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

Finitely generated infinite torsion groups that are residually finite simple

Eduard Schesler Erwin Schrödinger International Institute for Mathematics and Physics
University of Vienna
1090 Vienna
eduardschesler@googlemail.com
Abstract.

We show that every finitely generated residually finite torsion group GG embeds in a finitely generated torsion group Γ\Gamma that is residually finite simple. In particular we show the existence of finitely generated infinite torsion groups that are residually finite simple, which answers a question of Olshanskii and Osin.

Key words and phrases:
torsion groups
2010 Mathematics Subject Classification:
Primary 20E26

1. Introduction

Let 𝒞\mathcal{C} be a class of groups. A group GG is said to be residually 𝒞\mathcal{C} if the intersection of all normal subgroup NN of GG with G/N𝒞G/N\in\mathcal{C} is the trivial group. It is a classical problem in group theory to determine the classes of groups 𝒞\mathcal{C} for which a given group is residually 𝒞\mathcal{C} and a lot of research has been done in this direction, see e.g. [16, 17, 1, 2]. A special instant of this problem was formulated in 1987 by Gromov [6] and became a notorious open problems in geometric group theory: Is every hyperbolic group residually finite, i.e. residually \mathcal{F}, where \mathcal{F} denotes the class of finite groups. In 2008 it was shown by Olshanskii and Osin [14] that an affirmative answer to Gromov’s question would imply the existence of finitely generated infinite torsion groups that are residually 𝒮\mathcal{FS}, where 𝒮\mathcal{FS} denotes the class of finite simple groups. It was therefore natural for them to ask the following, see [14, Problem 3.4].

Problem 1.1.

Does there exist an infinite finitely generated torsion group that is residually 𝒮\mathcal{FS}?

Despite of a variety of techniques that are known to produce infinite finitely generated residually finite torsion groups that range from amenable [5, 7] and non-amenable branch groups [15, 8] to groups with property (T)(\operatorname{T}) [3, 4], and groups with positive first 2\ell^{2}-Betti number [13, 10], there was no construction known so far that produces infinite finitely generated torsion groups that are residually 𝒮\mathcal{FS}. In fact there is a big obstruction for infinite finitely generated torsion groups to be residually 𝒮\mathcal{FS}. To make this more precise, let us write 𝒮k\mathcal{FS}_{k} to denote the subclass of 𝒮\mathcal{FS} that consists groups that do not contain a subgroup isomorphic to Alt(k)\operatorname{Alt}(k). It was pointed out by Lubotzky and Segal [12, Theorem 16.4.2(i)] that every finitely generated group GG that is residually 𝒮k\mathcal{FS}_{k} for some kk can be realized as a subdirect product of finitely many linear groups. In particular, if such a group GG is infinite, it admits an infinite finitely generated linear quotient, which is virtually torsion free. It therefore follows that the class of finitely generated groups that are residually 𝒮k\mathcal{FS}_{k} for some kk does not contain an infinite torsion group. In view of this, it can be easily seen that an affirmative answer to Problem 1.1 implies the existence of a torsion group Γ\Gamma that is a subdirect subgroup of a product i=1Si\prod_{i=1}^{\infty}S_{i}, where Si𝒮S_{i}\in\mathcal{FS} contains an isomorphic copy of Alt(i)\operatorname{Alt}(i). We will show that such a group Γ\Gamma indeed exists and thereby answer Problem 1.1 affirmatively. In fact we will see that every finitely generated residually finite torsion group embeds in a group Γ\Gamma as above.

Theorem 1.2.

Every finitely generated residually finite torsion group embeds into a torsion group that is residually 𝒮\mathcal{FS}.

The proof of Theorem 1.2 is based on the following idea. Consider a group GG and a sequence of GG-sets (Ωi)i(\Omega_{i})_{i\in\mathbb{N}} that are represented by homomorphisms αi:GSym(Ωi)\alpha_{i}\colon G\rightarrow\operatorname{Sym}(\Omega_{i}). For each ii let τiSym(Ωi)\tau_{i}\in\operatorname{Sym}(\Omega_{i}) be a permutation of Ωi\Omega_{i}. Then, under suitable assumptions on τi\tau_{i} and αi\alpha_{i}, the subgroup Γ\Gamma of iSym(Ωi)\prod_{i\in\mathbb{N}}\operatorname{Sym}(\Omega_{i}) that is generated by (τi)i(\tau_{i})_{i\in\mathbb{N}} and the image of

α:GiSym(Ωi),g(αi(g))i\alpha\colon G\rightarrow\prod_{i\in\mathbb{N}}\operatorname{Sym}(\Omega_{i}),\ g\mapsto(\alpha_{i}(g))_{i\in\mathbb{N}}

will keep some of the properties of GG, e.g. being torsion, while gaining some extra properties, e.g. being residually 𝒮\mathcal{FS}. A related idea was recently applied in a work of Kionke and the author [9] in order to produce new examples of infinite finitely generated amenable simple groups.

Acknowledgments

This article arose from the author’s research stays at the Erwin Schrödinger International Institute for Mathematics and Physics in Vienna and the Institute of Mathematical Sciences in Madrid. The author would like to thank these institutes for their financial and organizational support. The author is grateful to Goulnara Arzhantseva, Andrei Jaikin-Zapirain, Steffen Kionke, and Markus Steenbock for helpful discussions.

2. Extending actions of torsion groups

For the rest of this section we fix a torsion group GG that acts on a set Ω\Omega. Let α:GSym(Ω)\alpha\colon G\rightarrow\operatorname{Sym}(\Omega) denote the corresponding homomorphism. Let us moreover fix an element pΩp\in\Omega and let Ω+:=Ω{q}\Omega^{+}\mathrel{\mathop{:}}=\Omega\cup\{q\} for some qΩq\notin\Omega. We are interested in the subgroup Γ\Gamma of Sym(Ω+)\operatorname{Sym}(\Omega^{+}) that is generated by α(G)\alpha(G) and the transposition τ=(p,q)\tau=(p,q).

Notation 2.1.

Let F(X)F(X) denote the free group over a set XX and let w=xi1xiF(X)w=x_{i_{1}}\ldots x_{i_{\ell}}\in F(X) be a reduced word of length 0\ell\in\mathbb{N}_{0}. For each 0k0\leq k\leq\ell we write wk:=xik+1xiw_{\geq k}\mathrel{\mathop{:}}=x_{i_{\ell-k+1}}\ldots x_{i_{\ell}} to denote the terminal subword of length kk in ww.

Let us now consider the free group F:=F(G{τ})F\mathrel{\mathop{:}}=F(G\cup\{\tau\}). To simplify the notation we will often interpret a word wFw\in F as an element of Γ\Gamma, respectively GG if wF(G)w\in F(G), as long as no ambiguity is possible.

Definition 2.2.

For each word wFw\in F of length 0\ell\in\mathbb{N}_{0} and each point ξΩ+\xi\in\Omega^{+}, we define the ww-trace of ξ\xi as the sequence

Trw(ξ):=(wiξ)i=1.\operatorname{Tr}_{w}(\xi)\mathrel{\mathop{:}}=(w_{\geq i}\cdot\xi)_{i=1}^{\ell}.

Note that the ww-trace of an element ξ\xi does not necessarily contain ξ\xi. Let us now fix a finite sequence g1,,gkg_{1},\ldots,g_{k} of elements in GG. In what follows we will study traces for the words

vn,i=(g1gk)ng1giv_{n,i}=(g_{1}\ldots g_{k})^{n}g_{1}\ldots g_{i}

and

wn,i=(τg1τgk)nτg1τgiw_{n,i}=(\tau g_{1}\ldots\tau g_{k})^{n}\tau g_{1}\ldots\tau g_{i}

in FF, where n0n\in\mathbb{N}_{0} and 0i<k0\leq i<k.

Notation 2.3.

Given a group HH and an element hHh\in H, we write oH(h){}\operatorname{o}_{H}(h)\in\mathbb{N}\cup\{\infty\} to denote the order of hh in HH.

Let us consider the element g:=g1gkGg\mathrel{\mathop{:}}=g_{1}\ldots g_{k}\in G and let N=oG(g)N=\operatorname{o}_{G}(g).

Lemma 2.4.

Let ξΩ+\xi\in\Omega^{+} and let 0i<k0\leq i<k. Suppose that pp is not contained in TrwN,i(ξ)\operatorname{Tr}_{w_{N,i}}(\xi). Then pp is not contained in Trwn,i(ξ)\operatorname{Tr}_{w_{n,i}}(\xi) for every n0n\in\mathbb{N}_{0}.

Proof.

If ξ=q\xi=q, then pp is clearly contained in TrwN,i(ξ)\operatorname{Tr}_{w_{N,i}}(\xi) so that there is nothing to show. Let us therefore assume that ξΩ\xi\in\Omega and that TrwN,i(ξ)\operatorname{Tr}_{w_{N,i}}(\xi) does not contain pp. Since τ\tau fixes every point in Ω{p}\Omega\setminus\{p\}, it follows that TrvN,i(ξ)\operatorname{Tr}_{v_{N,i}}(\xi) does not contain pp. Thus there is no non-trivial terminal subword uu of (g1gk)Ng1gi(g_{1}\ldots g_{k})^{N}g_{1}\ldots g_{i} that satisfies u(ξ)=pu(\xi)=p. Since

(g1gk)Ng1giξ=g1giξ,(g_{1}\ldots g_{k})^{N}g_{1}\ldots g_{i}\cdot\xi=g_{1}\ldots g_{i}\cdot\xi,

it follows that TrvaN+r,i(ξ)\operatorname{Tr}_{v_{aN+r,i}}(\xi) does not contain pp for every a0a\in\mathbb{N}_{0} and every r<kr<k. Thus the same is true for waN+r,iw_{aN+r,i}, which proves the lemma. ∎

Lemma 2.5.

The element pp is contained in TrwN,i(p)\operatorname{Tr}_{w_{N,i}}(p) for every 0i<k0\leq i<k.

Proof.

Suppose that pp is not contained in TrwN,i(p)\operatorname{Tr}_{w_{N,i}}(p). Since τ\tau fixes every point in Ω{p}\Omega\setminus\{p\}, it follows that pp is not contained in TrvN,i(p)\operatorname{Tr}_{v_{N,i}}(p). However, this is not possible since the word (gi+1gkg1gi)N(g_{i+1}\ldots g_{k}g_{1}\ldots g_{i})^{N}, which represents the trivial element in GG, is a non-trivial terminal subword of vN,iv_{N,i}. ∎

Lemma 2.6.

Let ξΩ+\xi\in\Omega^{+}, let n0n\in\mathbb{N}_{0}, and let 0i<k0\leq i<k. Suppose that Trwn,i(ξ)\operatorname{Tr}_{w_{n,i}}(\xi) contains pp. Then there are natural numbers m1,m2,jm_{1},m_{2},j with 0m1<m2<N(k+1)0\leq m_{1}<m_{2}<N(k+1) and 0j<k0\leq j<k such that

wN(k+1),0ξ=wm1,jp=wm2,jp.w_{N(k+1),0}\cdot\xi=w_{m_{1},j}\cdot p=w_{m_{2},j}\cdot p.
Proof.

From Lemma 2.4 we know that TrwN,0(ξ)\operatorname{Tr}_{w_{N,0}}(\xi) contains pp. Thus there are integers n1<Nn_{1}<N and i1<ki_{1}<k with

wN,0ξ=wn1,i1pw_{N,0}\cdot\xi=w_{n_{1},i_{1}}\cdot p

and therefore

w(k+1)N,0ξ=wkN+n1,i1p.w_{(k+1)N,0}\cdot\xi=w_{kN+n_{1},i_{1}}\cdot p.

Now an inductive application of Lemma 2.5 provides us with integers n2,n3,,nk<Nn_{2},n_{3},\ldots,n_{k}<N and i2,i3,,ik<ki_{2},i_{3},\ldots,i_{k}<k such that

wkN+n1,i1p\displaystyle w_{kN+n_{1},i_{1}}\cdot p =w(k1)N+n1+n2,i2p\displaystyle=w_{(k-1)N+n_{1}+n_{2},i_{2}}\cdot p
=w(k2)N+n1+n2+n3,i3p\displaystyle=w_{(k-2)N+n_{1}+n_{2}+n_{3},i_{3}}\cdot p
\displaystyle\ \vdots
=wn1++nk+1,ik+1p.\displaystyle=w_{n_{1}+\ldots+n_{k+1},i_{k+1}}\cdot p.

Regarding this, the lemma follows from the pigeonhole principle applied to the sequence of indices i1,,ik+1i_{1},\ldots,i_{k+1}. ∎

Lemma 2.7.

For every ξΩ+\xi\in\Omega^{+} there is a natural number mN(k+1)m\leq N(k+1) such that wm,0(ξ)=ξw_{m,0}(\xi)=\xi.

Proof.

Suppose first that pp is not contained in TrwN,0(ξ)\operatorname{Tr}_{w_{N,0}}(\xi). Then pp is not contained in TrvN,0(ξ)\operatorname{Tr}_{v_{N,0}}(\xi) and we obtain

wN,0ξ=(g1gk)Nξ=ξ.w_{N,0}\cdot\xi=(g_{1}\ldots g_{k})^{N}\cdot\xi=\xi.

Suppose next that pp is contained in TrwN,0(ξ)\operatorname{Tr}_{w_{N,0}}(\xi). From Lemma 2.6 we know that there are natural numbers m1,m2,jm_{1},m_{2},j with 0m1<m2<N(k+1)0\leq m_{1}<m_{2}<N(k+1) and 0j<k0\leq j<k such that

wN(k+1),0ξ=wm1,jp=wm2,jp.w_{N(k+1),0}\cdot\xi=w_{m_{1},j}\cdot p=w_{m_{2},j}\cdot p.

In view of this, we see that wm2m1,0ξ=ξw_{m_{2}-m_{1},0}\cdot\xi=\xi, where m2m1(k+1)Nm_{2}-m_{1}\leq(k+1)N. ∎

3. Embedding torsion groups

In this section we will apply Lemma 2.7 in the case where the involved groups are finitely generated and residually finite. This will enable us to prove Theorem 1.2 from the introduction.

3.1. The finitely generated case

Let GG, Γ\Gamma, and Ω+\Omega^{+} be as above. Suppose now that GG is finitely generated and let XX be a finite generating set of GG. In this case we can define the torsion growth function of GG with respect to XX as the function

TGX:,max{oG(g)gBGX()},T_{G}^{X}\colon\mathbb{N}\rightarrow\mathbb{N},\ \ell\mapsto\max\{\,\operatorname{o}_{G}(g)\mid g\in B_{G}^{X}(\ell)\,\},

where BGX()B_{G}^{X}(\ell) denotes the set of elements of GG whose word length with respect to XX is bounded above by \ell. We consider the generating set X+:=α(X){τ}X^{+}\mathrel{\mathop{:}}=\alpha(X)\cup\{\tau\} of Γ\Gamma.

Lemma 3.1.

Let \ell\in\mathbb{N}, let γBΓX+()\gamma\in B_{\Gamma}^{X^{+}}(\ell), and let ξΩ+\xi\in\Omega^{+}. The size of the orbit γξ\langle\gamma\rangle\cdot\xi is bounded above by TGX()(+1)T_{G}^{X}(\ell)\cdot(\ell+1).

Proof.

Since the claim is trivial otherwise, we may assume that γ\gamma does not lie in Bα(G)α(X)()B^{\alpha(X)}_{\alpha(G)}(\ell). Thus, up to conjugation, we may assume that γ\gamma is represented by a word of the form

w=τg1ττgr,w=\tau g_{1}\tau\ldots\tau g_{r},

where i=1r|gi|α(X)\sum_{i=1}^{r}|g_{i}|_{\alpha(X)}\leq\ell and therefore |g1gr|X|g_{1}\ldots g_{r}|_{X}\leq\ell. In this case we know from Lemma 2.7 that there is a natural number

mTG()(r+1)TG()(+1)m\leq T_{G}(\ell)(r+1)\leq T_{G}(\ell)(\ell+1)

such that γm(ξ)=ξ\gamma^{m}(\xi)=\xi. ∎

Note that Lemma 3.1 has the following immediate consequence.

Corollary 3.2.

Every element γΓ\gamma\in\Gamma satisfies

γ(TGX(|γ|Y)(|γ|Y+1))!=1,\gamma^{(T_{G}^{X}(|\gamma|_{Y})\cdot(|\gamma|_{Y}+1))!}=1,

where |γ|Y|\gamma|_{Y} denotes the word length of γ\gamma with respect to YY. In particular, Γ\Gamma is a torsion group and TΓYT_{\Gamma}^{Y} is bounded above by the function n(TGX(n)(n+1))!n\mapsto(T_{G}^{X}(n)\cdot(n+1))!.

3.2. Families of actions

The crucial point of Corollary 3.2 is that the function

n(TGX(n)(n+1))!n\mapsto(T_{G}^{X}(n)\cdot(n+1))!

does neither depend on the action of Γ\Gamma on Ω+\Omega^{+} nor on the choice of the point pΩp\in\Omega. This allows us to apply Corollary 3.2 simultaneously on a family of GG-actions. To do so, we consider a family (Ωi)iI(\Omega_{i})_{i\in I} of GG-sets Ωi\Omega_{i}. Let αi:GSym(Ωi)\alpha_{i}\colon G\rightarrow\operatorname{Sym}(\Omega_{i}) denote the homomorphism corresponding to the action of GG on Ωi\Omega_{i}. For each iIi\in I we fix an element piΩip_{i}\in\Omega_{i} and let (qi)iΩ(q_{i})_{i\in\Omega} be a family of pairwise different elements that do not lie in iΩi\cup_{i\in\mathbb{N}}\Omega_{i}. Let Ωi+:=Ωi{qi}\Omega_{i}^{+}\mathrel{\mathop{:}}=\Omega_{i}\cup\{q_{i}\} and let τi=(pi,qi)Sym(Ωi+)\tau_{i}=(p_{i},q_{i})\in\operatorname{Sym}(\Omega_{i}^{+}). We consider the homomorphism

αI:GiISym(Ωi+),g(αi(g))iI\alpha_{I}\colon G\rightarrow\prod\limits_{i\in I}\operatorname{Sym}(\Omega_{i}^{+}),\ g\rightarrow(\alpha_{i}(g))_{i\in I}

and the sequence τI:=(τi)iIiISym(Ωi+)\tau_{I}\mathrel{\mathop{:}}=(\tau_{i})_{i\in I}\in\prod_{i\in I}\operatorname{Sym}(\Omega_{i}^{+}). Let ΓI\Gamma_{I} denote the subgroup of iISym(Ωi+)\prod\limits_{i\in I}\operatorname{Sym}(\Omega_{i}^{+}) that is generated by αI(G)\alpha_{I}(G) and τI\tau_{I} and let XI:=αI(X){τI}X_{I}\mathrel{\mathop{:}}=\alpha_{I}(X)\cup\{\tau_{I}\}, which is a finite generating set of ΓI\Gamma_{I}.

Proposition 3.3.

The torsion function TΓIXIT_{\Gamma_{I}}^{X_{I}} of ΓI\Gamma_{I} with respect to XIX_{I} satisfies

TΓIXI(n)(TGX(n)(n+1))!T_{\Gamma_{I}}^{X_{I}}(n)\leq(T_{G}^{X}(n)\cdot(n+1))!

for every nn\in\mathbb{N}. In particular, ΓI\Gamma_{I} is a torsion group.

Proof.

The claim directly follows by applying Corollary 3.2 simultaneously to the actions of ΓI\Gamma_{I} on Ωi+\Omega_{i}^{+}, which are given by the canonical homomorphisms ΓISym(Ωi+)\Gamma_{I}\rightarrow\operatorname{Sym}(\Omega_{i}^{+}) for every iIi\in I. ∎

3.3. The residually finite case

Let us now assume that GG is an infinite finitely generated residually finite torsion group. In this case we can choose a properly decreasing chain (Ni)i(N_{i})_{i\in\mathbb{N}} of finite index normal subgroups of GG that satisfies iNi=1\cap_{i\in\mathbb{N}}N_{i}=1. Let Ωi:=G/Ni\Omega_{i}\mathrel{\mathop{:}}=G/N_{i} and let αi:GSym(Ωi)\alpha_{i}\colon G\rightarrow\operatorname{Sym}(\Omega_{i}) denote the action of GG that is given by left translation. Then, using the assumption that (Ni)i(N_{i})_{i\in\mathbb{N}} is properly decreasing, we see that the homomorphism

αn:GinSym(Ωi),g(αi(g))in\alpha_{\geq n}\colon G\rightarrow\prod\limits_{i\geq n}^{\infty}\operatorname{Sym}(\Omega_{i}),\ g\rightarrow(\alpha_{i}(g))_{i\geq n}

is injective for every nn\in\mathbb{N}. As before, we fix an element piΩip_{i}\in\Omega_{i} for each ii\in\mathbb{N} and a family (qi)i(q_{i})_{i\in\mathbb{N}} of pairwise different elements that do not lie in iΩi\cup_{i\in\mathbb{N}}\Omega_{i}. We write Ωi+:=Ωi{qi}\Omega_{i}^{+}\mathrel{\mathop{:}}=\Omega_{i}\cup\{q_{i}\} and consider the elements τi=(pi,qi)Sym(Ωi+)\tau_{i}=(p_{i},q_{i})\in\operatorname{Sym}(\Omega_{i}^{+}) and τ:=(τi)ii=1Sym(Ωi+)\tau\mathrel{\mathop{:}}=(\tau_{i})_{i\in\mathbb{N}}\in\prod_{i=1}^{\infty}\operatorname{Sym}(\Omega_{i}^{+}). Let ΓinSym(Ωi+)\Gamma\leq\prod\limits_{i\geq n}^{\infty}\operatorname{Sym}(\Omega_{i}^{+}) denote the subgroup that is generated by α1(G)\alpha_{\geq 1}(G) and τ\tau.

Lemma 3.4.

Let YY be a finite set and let HSym(Y)H\leq\operatorname{Sym}(Y) be a subgroup that acts transitively on YY. Let Y+=Y{z}Y^{+}=Y\cup\{z\}, where zYz\notin Y. For every yYy\in Y the group Sym(Y+)\operatorname{Sym}(Y^{+}) is generated by GG and the transposition (y,z)(y,z).

Proof.

Let HH denote the subgroup of Sym(Y+)\operatorname{Sym}(Y^{+}) that is generated by GG and (y,z)(y,z). Since GG acts transitively on YY it follows that every transposition of the form (x,z)(x,z) with xYx\in Y is a conjugate of (y,z)(y,z) in HH and therefore lies in HH. By conjugating such a transposition (x,z)(x,z) with a transposition (x,z)(x^{\prime},z), where x{x,z}x^{\prime}\notin\{x,z\}, we obtain (x,z)(x,z)=(x,x)H(x,z)^{(x^{\prime},z)}=(x,x^{\prime})\in H. Now the proof follows from the well-known fact that Sym(Y+)\operatorname{Sym}(Y^{+}) is generated by all transpositions in Sym(Y+)\operatorname{Sym}(Y^{+}). ∎

Recall that a subgroup HH of a product of groups P=iIKiP=\prod_{i\in I}K_{i} is called subdirect if the canonical map HKiH\rightarrow K_{i} is surjective for every iIi\in I.

Corollary 3.5.

The subgroup Γ\Gamma of i=1Sym(Ωi+)\prod_{i=1}^{\infty}\operatorname{Sym}(\Omega_{i}^{+}) is subdirect.

Proof.

This is a direct consequence of Lemma 3.4 and the definition of Γ\Gamma. ∎

Lemma 3.6.

Let (ni)i(n_{i})_{i\in\mathbb{N}} be a sequence of pairwise different natural numbers, let P:=i=1Sym(ni)P\mathrel{\mathop{:}}=\prod_{i=1}^{\infty}\operatorname{Sym}(n_{i}), and let HPH\leq P be a finitely generated subdirect subgroup. Let ι:HP\iota\colon H\rightarrow P denote the inclusion map. For each kk\in\mathbb{N} let prk:Pi=kSym(ni)\operatorname{pr}_{\geq k}\colon P\rightarrow\prod_{i=k}^{\infty}\operatorname{Sym}(n_{i}) denote the canonical projection. There is a natural number mm such that the image of the group K:=Hi=1Alt(ni)K\mathrel{\mathop{:}}=H\cap\prod_{i=1}^{\infty}\operatorname{Alt}(n_{i}) under prmι\operatorname{pr}_{\geq m}\circ\iota is a subdirect subgroup of i=mAlt(ni)\prod_{i=m}^{\infty}\operatorname{Alt}(n_{i}).

Proof.

Let π:i=1Sym(ni)i=1Sym(ni)abi=1𝔽2\pi\colon\prod_{i=1}^{\infty}\operatorname{Sym}(n_{i})\rightarrow\prod_{i=1}^{\infty}\operatorname{Sym}(n_{i})^{\operatorname{ab}}\cong\prod_{i=1}^{\infty}\mathbb{F}_{2} denote the abelianization. Note that KK is the kernel of πι\pi\circ\iota. Since HH is finitely generated, its image in i=1𝔽2\prod_{i=1}^{\infty}\mathbb{F}_{2} is finite and thus KK has finite index, say kk, in HH. Since there are only finitely many alternating groups that admit proper subgroups of index at most kk, it follows that the canonical map KAlt(ni)K\rightarrow\operatorname{Alt}(n_{i}) is surjective for almost every ii. Thus the lemma follows if mm is chosen big enough. ∎

We are now ready to prove the main result.

Theorem 3.7.

Every finitely generated residually finite torsion group GG embeds into a torsion group that is residually in the class 𝒮\mathcal{FS}.

Proof.

It is shown in [11, Theorem 1.1] that every finitely generated, residually finite torsion group embeds into a finitely generated, residually finite perfect torsion group. Regarding this, we can assume that GG is perfect. Let (Ni)i(N_{i})_{i\in\mathbb{N}} be a strictly decreasing sequence of finite index normal subgroups of GG with iNi=1\cap_{i\in\mathbb{N}}N_{i}=1 and let Ωi=G/Ni\Omega_{i}=G/N_{i}. From Corollary 3.5 we know that GG embeds in a finitely generated subdirect subgroup Γ\Gamma of i=1Sym(ni)\prod_{i=1}^{\infty}\operatorname{Sym}(n_{i}), where ni=|Ωi|+1n_{i}=|\Omega_{i}|+1. In this case Lemma 3.6 provides us with a number mm\in\mathbb{N} such that the projection image of K:=Γi=1Alt(ni)K\mathrel{\mathop{:}}=\Gamma\cap\prod_{i=1}^{\infty}\operatorname{Alt}(n_{i}) in i=mSym(ni)\prod_{i=m}^{\infty}\operatorname{Sym}(n_{i}) is a subdirect subgroup of i=mAlt(ni)\prod_{i=m}^{\infty}\operatorname{Alt}(n_{i}). Since GG is perfect we have GKG\leq K. Moreover the restriction of the projection i=1Sym(ni)i=mSym(ni)\prod_{i=1}^{\infty}\operatorname{Sym}(n_{i})\rightarrow\prod_{i=m}^{\infty}\operatorname{Sym}(n_{i}) to GG is injective since the sequence (Ni)i(N_{i})_{i\in\mathbb{N}} was chosen to be decreasing. Thus GG embeds into the image of KK in i=mAlt(ni)\prod_{i=m}^{\infty}\operatorname{Alt}(n_{i}), which is a finitely generated subdirect torsion subgroup, which completes the proof. ∎

References

  • [1] Ian Agol, Criteria for virtual fibering, J. Topol. 1 (2008), no. 2, 269–284 (English).
  • [2] Martin R. Bridson, David M. Evans, Martin W. Liebeck, and Dan Segal, Algorithms determining finite simple images of finitely presented groups, Invent. Math. 218 (2019), no. 2, 623–648 (English).
  • [3] Mikhail Ershov, Golod-Shafarevich groups with property (T)(T) and Kac-Moody groups, Duke Math. J. 145 (2008), no. 2, 309–339. MR 2449949
  • [4] Mikhail Ershov and Andrei Jaikin-Zapirain, Groups of positive weighted deficiency and their applications, J. Reine Angew. Math. 677 (2013), 71–134. MR 3039774
  • [5] R. I. Grigorčuk, On Burnside’s problem on periodic groups, Funktsional. Anal. i Prilozhen. 14 (1980), no. 1, 53–54. MR 565099
  • [6] M. Gromov, Hyperbolic groups, Essays in group theory, Publ., Math. Sci. Res. Inst. 8, 75-263 (1987)., 1987.
  • [7] Narain Gupta and Saïd Sidki, On the Burnside problem for periodic groups, Math. Z. 182 (1983), no. 3, 385–388. MR 696534
  • [8] Steffen Kionke and Eduard Schesler, Amenability and profinite completions of finitely generated groups, Groups Geom. Dyn. 17 (2023), no. 4, 1235–1258 (English).
  • [9] by same author, From telescopes to frames and simple groups, arXiv preprint arXiv:2304.09307 (2023).
  • [10] by same author, Hereditarily just-infinite torsion groups with positive first ell2ell^{2}-betti number, arXiv preprint arXiv:2401.04542 (2024).
  • [11] by same author, Realising residually finite groups as subgroups of branch groups, Bull. Lond. Math. Soc. 56 (2024), no. 2, 536–550 (English).
  • [12] Alexander Lubotzky and Dan Segal, Subgroup growth., Prog. Math., vol. 212, Basel: Birkhäuser, 2003 (English).
  • [13] W. Lück and D. Osin, Approximating the first L2L^{2}-Betti number of residually finite groups., J. Topol. Anal. 3 (2011), no. 2, 153–160 (English).
  • [14] A. Yu. Olshanskii and D. V. Osin, Large groups and their periodic quotients, Proc. Amer. Math. Soc. 136 (2008), no. 3, 753–759. MR 2361846
  • [15] Said Sidki and J. S. Wilson, Free subgroups of branch groups, Arch. Math. (Basel) 80 (2003), no. 5, 458–463. MR 1995624
  • [16] T. S. Weigel, Residual properties of free groups. III, Isr. J. Math. 77 (1992), no. 1-2, 65–81 (English).
  • [17] Daniel T. Wise, The residual finiteness of negatively curved polygons of finite groups., Invent. Math. 149 (2002), no. 3, 579–617 (English).