Home On the profinite topology on solvable groups
Article Publicly Available

On the profinite topology on solvable groups

  • Khadijeh Alibabaei EMAIL logo
Published/Copyright: October 11, 2016

Abstract

We show that the wreath product of a finitely generated abelian group with a polycyclic group is a LERF group. This theorem yields as a corollary that finitely generated free metabelian groups are LERF, a result due to Coulbois. We also show that a free solvable group of class 3 and rank at least 2 does not contain a strictly ascending HNN-extension of a finitely generated group. Since such groups are known not to be LERF, this settles, in the negative, a question of J. O. Button.

1 Introduction

In [10], Hall defined the profinite topology on a group G. In the profinite topology, the set of all normal subgroups with finite index is taken as a base of neighbourhoods of the identity. For every subgroup H of G, the profinite closure of H in G is the intersection of all subgroups of finite index of G containing H. If every subgroup G is closed in the profinite topology of G, then G is called extended residually finite (ERF) group. It is clear that the class of ERF groups is closed under taking subgroups and quotient groups. Mal’cev showed that every polycyclic-by-finite group is ERF (see [13] or [17, p. 148]). Two weaker properties for the profinite topology are as follows: a group is said to be locally extended residually finite (LERF) (some authors call LERF groups “subgroup separable groups” [16, Section 3.7]) if every finitely generated (f.g.) subgroup of G is closed in the profinite topology of G and a group is said to be residually finite (RF) if the trivial subgroup of G is closed in the profinite topology of G. In [13], Mal’cev showed that a finitely presented LERF group has decidable membership problem with respect to its f.g. subgroups. The classes of RF groups and LERF groups are closed under taking subgroups but they are not closed under taking quotients. In [10], Hall showed that a f.g. free group is LERF. The noncyclic f.g. free solvable groups of solvable class greater than 3 are RF groups but not LERF (see [9] and [1]).

In general the properties LERF and RF fail to be preserved by many natural operations. For example, in [9], Gruenberg showed that the wreath product AB of groups A and B is RF if and only if A and B are RF groups and either A is abelian or B is finite. It is natural to ask under what conditions the wreath product of a group A with a group B is LERF. In general, it is an open question. Since every LERF group is RF, by the Gruenberg theorem, A must be an abelian group. Moreover, as the class of LERF groups is closed under taking subgroups, A and B must be LERF. In [8, Proposition 3.19], it is shown that for every f.g. abelian group A, the wreath product A is LERF. In the third section of this paper, we generalize this result to AQ where Q is a polycyclic group. The fact that every f.g. abelian-by-polycyclic group is RF [11], plays an important role in the proof of our theorem. This theorem yields as a corollary that the f.g. free metabelian group is LERF, which was proved before by Coulbois in his Ph.D. thesis.

For determining that a group G is not ERF (respectively LERF), a useful method is using [5, Theorem 1], which says that if a group G contains a strictly ascending HNN-extension H=B,ttBt-1B with (respectively f.g.) base B and stable letter t, then G is not ERF (respectively LERF). This is because in every finite quotient of G the images of the subgroups B and tBt-1 are conjugate subgroups and so they are of the same order, whence they must be equal. In [2], Alperin showed that a f.g. solvable group which is not polycyclic contains a subgroup that is a strictly ascending HNN-extension. Hence, by the Mal’cev theorem, a f.g. solvable group is not ERF if and only if it contains a strictly ascending HNN-extension. In [14], it is asked if a f.g. solvable group which is not LERF must contain a f.g. strictly ascending HNN-extension H=t,BtBt-1B where B is a cyclic subgroup of G. In [6], it is shown that this is not the case; furthermore, there exists a finitely presented solvable group that is not LERF but does not contain a f.g. strictly ascending HNN-extension subgroup H=t,B where B is a polycyclic subgroup of G. Also, if we take G as a finite non-abelian solvable group, then G is a f.g. solvable group and by the aforementioned result of [9], this group is not RF and hence, G is not LERF. As mentioned in [6], this group does not contain a strictly ascending HNN-extension of a f.g. group. But the following question is still open:

Question 1.1.

If G is a f.g. (or finitely presented) solvable group which is RF but not LERF, then must it contain a strictly ascending HNN-extension subgroup H=t,B where B is a f.g. subgroup of G?

In the fourth section, we show that the free solvable group of solvability class 3 is a counter-example. This does not completely settle the above question since the group is not finitely presented (see [19] or [8]).

2 Preliminaries

In this section we briefly present general results on the (restricted) wreath product of two groups, the HNN construction and profinite topology. Let A and B be nontrivial groups. We construct their (restricted) wreath product as follows. Denote by A(B) the set of functions from B to A with finite support, and equip it with pointwise multiplication to make it a group. The wreath product AB is a semidirect product A(B)B. The elements of AB are pairs (f,b) where fA(B) and bB. Multiplication in AB is given by

(f,b)(g,c)=(fgb,bc)

where gb(x)=g(b-1x) for every xB.

For every aA and bB, let δa,bA(B) be defined by δa,b(b)=a and δa,b(g)=1 for every gB{b}. Note that the set

{δa,baA and bB}

is a generating set for the group A(B).

Lemma 2.1 ([8, Proposition 2.1]).

The wreath product W=AB of the groups A and B is f.g. if and only if both A and B are f.g.

Theorem 2.2 ([9, Theorem 3.2]).

The wreath product AB is RF if and only if A and B are RF and either A is abelian or B is finite.

Lemma 2.3 ([15, Statement 22.14]).

Let A and Q be two groups and NQ. Let T be a right transversal to N in Q. Then there is an isomorphism Θ between A(Q)N and A(T)N which is defined as follows: For every element (f,y) in A(Q)N, Θ(f,y)=(F,y) where, for every xN and tT, F(x)(t)=f(xt).

The following lemma is the main tool for the proof of Theorem 2.2 and we shall use it several times.

Lemma 2.4 ([9, Lemma 3.2]).

Let Q be a group and A be an abelian group. If N is a normal subgroup of Q, then there is a natural homomorphism φ from AQ onto A(Q/N) which is defined by

φ(i=1nδxi,yi,y)=(i=1nδxi,yiN,yN),

for every (i=1nδxi,yi,y)AQ. In particular, if A1A and R is a subgroup of Q, then we have

φ(A1R)=A1(RN/N).

Remark 2.5.

Let F(n) be the free group of rank n2. Denote by Fd(n) the dth derived subgroup, that is Fd(n)=[Fd-1(n),Fd-1(n)] where F0(n)=F(n). The free solvable group of rank n and solvable class d is Sn,d=F(n)/Fd(n). Let T be the free left module of rank n over the ring [Sn,d]. In [12], Magnus showed that the free solvable group Sn,d+1 of solvable class d+1 can be embedded into the group of matrices M of the form (ft01), where fSn,d and tT. The group M is isomorphic to the wreath product of the free abelian group of rank n with Sn,d.

Theorem 2.6 ([9, Theorem 6.3]).

Every f.g. free solvable group of class d1 is an RF group.

Definition 2.7.

Let B be a group and let A1 and A2 be subgroups of B with φ:A1A2 an isomorphism. Let t be the infinite cyclic group generated by a new element t. The HNN-extension H is formed by adjoining to the free product B*t the relations tat-1=φ(a) for all aA1. Hence, H admits the presentation

H=B,ttat-1=φ(a),aA1.

The group B is called the base ofH, t is called the stable letter, and A1 and A2 are called the associated groups.

An HNN-extension H is called an ascending HNN-extension if at least one of the subgroups A1 and A2 is equal to the base B and it is called a strictly ascending HNN-extension, if at least one of the subgroups A1 and A2 is equal to the base B and the other one is strictly contained in B, so that conjugation by t induces an injective but not surjective endomorphism of B.

If H is an HNN-extension that is not ascending, then H contains a free subgroup on two generators [4, Lemma 5.1]. Hence, if an HNN-extension H is a solvable group, then H is an ascending HNN-extension.

An example of HNN-extension is the f.g. Baumslag–Solitar group

BS(n,m)=a,ttamt-1=an,

for each pair of nonzero integers m and n.

Remark 2.8.

If H is a subgroup of the group G, we can consider two topologies on H: its own profinite topology as a group and the restriction to H of the profinite topology of G. It is clear that the restriction to H of the profinite topology of G is contained in the profinite topology of H. In particular, if H1 is a subgroup of H which is closed in the profinite topology of G, then H1 is closed in the profinite topology of H. If H is a finite index subgroup of G, then the induced topology on H from G coincides with its own profinite topology of H.

Lemma 2.9 ([16, Lemma 3.1.5]).

Let G=NH be an RF group which is a semidirect product of its normal subgroup N with a subgroup H. Then the induced topology on H coincides with its own profinite topology and H is closed in the profinite topology of G.

Remark 2.10.

Let N be a normal subgroup of a group G. Then N is closed in the profinite topology of G if and only if G/N is RF.

As we mentioned in the introduction, for determining that a group G is not ERF (respectively LERF), a useful method is the following theorem due to Neumann and Blass.

Theorem 2.11 ([5]).

Let G be a group and H a subgroup of G that is closed in the profinite topology of G. Then H is not conjugate in G to a proper subgroup of itself. In other words, if G contains a (respectively f.g.) strictly ascending HNN-extension H=B,ttBt-1B, then B is not closed in the profinite topology of G.

3 Wreath products of f.g. abelian groups with polycyclic groups are LERF

A group G is called abelian-by-polycyclic if G has a normal subgroup N such that N is abelian and G/N is polycyclic. Clearly the class of abelian-by-polycyclic groups is closed under taking subgroups and quotients.

Theorem 3.1 ([11, Theorem 3]).

Every f.g. abelian-by-polycyclic group is RF.

However, there exist f.g. abelian-by-polycyclic groups that are not LERF. For example the f.g. Baumslag–Solitar group

BS(1,2)=a,tat=a2[12],

is not LERF because we have ata, whence by Theorem 2.11, the subgroup a is not closed in the profinite topology of BS(1,2).

As mentioned in [8], it would be interesting to characterise LERF groups amongst f.g. solvable groups but this is open even when restricted to the class of metabelian groups. In [8, Proposition 3.19], it is shown that A is LERF but not ERF when A is a f.g. abelian group. In this section, we generalize this result and we show that W=AQ is a LERF group for every polycyclic group Q. We need the following lemmas and propositions.

In [18], it is shown that if a group G has a finite index subgroup that is ERF, then G is ERF. The proof generalises immediately for the property LERF.

Lemma 3.2.

Let G be a group with a normal subgroup of finite index N. If N is ERF (respectively LERF), then so is G.

Proof.

In the case ERF, the result is [18, Lemma 4.2]. In the LERF case, the proof is essentially the same; one needs only to observe that if H is a f.g. subgroup of G, then HN is a f.g. subgroup of N: since N has finite index in G, the group

HN/NH/(HN),

is finite and so, HN is a finite index subgroup of a f.g. group H. Hence, by Schreier’s theorem [17, statement 1.6.11], HN is f.g. ∎

The ERF case of the following result is [18, Corollary 4.3].

Corollary 3.3.

Let G be a group with a finite normal subgroup N. If G/N is ERF (respectively LERF) and G is an RF group, then G is ERF (respectively LERF).

For a subgroup H of a group G denote by ClG(H) the profinite closure of H in G.

Lemma 3.4.

Let G be a group and N be a normal subgroup of G. Let H be a subgroup of G such that HN/N is closed in the profinite topology of G/N. Then ClG(H)=H(ClG(H)N). In particular, if for every nNH, n is not in the closure of H in G, then H is closed in the profinite topology of G.

Proof.

Let Φ:GG/N be natural projection. Since HN/N is closed in the profinite topology of G/N, the subgroup HN is closed in the profinite topology of G. Hence, we have

ClG(H)HN.

So, by Dedekind’s Modular Law [17, Statement 1.3.14], we have

ClG(H)=ClG(H)HN=H(ClG(H)N).

If for every nNH, n is not in the closure of H in G, then we have

ClG(H)=H(ClG(H)N)=H(HN)=H.

The following technical result is used in the proof of the main theorem of this section.

Lemma 3.5.

Let G be a group with normal subgroup N such that G/N is an ERF (respectively a LERF) group. Let H be a (respectively f.g.) subgroup of G. Then H is closed in the profinite topology of G if and only if HN is closed in the profinite topology of G.

Proof.

Since G/N is an ERF (respectively LERF) group, this quotient group is RF, so, by Remark 2.10, N is closed in the profinite topology of G.

Suppose that H is a closed (respectively f.g.) subgroup of G. Hence, HN is closed in the profinite topology of G.

Conversely, suppose that HN is a closed subgroup in G. By the preceding lemma, it is enough to show that for every nNH, n does not belong to ClG(H). Since nHN and this subgroup is closed in the profinite topology of G, there is a finite index normal subgroup M of G such that n(HN)M. The subgroup L=MN is a closed normal subgroup of G such that L is of finite index in N and nL.

Now we claim that the subgroup LH is a closed subgroup of G containing H but not n, which entails that n is not in the closure of H in G. First if n is in LH, then n=lh with lL and hH. Then l-1n=hHN(HN)M, whence n is in (HN)M, which is a contradiction with the choice of M. So, n is not in LH. Finally, we show that LH is a closed subgroup of G. As L is closed in the profinite topology of G, by Remark 2.10, G/L is RF. The RF group G/L contains the finite normal subgroup N/L such that (G/L)/(N/L)G/N is ERF (respectively LERF). By Corollary 3.3, G/L is ERF (respectively LERF). Hence (HL)/L is closed in the profinite topology of G/L; so HL is closed in the profinite topology of G. ∎

Here is a modification of [8, Lemma 3.18].

Corollary 3.6.

Let G be a f.g. group which has a surjective morphism φ onto a polycyclic group Q, with abelian kernel K. Let H be a subgroup of G such that φ(H) has finite index in Q. Then the induced topology on H coincides with its own profinite topology and H is closed in the profinite topology of G.

Proof.

Let H1 be a clopen subgroup of H. By the definition of the profinite topology, H1 is a finite index subgroup of H. Hence, the assumptions of the corollary imply that H1K is a finite index subgroup of G containing H1. So, by Remark 2.8, it is enough to show that H1 is closed in the profinite topology of G1=H1K. By Schreier’s theorem [17, statement 1.6.11], G1 is a f.g. group and as the class of abelian-by-polycyclic groups is closed under taking subgroups, G1 is a f.g. abelian-by-polycyclic group. Since K is an abelian group, H1K is a normal subgroup of G1. By Theorem 3.1, the group G1/(H1K) is RF. Hence, by Remark 2.10, H1K is closed in the profinite topology of G1. By the preceding lemma, H1 is closed in G1. ∎

The next lemma is a key tool for the proof of the main theorems of this section and the next section.

Lemma 3.7.

Let A and Q be arbitrary groups, W=AQ and let Φ:WQ be the natural projection. Let H be a f.g. subgroup of W such that R=Φ(H) is closed in the profinite topology of Q. Then there are a finite index subgroup N and a finite subset T of Q such that HA(T)RA(T)N and A(T)N is a finite index subgroup of W.

Proof.

Let

B=A(Q),H=(f1,r1),,(fn,rn)andS=i=1nSupp(fi).

For every sS, let tsQR and rsR such that s=rsts. Let T be a complete set of representatives of R-right cosets of the set {tssS}. Since R is closed in the profinite topology of Q, for all distinct ts,tsT, there is a subgroup Ns,s of finite index of Q containing R such that tsNs,stsNs,s. Since the set T is a finite subset of Q, the subgroup

N=ts,tsTtstsNs,s

is a subgroup of finite index of Q containing R such that for all distinct ts,tsT, tsNtsN. The subgroup H is contained in BN. We can choose a right transversal T to N in Q such that TT. By Lemma 2.3, the subgroup BN is isomorphic to A(T)N and it is clear that this subgroup has finite index in W. The group A(T)R is a subgroup of A(T)N. We claim that every generator of H is in A(T)R. First, note that for every generator (f,r) of H and sSupp(f)S, we have s=rsts with rsR and tsQR. By our definition of T, there is tT such that tst-1R, and so ts=rt for some rR. Therefore, by Lemma 2.3, we have

f(s)=f(rsts)=f(rsrt)=F(rsr)(t).

Hence, F belongs to A(T)(R). So H is contained in A(T)R. ∎

Theorem 3.8.

Let W=AQ where A is a f.g. abelian group and Q is a polycyclic group. Then W is a LERF group.

Proof.

Let B=A(Q), H be a f.g. subgroup of W and let Φ:WQ be the natural projection, and R=Φ(H)(HB/B). By the Mal’cev theorem [13], Q is an ERF group, whence R is closed in the profinite topology of Q. Hence, by Lemma 3.7, there is a finite index subgroup A(T)N of W such that

HA(T)RA(T)N.

By Remark 2.8, it is enough to show that H is closed in the subgroup of finite index A(T)N. Hence, by replacing W with its finite index subgroup A(T)N, we can assume that

HARAQ.

So, we have HA(Q)=HA(R), where we are identifying A(Q) with A(Q){1} and A(R) with A(R){1}.

To show that H is closed in the profinite topology of W, by Lemma 3.5, it is enough to show that A(Q)H=A(R)H is closed in the profinite topology of W. First we show that A(R) is closed in the profinite topology of W. As W/BQ is an ERF group and therefore W/B is RF, by Remark 2.10, B is a closed subgroup of W. Hence, it is enough to show that for every gBA(R), g is not in the closure of A(R). Since gA(R), there is ySupp(g) such that y is not in R. As Q is an ERF group, there exist a subgroup N of finite index of Q containing R such that yN. The subgroup AN contains A(R) but not g. Since N is of finite index in Q, by Corollary 3.6, the subgroup AN is closed in the profinite topology of W. So, g is not in the closure of A(R). This establishes that A(R) is a closed subgroup of W, which contains HA(R). Hence, we have

ClW(HA(R))A(R).

Take gA(R)(HA(R)). We have the natural projection π:ARR. By the second isomorphism theorem and the fact that R=π(H) we have

π(AR)=R=π(H)=HA(R)A(R).

Hence, HA(R)=AR. Since A(R) is abelian, the subgroup HA(R) is normal in HA(R)=AR. Since the class of abelian-by-polycyclic groups is closed under taking quotients, (AR)/(HA(R)) is a f.g. abelian-by-polycyclic group and so RF, by Theorem 3.1. Since gHA(R), there is a normal subgroup M of finite index of AR containing HA(R) but not g. The subgroup L=MA(R) is a normal subgroup of AR containing HA(R) but not g.

Let LW be the normal closure of L in W. By Theorem 3.1, the f.g. abelian-by-polycyclic group W/LW is an RF group, and so, by Remark 2.10, LW is closed in the profinite topology of W. We claim that L=LWA(R). It follows from this claim that L is a closed subgroup of W containing HA(R) but not g, whence g is not in ClW(HA(R)). Hence, HA(R) is closed in the profinite topology of W, and so by Lemma 3.5, H is closed in the profinite topology of W.

It remains to establish the preceding claim. As B is an abelian normal subgroup of W and LA(R)B, B normalizes L and LW=LBQ=LQB where

LQ=(f,1)(1,q)(f,1)L,qQ.

Let (f,1)LQA(R) and

(f,1)=(f1,1)(1,q1)(fm,1)(1,qm)=(f1q1,1)(fmqm,1)

where (fi,1)L and qiQ for every i with 1im. For every element xSupp(f)R, there exists an i{1,,m} such that fiqi(x)=fi(qi-1x) is nontrivial. As (fi,1) is an element of LA(R), we must have qiR. Hence, (fiqi,1) is in L.

Since B is an abelian group, we can write (f,1)=(g1,1)(g2,1) such that (g1,1)L and (g2,1)=(fi1qi1,1)(fikqik,1) where qijR. For each element xSupp(f) if g2(x) is nontrivial, then by similar arguments as above there is j{1,,k} such that qijR, which is in contradiction with the choice of g2; whence, we have g2(x)=1. For each xSupp(f), we have two cases:

  1. g1(x)=g2(x)=1,

  2. g1(x)=g2-1(x)1.

In case (ii), since Supp(g1)R, by a similar argument as above there is an index j{1,,k} such that qijR, which again contradicts the choice of g2. Therefore, g2 is trivial and fL. So, LQA(R) is contained in L. The reverse inclusion is trivial. ∎

The following result is proved by Coulbois in his Ph.D. thesis [7]. We give another proof based on Theorem 3.8.

Corollary 3.9.

A f.g. free metabelian group is LERF.

Proof.

Via the Magnus embedding, the free metabelian group can be embedded in a group AQ where A and Q are f.g. abelian groups. The corollary follows now since the class of LERF groups is closed under taking subgroups. ∎

A group G is called Hopfian if G is not isomorphic to a proper quotient of itself. The group Baumslag–Solitar group BS(2,3) is not Hopfian [3]. Mal’cev proved the following theorem.

Theorem 3.10 ([17, Statement 6.1.11]).

Every f.g. residually finite group is Hopfian.

Lemma 3.11.

Let A be a f.g. abelian group and let Q be an RF group. Then every f.g. subgroup H of A(Q) is closed in the profinite topology of W=AQ.

Proof.

By Lemma 3.7, there is a finite subset T of Q and a finite index subgroup N of Q such that HA(T){1}A(T)N and A(T)N is a finite index subgroup of W. By Remark 2.8, it is enough to show that H is closed in the profinite topology of ATN.

By Lemma 2.4, there is a projection φ:A(T)NA(T){1} such that

φ(A(T){1})=A(T){1}.

By Theorem 3.10, the f.g. and RF group A(T){1} is Hopfian. Hence,

A(T){1}kerφ={1},

which entails that ATN=kerφ(A(T){1}). By Theorem 2.2, W is RF and so, the subgroup ATN is RF. Hence, by Lemma 2.9, the induced topology on A(T){1} from ATN coincides with its own profinite topology and A(T){1} is closed in the profinite topology of ATN. Since A(T){1} is a f.g. abelian group and so, a LERF group, it follows that H is closed in A(T)N. ∎

Therefore, to determine whether a wreath product of a f.g. abelian group A with a LERF group Q is LERF, we only need to find conditions on Q such that for every f.g. subgroup H with π(H)1 (π is the natural projection from W onto Q), H is closed in the profinite topology of W=AQ.

4 On a question of J. O. Button

In [1], it is shown that the free solvable group Sn,d of solvable class d with n generators is not LERF for d3 and n2. In the following, we show that Sn,3 does not contain a strictly ascending HNN-extension.

The following lemma is an extension of [6, Proposition 2.2] obtain by dropping the requirement that the quotient group is an abelian group.

Lemma 4.1.

Let G be a group with a normal subgroup N such that G/N is an ERF (respectively a LERF) group. Let H=B,t be a subgroup of G which is a strictly ascending HNN-extension with (respectively f.g.) base B and stable letter t. Then the subgroup BN,t is also a strictly ascending HNN-extension.

Proof.

Let C=BN. Since N is a normal subgroup of G,

tCt-1tBt-1tNt-1BN=C.

Let us suppose that tCt-1=C. Since G/N is an ERF (respectively a LERF) group, the subgroup BN/N is closed in the profinite topology of G/N and so by Theorem 2.11, we have

BNN=(tBt-1)NN.

Hence, for every bB, there is an element b0B such that tb0t-1N=bN and so tb0-1t-1bNB=C as tBt-1<B. Since we assumed that tCt-1=C, there is c0C such that tb0-1t-1b=tc0t-1, whence b=tb0c0t-1tBt-1, which entails that B is equal to tBt-1 in contradiction with the assumption that tBt-1<B. Thus, we have tCt-1<C. ∎

Theorem 4.2.

The free solvable group Sn,3(n2) is not LERF but it does not contain a strictly ascending HNN-extension H=t,B where B is a f.g. subgroup.

Proof.

Let A be the f.g. free abelian group of rank n, Sn,2 be the free metabelian group of rank n, and let W=ASn,2. By the Magnus embedding, Sn,3 can be embedded in W. We will show that W does not contain a strictly ascending HNN-extension with a f.g. base. We assume by contradiction that the group H=B,ttBt-1B is a strictly ascending HNN-extension with f.g. base B. Let π:WSn,2 be the natural projection, N=kerπ, R=π(B), t1=π(t). Since, by Corollary 3.9, Sn,2 is a LERF group, by Lemma 4.1, BN,t is a strictly ascending HNN-extension. If t belongs to N, then as N is an abelian group, BN=(BN)t which contradicts the fact that BN,t is a strictly ascending HNN-extension. Hence, we have that π(t) is nontrivial. Since Sn,2 is a LERF group, R is a closed subgroup of Sn,2. So, Theorem 2.11 implies that the equalities R=π(B)=π(Bt)=Rt1 hold. Therefore, R is a normal subgroup of π(H)=R,t1. The f.g. subgroup R,t1 is closed in the profinite topology of the LERF group Sn,2. Hence, by Lemma 3.7 there is a finite set T of Sn,2 such that HA(T)R,t1. Since π(B)=R, it follows that B is contained in A(T)(R,t1)R. Let A1=A(T) and W1=A1R,t1.

Suppose that there is a positive integer j such that t1jR. Then A1R,t1R is a finite index subgroup of the group W1 containing B. Let

π1:A1R,t1RR

be the natural projection. We observe that π1 is a restriction of π to the subgroup A1R,t1R. Since π1(B)=R, we have

BA1R,t1=A1R,t1R.

Since A1R,t1 is an abelian group, BA1R,t1 is a normal subgroup of A1R,t1R. We have

BA1R,t1=BNandt1jA1R,t1R.

So, we obtain the following equalities:

BN=(BN)t1j=(BN)tj

where the last equality holds because t=(f,t1) for some fN and N is an abelian group. Since (BN)t is contained in (BN)tj, we have

(BN)t=BN,

which contradicts the assumption that BN,t is a strictly ascending HNN-extension. We thus conclude that Rt1={1} and so R,t1=Rt1.

Let B=(f1,(r1,1)),,(fm,(rm,1)) and

S=i=1nSupp(fi)={(r1,t1α1),,(rk,t1αk)},

for some elements f1,,fmA1(Rt1), (r1,t1α1),,(rk,t1αk)Rt1, and r1,,rmR. Let

T={(1,t1α1),,(1,t1αk)}.

The subset A1(T) is a subgroup of A1(Rt1) because for every f and g in A1(Rt1) we always have

  1. Supp(f)=Supp(f-1),

  2. Supp(fg)Supp(f)Supp(g).

The group A1(T)R is a subgroup of W1 and every generator of B is in this subgroup, whence the group B is contained in A1(T)R. Let α be an element of the set {α1,,αk} such that for every 1ik, |α||αi|. We take (g,(1,1)) in BN=BA1(T)(R). Since N is an abelian group and t=(f,(1,t1)) for some fN, we have

t(BN)t-1=(1,(1,t1))(BN)(1,(1,t1-1)).

Hence, for every positive integer i, the element

(1,(1,t1i))(g,(1,1))(1,(1,t1-i))=(g(1,t1i),1)

is in BN. Let (r,t1αj)Supp(g) with rR and t1αjT. For every positive integer i, we have that the element (rt1i,t1αj+i) is in the support of g(1,t1i) because

g(1,t1i)(rt1i,t1αj+i)=g((1,t1-i)(rt1i,t1αj+i))=g(r,t1αj).

We can take the positive integer i large enough so that |αj+i|>|α|. Hence, the element (g(1,t1i),1) is not in BN=BA1(T)(R), whence ti(BN)t-i is not contained in BN. This contradicts the fact that BN,t is a strictly ascending HNN-extension. ∎


Communicated by John S. Wilson


Award Identifier / Grant number: SFRH/BD/98202/2013

Award Identifier / Grant number: UID/MAT/00144/2013

Funding statement: The work was partially supported by the FCT Doctoral Grant with reference (SFRH/BD/98202/2013). It was also partially supported by CMUP (UID/MAT/00144/2013), which is funded by FCT (Portugal) with national (MEC) and European structural funds (FEDER), under the partnership agreement PT2020.

Acknowledgements

This work is part of the author’s preparation of a doctoral thesis under the supervision of Prof. Jorge Almeida, whose advice is gratefully acknowledged.

References

[1] S. A. Agalakov, Finite separability of groups and Lie algebras, Algebra i Logika 22 (1985), no. 4, 363–371. 10.1007/BF01979673Search in Google Scholar

[2] R. C. Alperin, Solvable groups of exponential growth and HNN extensions, Groups Korea ’98 (Pusan 1998), De Gruyter, Berlin (2000), 1–5. 10.1515/9783110807493-003Search in Google Scholar

[3] G. Baumslag and D. Solitar, Some two-generator one-relator non-Hopfian groups, Bull. Amer. Math. Soc. 68 (1962), 199–201. 10.1090/S0002-9904-1962-10745-9Search in Google Scholar

[4] K. A. Bencsáth, M. C. Bonanome, M. H. Dean and M. Zyman, Lectures on Finitely Generated Solvable Groups, Springer Briefs Math., Springer, New York, 2013. 10.1007/978-1-4614-5450-2Search in Google Scholar

[5] A. Blass and P. M. Neumann, An application of universal algebra in group theory, Michigan Math. J. 21 (1974), 167–169. 10.1307/mmj/1029001263Search in Google Scholar

[6] J. O. Button, Strictly ascending HNN extensions in soluble groups, Ric. Mat. 61 (2012), no. 1, 139–145. 10.1007/s11587-011-0120-2Search in Google Scholar

[7] T. Coulbois, Proprié,tés de Ribes–Zalesskiĭ, topologie profinie produit libre et généralisations, Ph.D. thesis, Université Paris 7, Paris, 2000. Search in Google Scholar

[8] Y. de Cornulier, Finitely presented wreath products and double coset decompositions, Geom. Dedicata 122 (2006), 89–108. 10.1007/s10711-006-9061-4Search in Google Scholar

[9] K. W. Gruenberg, Residual properties of infinite soluble groups, Proc. Lond. Math. Soc. (3) 7 (1957), 29–62. 10.1112/plms/s3-7.1.29Search in Google Scholar

[10] M. Hall, Jr., A topology for free groups and related groups, Ann. of Math. (2) 52 (1950), 127–139. 10.2307/1969513Search in Google Scholar

[11] A. V. Jategaonkar, Integral group rings of polycyclic-by-finite groups, J. Pure Appl. Algebra 4 (1974), 337–343. 10.1016/0022-4049(74)90013-9Search in Google Scholar

[12] W. Magnus, On a theorem of Marshall Hall, Ann. of Math. (2) 40 (1939), 764–768. 10.2307/1968892Search in Google Scholar

[13] A. L. Mal’cev, On homomorphisms onto finite groups, Uchen. Zap. Ivanov Gos. Ped. Inst. 18 (1956). 10.1090/trans2/119/08Search in Google Scholar

[14] V. D. Mazurov and E. I. Khukhro, The Kourovka Notebook. Unsolved Problems in Group Theory. 16th Edition, Institute of Mathematics, Russian Academy of Sciences, Novosibirsk, 2006. Search in Google Scholar

[15] H. Neumann, Varieties of Groups, Springer, New York, 1967. 10.1007/978-3-642-88599-0Search in Google Scholar

[16] L. Ribes and P. A. Zalesskiĭ, Profinite Groups, Ergeb. Math. Grenzgeb. (3) 40, Springer, Berlin, 2000. 10.1007/978-3-662-04097-3Search in Google Scholar

[17] D. J. S. Robinson, A Course in the Theory of Groups, Grad. Texts in Math. 80, Springer, New York, 1993. 10.1007/978-1-4684-0128-8_11Search in Google Scholar

[18] D. J. S. Robinson, A. Russo and G. Vincenzi, On groups whose subgroups are closed in the profinite topology, J. Pure Appl. Algebra 213 (2009), no. 4, 421–429. 10.1016/j.jpaa.2008.07.015Search in Google Scholar

[19] A. Shmelkin, On soluble products of groups, Sibirsk. Mat. Zh. 6 (1965), 212–220. Search in Google Scholar

Received: 2016-7-11
Revised: 2016-8-30
Published Online: 2016-10-11
Published in Print: 2017-7-1

© 2017 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 18.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/jgth-2016-0048/html
Scroll to top button