n}}) with eiEe_{i}\in E, aiG^a_{i}\in\widehat{G} and εi{1,1}\varepsilon_{i}\in\{-1,1\}. Then it is possible to choose, for each eEe\in E an element xe[G^]x_{e}\in\mathbb{C}[\widehat{G}] and an element δ[G^]\delta\in\mathbb{C}[\widehat{G}] so that

1=(1z)δ+i=1nεiaiN(G^ei)xei.1=(1-z)\delta+\sum_{i=1}^{n}\varepsilon_{i}a_{i}N(\widehat{G}_{e_{i}})x_{e_{i}}.

Therefore, for any complex representation VV of G^\widehat{G} we have V=(1z)V+eEseVG^eV=(1-z)V+\sum_{e\in E}s_{e}V^{\widehat{G}_{e}}, where se=iIeεiais_{e}=\sum_{i\in I_{e}}\varepsilon_{i}a_{i} and Ie={i:ei=e}I_{e}=\{i\mathrel{{\>:\>}}e_{i}=e\}.

Proof.

Consider the ring homomorphism π:[G^][G]\pi\colon\mathbb{C}[\widehat{G}]\to\mathbb{C}[G]. By [SC21, Lemma 5.6], there are elements x~e[G^]\widetilde{x}_{e}\in\mathbb{C}[\widehat{G}] such that 1=i=1nεiπ(ai)N(Gei)π(x~ei)1=\sum_{i=1}^{n}\varepsilon_{i}\pi(a_{i})N(G_{e_{i}})\pi(\widetilde{x}_{e_{i}}). Let xe=12x~ex_{e}=\frac{1}{2}\widetilde{x}_{e}. Note that π(N(G^e))=2N(Ge)\pi(N(\widehat{G}_{e}))=2\cdot N(G_{e}) and then π(i=1nεiaiN(G^ei)xei)=1\pi(\sum_{i=1}^{n}\varepsilon_{i}a_{i}N(\widehat{G}_{e_{i}})x_{e_{i}})=1. Therefore, since the kernel of π\pi is the ideal generated by 1z1-z, there is an element δ[G^]\delta\in\mathbb{C}[\widehat{G}] such that 1=(1z)δ+i=1nεiaiN(G^ei)xei1=(1-z)\delta+\sum_{i=1}^{n}\varepsilon_{i}a_{i}N(\widehat{G}_{e_{i}})x_{e_{i}}. ∎

4. Representations and centralizers

In this section we fix a suitable irreducible representation ρ0:G^𝔾{\rho_{0}}\colon\widehat{G}\to\mathbb{G} and compute the dimension of the centralizers 𝐂𝔾(ρ0(G^ηi)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{i}})) and 𝐂𝔾(ρ0(G^vi)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{i}})). To perform these computations, we first need to know how many elements of each conjugacy class of G^\widehat{G} appear in each of the subgroups G^ηi\widehat{G}_{\eta_{i}} and G^vi\widehat{G}_{v_{i}}. Recall that if xx is an element of G^\widehat{G}, then (x)(x) denotes its conjugacy class. For the structure and conjugacy classes of the groups SL2(q)\mathrm{SL}_{2}(q) we refer to [SC21, Appendix A.5].

Proposition 4.0.1.

Let G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q), with q±1(mod3)q\equiv\pm 1\pmod{3} and q1(mod4)q\equiv 1\pmod{4}. Then

  1. (i)

    Cq1C_{q-1} contains 11 and zz; and 22 elements of each class (al)(a^{l}), for l=1,,q32l=1,\ldots,\frac{q-3}{2}.

  2. (ii)

    C4C_{4} contains 11 and zz; and 22 elements of the class (aq14)(a^{\frac{q-1}{4}}).

  3. (iii)

    Q8Q_{8} contains 11 and zz; and 66 elements of the class (aq14)(a^{\frac{q-1}{4}}).

  4. (iv)

    If q1(mod3)q\equiv 1\pmod{3} then C6C_{6} contains 11 and zz; and 22 elements of each class (al)(a^{l}) for l=q13l=\frac{q-1}{3}, q16\frac{q-1}{6}. If q2(mod3)q\equiv 2\pmod{3} then C6C_{6} contains 11 and zz; and 22 elements of each class (bm)(b^{m}) for m=q+13m=\frac{q+1}{3}, q+16\frac{q+1}{6}.

  5. (v)

    The Borel subgroup BB contains 11 and zz; q12\frac{q-1}{2} elements of each of the classes (c)(c), (d)(d), (zc)(zc) and (zd)(zd); and 2q2q elements of each class (al)(a^{l}), for l=1,,q32l=1,\ldots,\frac{q-3}{2}.

  6. (vi)

    2Dq12D_{q-1} contains 11 and zz; 22 elements of each class (al)(a^{l}), for l=1,,q32l=1,\ldots,\frac{q-3}{2}; and q1q-1 extra elements of the class (aq14)(a^{\frac{q-1}{4}}).

  7. (vii)

    2Dq+12D_{q+1} contains 11 and zz; 22 elements of each class (bm)(b^{m}), for m=1,,q12m=1,\ldots,\frac{q-1}{2}; and q+1q+1 elements of the class (aq14)(a^{\frac{q-1}{4}}).

  8. (viii)

    If q1(mod3)q\equiv 1\pmod{3} then SL2(3)\mathrm{SL}_{2}(3) contains 11 and zz; 66 elements of the class (aq14)(a^{\frac{q-1}{4}}); and 88 elements of each class (al)(a^{l}) for l=q13l=\frac{q-1}{3}, q16\frac{q-1}{6}. If q2(mod3)q\equiv 2\pmod{3} then SL2(3)\mathrm{SL}_{2}(3) contains 11 and zz; 66 elements of the class (aq14)(a^{\frac{q-1}{4}}); and 88 elements of each class (bm)(b^{m}) for m=q+13m=\frac{q+1}{3}, q+16\frac{q+1}{6}.

Proof.

Note that if q1(mod4)q\equiv 1\pmod{4}, then every element of SL2(q)\mathrm{SL}_{2}(q) is conjugate to its inverse (cf. [Dor71, p.234]). The above computation follows then from the structure description of each one of the subgroups appearing in the above list and by [SC21, Theorem A.5.1]. For example, note that B=𝔽qCq1B=\mathbb{F}_{q}\rtimes C_{q-1}, and that 2Dq12D_{q-1} and 2Dq+12D_{q+1} can be described as follows:

2Dq1C(q1)/4Q8,2Dq+1C(q+1)/2C4.2D_{q-1}\simeq C_{(q-1)/4}\rtimes Q_{8},\quad 2D_{q+1}\simeq C_{(q+1)/2}\rtimes C_{4}.

More concretely, 2Dq1=a,α2D_{q-1}=\langle a,\alpha\rangle (resp. 2Dq+1=b,α2D_{q+1}=\langle b,\alpha\rangle), where aa (resp. bb) has order q1q-1 (resp. q+1q+1), α\alpha has order 44, aα=a1a^{\alpha}=a^{-1} (resp. bα=b1b^{\alpha}=b^{-1}) and α2=z\alpha^{2}=z. ∎

For i=1,3i=1,3 and each of the groups GG in Theorem C, we fix a generator g^i\hat{g}_{i} of G^ηi\widehat{G}_{\eta_{i}}.

Proposition 4.0.2.

Let G^=SL2(q)\widehat{G}=\mathrm{SL}_{2}(q) where q5(mod8)q\equiv 5\pmod{8} and let 𝔾=𝐔(q12)\mathbb{G}=\mathbf{U}\left(\frac{q-1}{2}\right). There is an irreducible representation ρ0:G^𝔾{\rho_{0}}\colon\widehat{G}\to\mathbb{G} satisfying the following properties:

  1. (i)

    The centralizer 𝐂𝔾(ρ0(G^η0)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{0}})) has dimension q12\frac{q-1}{2}.

  2. (ii)

    The eigenvalues of ρ0(g^1){\rho_{0}}(\hat{g}_{1}) are 𝐢\mathbf{i} and 𝐢-\mathbf{i}. The centralizer 𝐂𝔾(ρ0(G^η1)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{1}})) has dimension (q1)28\frac{(q-1)^{2}}{8}.

  3. (iii)

    The centralizer 𝐂𝔾(ρ0(G^η2)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{2}})) has dimension (q1)216\frac{(q-1)^{2}}{16}.

  4. (iv)

    The eigenvalues of ρ0(g^3){\rho_{0}}(\hat{g}_{3}) are ω\omega, ω5\omega^{5} and 1-1, where ω=e2π𝐢/6\omega=e^{2\pi\mathbf{i}/6}. The dimension of 𝐂𝔾(ρ0(G^η3)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{3}})) is given by

    dim𝐂𝔾(ρ0(G^η3))={(q1)212 if q1(mod3)q22q+912 if q2(mod3).\dim{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{3}}))=\begin{cases}\frac{(q-1)^{2}}{12}&\text{ if }q\equiv 1\pmod{3}\\ \frac{q^{2}-2q+9}{12}&\text{ if }q\equiv 2\pmod{3}.\end{cases}
  5. (v)

    The restriction of ρ0{\rho_{0}} to the Borel subgroup G^v0\widehat{G}_{v_{0}} is irreducible.

  6. (vi)

    The centralizer 𝐂𝔾(ρ0(G^v1)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{1}})) has dimension q14\frac{q-1}{4}.

  7. (vii)

    The centralizer 𝐂𝔾(ρ0(G^v2)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{2}})) has dimension q14\frac{q-1}{4}.

  8. (viii)

    The dimension of 𝐂𝔾(ρ0(G^v3)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{3}})) is given by

    dim𝐂𝔾(ρ0(G^v3))={(q1)248 if q1(mod3)(q1)2+3248 if q2(mod3).\dim{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{3}}))=\begin{cases}\frac{(q-1)^{2}}{48}&\text{ if }q\equiv 1\pmod{3}\\ \frac{(q-1)^{2}+32}{48}&\text{ if }q\equiv 2\pmod{3}.\end{cases}
Proof.

We take ρ0{\rho_{0}} a representation realizing the degree q12\frac{q-1}{2} character η1\eta_{1} of [SC21, Theorem A.5.1]. By [SC21, Theorem 4.1], we can take ρ0{\rho_{0}} to be unitary. By [SC21, Lemma A.1.1] and [SC21, Lemma 4.5] we can prove parts (i) to (viii) by computing inner products of the restrictions of η1\eta_{1}. These restrictions are computed using Proposition 4.0.1. ∎

5. The proof of Theorem C

For each of the groups GG in Theorem C, we consider a closed edge path ξ\xi in X1OS(G)X_{1}^{OS}(G) such that attaching a free orbit of 22-cells along this path gives an acyclic 22-complex. We define x0=i(ξ)x_{0}=i(\xi), where i:π1(X1OS(G),v0)Γ0i\colon\pi_{1}(X_{1}^{OS}(G),v_{0})\to\Gamma_{0} is the inclusion given by Brown’s theorem. We set xi=xηix_{i}=x_{\eta^{\prime}_{i}} for i=1,,ki=1,\ldots,k. Let η~{\tilde{\eta}} be the unique edge of X1OS(G)X_{1}^{OS}(G) which lies in ETE-T. We define y0=xη~y_{0}=x_{{\tilde{\eta}}} and yi=xηiy_{i}=x_{\eta^{\prime}_{i}} for i=1,,ki=1,\ldots,k.

Let k\mathcal{M}_{k} be the moduli of representations of Γk\Gamma_{k} obtained from the representation ρ0:G^𝐔(m){\rho_{0}}\colon\widehat{G}\to\mathbf{U}(m) of Proposition 4.0.2 using [SC21, Theorem 3.2]. Let ¯k\overline{\mathcal{M}}_{k} be the corresponding quotient obtained using [SC21, Proposition 3.3]. Note that the equalities k=0×𝔾k\mathcal{M}_{k}=\mathcal{M}_{0}\times\mathbb{G}^{k} and ¯k=¯0×𝔾k\overline{\mathcal{M}}_{k}=\overline{\mathcal{M}}_{0}\times\mathbb{G}^{k} still hold, because ρ0(z)𝐙(𝔾){\rho_{0}}(z)\in{\mathbf{Z}}(\mathbb{G}).

In what follows we consider the induced maps Xi(τ)=ρτ(xi)X_{i}(\tau)=\rho_{\tau}(x_{i}), Yi(τ)=ρτ(yi)Y_{i}(\tau)=\rho_{\tau}(y_{i}).

Proof of Theorem C.

By [SC21, Corollary 3.4], k\mathcal{M}_{k} and ¯k\overline{\mathcal{M}}_{k} are connected and orientable. A computation using Proposition 4.0.2 shows dim¯k=dim𝔾k+1\dim\overline{\mathcal{M}}_{k}=\dim\mathbb{G}^{k+1} (alternatively, note that this also follows from [SC21, Lemma 7.2]). By Lemma 6.0.2, 𝟏\mathbf{1} is a regular point of X0X_{0}. By Propositions 7.0.2 and 7.0.1, Y¯0:¯0𝔾\overline{Y}_{0}\colon\overline{\mathcal{M}}_{0}\to\mathbb{G} has degree 0. The rest of the proof now continues in exactly the same way as the proof of [SC21, Theorem B]. See [SC21, Sections 8, 9 and 10] for more details. ∎

6. The differential of X0X_{0} at 𝟏\mathbf{1}

Proposition 6.0.1.

The representation ρ0{\rho_{0}} satisfies Adρ0(z)=1\mathrm{Ad}\circ{\rho_{0}}(z)=1, where Ad:𝔾GL(T𝟏𝔾)\mathrm{Ad}\colon\mathbb{G}\to\mathrm{GL}(T_{\mathbf{1}}\mathbb{G}) is the adjoint representation.

Proof.

This is immediate, for Ad(g)\mathrm{Ad}(g) is the differential of the map xgxg1x\mapsto gxg^{-1} and ρ0(z)=𝟏{\rho_{0}}(z)=-\mathbf{1} is central. ∎

Lemma 6.0.2.

For each of the groups in Theorem C, 𝟏\mathbf{1} is a regular point of X0:0𝔾X_{0}\colon\mathcal{M}_{0}\to\mathbb{G}.

Proof.

Consider the representation Adρ0:G^GL(T𝟏𝔾)\mathrm{Ad}\circ{\rho_{0}}\colon\widehat{G}\to\mathrm{GL}(T_{\mathbf{1}}\mathbb{G}) which is given by gv=dρ0(g)1Lρ0(g)d𝟏Rρ0(g)1(v)g\cdot v=d_{{\rho_{0}}(g)^{-1}}L_{{\rho_{0}}(g)}\circ d_{\mathbf{1}}R_{{\rho_{0}}(g)^{-1}}(v). By [SC21, Proposition 2.4], we have T𝟏𝐂𝔾(ρ0(H))=(T𝟏𝔾)HT_{\mathbf{1}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(H))=(T_{\mathbf{1}}\mathbb{G})^{H}. By Proposition 6.0.1 we have (1z)T𝟏𝔾=0(1-z)\cdot T_{\mathbf{1}}\mathbb{G}=0 and then Lemma 3.0.1 gives T𝟏𝔾=eEseT𝟏𝐂𝔾(ρ0(G^e))T_{\mathbf{1}}\mathbb{G}=\sum_{e\in E}s_{e}\cdot T_{\mathbf{1}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{e})). Then the result follows from [SC21, Theorem 3.7]. ∎

7. The degree of Y¯0\overline{Y}_{0}

We now prove the degree of Y¯0\overline{Y}_{0} is 0 for each of the groups in Theorem C. When q13(mod24)q\equiv 13\pmod{24}, the approach is similar to that of [SC21, Propositions 9.1 and 9.2]. When q5(mod24)q\equiv 5\pmod{24} the approach needs to be modified. Table 3 gives the value of Y0Y_{0} in the different cases that we consider.

G^\widehat{G} q{q} Y0(τ)Y_{0}(\tau)
SL2(q)\mathrm{SL}_{2}(q) q13(mod24)q\equiv 13\pmod{24} τη01τη21τη31ρ0(gη3)\tau_{\eta_{0}}^{-1}\tau_{\eta_{2}}^{-1}\tau_{\eta_{3}}^{-1}{\rho_{0}}(g_{\eta_{3}})
SL2(q)\mathrm{SL}_{2}(q) q5(mod24)q\equiv 5\pmod{24} τη01τη21τη31τη11ρ0(gη1)\tau_{\eta_{0}}^{-1}\tau_{\eta_{2}}^{-1}\tau_{\eta_{3}}^{-1}\tau_{\eta_{1}}^{-1}{\rho_{0}}(g_{\eta_{1}})
Table 3. The map Y0Y_{0}, for each of the groups GG that we consider.
Proposition 7.0.1.

In the case of q13(mod24)q\equiv 13\pmod{24} the degree of Y¯0:¯0𝔾\overline{Y}_{0}\colon\overline{\mathcal{M}}_{0}\to\mathbb{G} is 0.

Proof.

Consider the manifold M=𝐂𝔾(ρ0(G^η0))×𝐂𝔾(ρ0(G^η2))×𝐂𝔾(ρ0(G^η3))M={\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{0}}))\times{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{2}}))\times{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{3}})), the group H=𝐂𝔾(ρ0(G^v1))×𝐂𝔾(ρ0(G^v3))H={\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{1}}))\times{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{3}})) and the free right action MHM\curvearrowleft H given by

(τη0,τη2,τη3)(αv1,αv3)=(αv11τη0,αv31τη2αv1,τη3αv3).(\tau_{\eta_{0}},\tau_{\eta_{2}},\tau_{\eta_{3}})\cdot(\alpha_{v_{1}},\alpha_{v_{3}})=(\alpha_{v_{1}}^{-1}\tau_{\eta_{0}},\alpha_{v_{3}}^{-1}\tau_{\eta_{2}}\alpha_{v_{1}},\tau_{\eta_{3}}\alpha_{v_{3}}).

For q>3q>3 we have

dimM/H\displaystyle\dim M/H =dimMdimH\displaystyle=\dim M-\dim H
=dim0dimdim𝐂𝔾(ρ0(G^η1))+dim𝐂𝔾(ρ0(G^v2))\displaystyle=\dim\mathcal{M}_{0}-\dim\mathcal{H}-\dim{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{1}}))+\dim{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{2}}))
=dim𝔾(q1)28+q14\displaystyle=\dim\mathbb{G}-\frac{(q-1)^{2}}{8}+\frac{q-1}{4}
<dim𝔾.\displaystyle<\dim\mathbb{G}.

Note that the image of Y0¯\overline{Y_{0}} is the image of the map M/H𝔾M/H\to\mathbb{G} given by (τη0,τη2,τη3)τη01τη21τη31ρ0(gη3)(\tau_{\eta_{0}},\tau_{\eta_{2}},\tau_{\eta_{3}})\mapsto\tau_{\eta_{0}}^{-1}\tau_{\eta_{2}}^{-1}\tau_{\eta_{3}}^{-1}{\rho_{0}}(g_{\eta_{3}}). Since this map is differentiable we conclude that Y¯0\overline{Y}_{0} is not surjective and therefore has degree 0. ∎

Proposition 7.0.2.

In the case of q5(mod24)q\equiv 5\pmod{24} and q>5q>5 the degree of Y¯0:¯0𝔾\overline{Y}_{0}\colon\overline{\mathcal{M}}_{0}\to\mathbb{G} is 0.

Proof.

By Lemma 2.0.1 (and parts (ii) and (iv) of Proposition 4.0.2) there are matrices Aη1,Aη3𝔾A_{\eta_{1}},A_{\eta_{3}}\in\mathbb{G} such that Aη31Aη1A_{\eta_{3}}^{-1}A_{\eta_{1}} commutes with 𝐂𝔾(ρ0(G^v2)){\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{2}})) and letting

K=(Aη3𝐂𝔾(ρ0(G^η3))Aη31)(Aη1𝐂𝔾(ρ0(G^η1))Aη11)K=\left(A_{\eta_{3}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{3}}))A_{\eta_{3}}^{-1}\right)\cap\left(A_{\eta_{1}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{1}}))A_{\eta_{1}}^{-1}\right)

we have dimK(q1)224\dim K\geq\frac{(q-1)^{2}}{24}. Consider the \mathcal{H}-equivariant map Z:0𝔾Z\colon\mathcal{M}_{0}\to\mathbb{G} defined by

τAη3τη01τη21τη31Aη31Aη1τη11Aη11ρ0(gη1).\tau\mapsto A_{\eta_{3}}\tau_{\eta_{0}}^{-1}\tau_{\eta_{2}}^{-1}\tau_{\eta_{3}}^{-1}A_{\eta_{3}}^{-1}\cdot A_{\eta_{1}}\tau_{\eta_{1}}^{-1}A_{\eta_{1}}^{-1}{\rho_{0}}(g_{\eta_{1}}).

By [SC21, Proposition 3.11], the induced maps Y¯0,Z¯:¯0𝔾\overline{Y}_{0},\overline{Z}\colon\overline{\mathcal{M}}_{0}\to\mathbb{G} are homotopic. To conclude, we will prove that ZZ is not surjective. Let

M\displaystyle M =(Aη3𝐂𝔾(ρ0(G^η0))Aη31)×(Aη1𝐂𝔾(ρ0(G^η1))Aη11)\displaystyle=\left(A_{\eta_{3}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{0}}))A_{\eta_{3}}^{-1}\right)\times\left(A_{\eta_{1}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{1}}))A_{\eta_{1}}^{-1}\right)
×(Aη3𝐂𝔾(ρ0(G^η2))Aη31)×(Aη3𝐂𝔾(ρ0(G^η3))Aη31)\displaystyle\qquad\times\left(A_{\eta_{3}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{2}}))A_{\eta_{3}}^{-1}\right)\times\left(A_{\eta_{3}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{\eta_{3}}))A_{\eta_{3}}^{-1}\right)
H\displaystyle H =(Aη3𝐂𝔾(ρ0(G^v1))Aη31)×(Aη3𝐂𝔾(ρ0(G^v3))Aη31)×K\displaystyle=\left(A_{\eta_{3}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{1}}))A_{\eta_{3}}^{-1}\right)\times\left(A_{\eta_{3}}{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{3}}))A_{\eta_{3}}^{-1}\right)\times K

and consider the free right action MHM\curvearrowleft H given by

(τ0,τ1,τ2,τ3)(α1,α3,αK)=(α11τ0,τ1αK,α31τ2α1,αK1τ3α3).(\tau_{0},\tau_{1},\tau_{2},\tau_{3})\cdot(\alpha_{1},\alpha_{3},\alpha_{K})=(\alpha_{1}^{-1}\tau_{0},\tau_{1}\alpha_{K},\alpha_{3}^{-1}\tau_{2}\alpha_{1},\alpha_{K}^{-1}\tau_{3}\alpha_{3}).

Finally, note that the image of ZZ is the image of the HH-equivariant map T:M𝔾T\colon M\to\mathbb{G} given by (τ0,τ1,τ2,τ3)τ01τ21τ31τ11ρ(gη1)(\tau_{0},\tau_{1},\tau_{2},\tau_{3})\mapsto\tau_{0}^{-1}\tau_{2}^{-1}\tau_{3}^{-1}\tau_{1}^{-1}\rho(g_{\eta_{1}}) which cannot be surjective since we have

dimM/H\displaystyle\dim M/H =dimMdimH\displaystyle=\dim M-\dim H
=dim0dim+dim𝐂𝔾(ρ0(G^v2))dimK\displaystyle=\dim\mathcal{M}_{0}-\dim\mathcal{H}+\dim{\mathbf{C}}_{\mathbb{G}}({\rho_{0}}(\widehat{G}_{v_{2}}))-\dim K
dim𝔾+q14(q1)224\displaystyle\leq\dim\mathbb{G}+\frac{q-1}{4}-\frac{(q-1)^{2}}{24}
(since q>7q>7) <dim𝔾.\displaystyle<\dim\mathbb{G}.

References

  • [Dor71] Larry Dornhoff. Group representation theory. Part A: Ordinary representation theory. Marcel Dekker, Inc., New York, 1971. Pure and Applied Mathematics, 7.
  • [SC20] Iván Sadofschi Costa. Group actions of A5A_{5} on contractible 22-complexes. Preprint, arXiv:2009.01755, 2020.
  • [SC21] Iván Sadofschi Costa. Group actions on contractible 22-complexes I. With an appendix by Kevin I. Piterman. Preprint, 2021.