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

pp-nilpotency criteria for some verbal subgroups

Yerko Contreras Rojas Faculty of Mathematics – Institute of exact sciences
Universidade Federal do Sul e Sudeste do Pará
Avenida dos Ipês, Cidade Universitária, Marabá - Pará
Brazil
[email protected]
Valentina Grazian Department of Mathematics and Applications
University of Milano – Bicocca
Via Roberto Cozzi 55, 20125 Milano
Italy
[email protected]
 and  Carmine Monetta Department of Mathematics
University of Salerno
via Giovanni Paolo II 132, 84084 Fisciano (SA)
Italy
[email protected]
Abstract.

Let GG be a finite group, let pp be a prime and let ww be a group-word. We say that GG satisfies P(w,p)P(w,p) if the prime pp divides the order of xyxy for every ww-value xx in GG of pp^{\prime}-order and for every non-trivial ww-value yy in GG of order divisible by pp. With k2k\geq 2, we prove that the kkth term of the lower central series of GG is pp-nilpotent if and only if GG satisfies P(γk,p)P(\gamma_{k},p). In addition, if GG is soluble, we show that the kkth term of the derived series of GG is pp-nilpotent if and only if GG satisfies P(δk,p)P(\delta_{k},p).

Key words and phrases:
pp-nilpotency, lower central word, derived word
2010 Mathematics Subject Classification:
20D12, 20F15

1. Introduction

Let GG be a finite group and let pp be a prime. We say that GG is pp-nilpotent if GG has a normal pp-complement, that is, a normal subgroup HH of GG having pp^{\prime}-order and pp-power index in GG. There are well known results giving sufficient conditions for the pp-nilpotecy of a finite group, such as Burniside’s theorem [9, Theorem 7.4.3], and remarkable criteria like Frobenius’ normal pp-complement theorem [9, Theorem 7.4.5], which says that a group GG is pp-nilpotent if and only if NG(H)/CG(H)N_{G}(H)/C_{G}(H) is a pp-group for every pp-subgroup HH of GG. These results mainly deal with properties of pp-local subgroups and they motivated other works in the subject (see for instance [1] and [2]). In recent years, the problems of nilpotency and pp-nilpotency of a group have been studied from a different point of view. In an unpublished work of 2014 [6], Baumslag and Wiegold studied the nilpotency of a finite group looking at the behaviour of elements of coprime orders. Beltrán and Sáez [7] took inspiration to characterize the pp-nilpotency of a finite group by means of the orders of its elements:

Theorem A.

[7, Theorem A] Let GG be a finite group and let pp be a prime. Then GG is pp-nilpotent if and only if for every pp^{\prime}-element xx of GG of prime power order and for every pp-element y1y\neq 1 of GG, pp divides o(xy)o(xy).

Here o(x)o(x) denotes the order of the group element xx. We point out that, instead of studying the couple of elements (x,y)(x,y), where xx is a pp^{\prime}-element of prime power order and yy is a non-trivial pp-element, one can focus on the couple (x,y)(x,y), where xx is a pp^{\prime}-element and yy is a non-trivial element of order divisible by pp.

Corollary B.

Let GG be a finite group and let pp be a prime. Then GG is pp-nilpotent if and only if for every xGx\in G such that pp does not divide o(x)o(x) and for every 1yG1\neq y\in G such that pp divides o(y)o(y), pp divides o(xy)o(xy).

The work of Baumslag and Wiegold has been extended to the realm of group-words in a series of papers [3, 4, 5, 8, 12]. In a similar matter, the aim of this work is to generalize Theorem A obtaining a verbal version.

A group-word is any nontrivial element of a free group FF on free generators x1,x2,x_{1},x_{2},\dots, that is, a product of finitely many xix_{i}’s and their inverses. The elements of the commutator subgroup of FF are called commutator words. Let w=w(x1,,xk)w=w(x_{1},\dots,x_{k}) be a group-word in the variables x1,,xkx_{1},\dots,x_{k}. For any group GG and arbitrary g1,,gkGg_{1},\dots,g_{k}\in G, the elements of the form w(g1,,gk)w(g_{1},\dots,g_{k}) are called ww-values in GG. We denote by GwG_{w} the set of all ww-values in GG. The verbal subgroup of GG corresponding to ww is the subgroup w(G)w(G) of GG generated by GwG_{w}.

Definition.

Let GG be a group, let pp be a prime and let ww be a group-word. We say that GG satisfies P(w,p)P(w,p) if the prime pp divides o(xy)o(xy), for every xGwx\in G_{w} of pp^{\prime}-order and for every non-trivial yGwy\in G_{w} of order divisible by pp.

Let ww be a group-word and let pp be a prime. If GG is a finite group and w(G)w(G) is pp-nilpotent, then Corollary B implies that GG satisfies P(w,p)P(w,p). An interesting question is the following:

Question 1.

If GG is a finite group satisfying P(w,p)P(w,p), is w(G)w(G) pp-nilpotent?

In general the answer is negative. For instance, one may consider any non-abelian simple group GG, say of exponent ee, and the word xnx^{n}, where nn is a divisor of ee such that e/ne/n is prime. If pp is a prime dividing the order of GG, then GG satisfies P(w,p)P(w,p), but w(G)=Gw(G)=G is not pp-nilpotent.

Even in the case of commutator words we can find counterexamples. Indeed, if G=Alt(5)G=Alt(5) is the alternating group of degree 55 and ww is the word considered in [12, Example 4.2], then GwG_{w} consists of the identity and all products of two transpositions. In particular if p{2,3,5}p\in\{2,3,5\} then GG satisfies P(w,p)P(w,p), but w(G)=Gw(G)=G is a simple group and therefore not pp-nilpotent.

However, if we consider the group-word w=xw=x, Corollary B says exactly that w(G)w(G) is pp-nilpotent if and only if G satisfies P(w,p)P(w,p). Actually, this situation is a particular instance of a more general result concerning the commutator word γk\gamma_{k}.

Given an integer k1k\geq 1, the word γk=γk(x1,,xk)\gamma_{k}=\gamma_{k}(x_{1},\dots,x_{k}) is defined inductively by the formulae

γ1=x1,γk=[γk1,xk]=[x1,,xk] for k2,\gamma_{1}=x_{1},\quad\gamma_{k}=[\gamma_{k-1},x_{k}]=[x_{1},\dots,x_{k}]\leavevmode\nobreak\ \text{ for }k\geq 2,

where [x,y]=x1y1xy[x,y]=x^{-1}y^{-1}xy, for any group elements xx and yy. Note that γ2(G)=G\gamma_{2}(G)=G^{\prime} is the derived subgroup of GG and in general the subgroup γk(G)\gamma_{k}(G) of GG corresponds to the kk-th term of the lower central series of GG.

Our first main theorem is the following:

Theorem C.

Let GG be a finite group, let k1k\geq 1 be an integer and let pp be a prime. Then γk(G)\gamma_{k}(G) is pp-nilpotent if and only if GG satisfies P(γk,p)P(\gamma_{k},p).

It is worth mentioning that Theorem C gives a positive answer to Question 1 for γk\gamma_{k}-words. Such words belong to the larger class of multilinear commutator words, which are words obtained by nesting commutators but using always different variables. For example the word [[x1,x2],[x3,x4,x5],x6][[x_{1},x_{2}],[x_{3},x_{4},x_{5}],x_{6}] is a multilinear commutator, while the Engel word [x1,x2,x2][x_{1},x_{2},x_{2}] is not. Hence, it is natural to ask if the answer to Question 1 remains positive if GG is any finite group and the word considered is any multilinear commutator word. In this direction, we provide another positive answer when ww belongs to the family of δk\delta_{k}-words. Given an integer k0k\geq 0, the word δk=δk(x1,,x2k)\delta_{k}=\delta_{k}(x_{1},\dots,x_{2^{k}}) is defined inductively by the formulae

δ0=x1,δk=[δk1(x1,,x2k1),δk1(x2k1+1,,x2k)] for k1.\delta_{0}=x_{1},\quad\delta_{k}=[\delta_{k-1}(x_{1},\dots,x_{2^{k-1}}),\delta_{k-1}(x_{2^{k-1}+1},\dots,x_{2^{k}})]\leavevmode\nobreak\ \text{ for }k\geq 1.

We have δ1(G)=G\delta_{1}(G)=G^{\prime} and δk(G)\delta_{k}(G) corresponds to the kk-th term G(k)G^{(k)} of the derived series of GG. Since δh=γh+1\delta_{h}=\gamma_{h+1} for 0h10\leq h\leq 1, Theorem C gives an affirmative answer to Question 1 when w=δ0,δ1w=\delta_{0},\delta_{1}. For w=δkw=\delta_{k} with k2k\geq 2, we prove the following:

Theorem D.

Let GG be a finite soluble group, let k2k\geq 2 be an integer and let pp be a prime. Then G(k)G^{(k)} is pp-nilpotent if and only if GG satisfies P(δk,p)P(\delta_{k},p).

2. Preliminaries

Let GG be a finite group and let pp be a prime. We denote by Op(G)O_{p}(G) the largest normal pp-subgroup of GG and by Op(G)O_{p^{\prime}}(G) the largest normal pp^{\prime}-subgroup of GG. If GG is pp-nilpotent, then the normal pp-complement of GG is unique and corresponds to the group Op(G)O_{p^{\prime}}(G). In particular we can write G=POp(G)G=PO_{p^{\prime}}(G) for some Sylow pp-subgroup PP of GG. Moreover, if GG is pp-nilpotent then Op(G)O_{p^{\prime}}(G) contains all elements of GG having pp^{\prime}-order. Now, Corollary B is a direct consequence of Theorem A and basic properties of pp-nilpotent groups.

Proof of Corollary B.

Suppose GG is pp-nilpotent. Then it has a normal pp-complement HH. Let x,yGx,y\in G be such that pp does not divide o(x)o(x) and pp divides o(y)o(y), with y1y\neq 1. Then xHx\in H and xyHxy\not\in H as yHy\not\in H. Hence pp divides o(xy)o(xy). The other implication follows immediately from Theorem A. ∎

If GG is a group, we denote by F(G)F(G) the Fitting subgroup of GG, that is the largest normal nilpotent subgroup of GG, and by Fitp(G)Fit_{p}(G) the pp-Fitting subgroup of GG, that is the largest normal pp-nilpotent subgroup of GG. Note that F(G)Fitp(G)F(G)\leq Fit_{p}(G).

Lemma 2.1.

Let GG be a finite group and let pp be a prime. If Op(G)=1O_{p^{\prime}}(G)=1 then Fitp(G)=Op(G)Fit_{p}(G)=O_{p}(G) is a pp-group.

Proof.

We have Op(Fitp(G))Op(G)=1O_{p^{\prime}}(Fit_{p}(G))\leq O_{p^{\prime}}(G)=1. Since Fitp(G)Fit_{p}(G) is pp-nilpotent, we also have Fitp(G)=TOp(Fitp(G))Fit_{p}(G)=TO_{p^{\prime}}(Fit_{p}(G)) for some Sylow pp-subgroup TT of Fitp(G)Fit_{p}(G). Hence Fitp(G)=TFit_{p}(G)=T is a normal pp-subgroup of GG. Thus Fitp(G)Op(G)Fit_{p}(G)\leq O_{p}(G). On the other hand, Op(G)O_{p}(G) is a normal pp-nilpotent subgroup of GG, and so we conclude that Fitp(G)=Op(G)Fit_{p}(G)=O_{p}(G). ∎

A group GG is said to be metanilpotent if it has a normal subgroup NN such that both NN and G/NG/N are nilpotent.

Lemma 2.2.

[4, Lemma 3] Let pp be a prime and GG a finite metanilpotent group. Suppose that xx is a pp-element in GG such that [Op(F(G)),x]=1[O_{p^{\prime}}(F(G)),x]=1. Then xF(G)x\in F(G).

We recall that a subset XX of a group GG is said to be commutator-closed if [X,X]X[X,X]\subseteq X, and symmetric if X1=XX^{-1}=X. In [8, Lemma 2.1], it has been showed that a finite soluble group GG admits a commutator-closed subset XX such that G=XG=\langle X\rangle and every element of XX has prime power order. From the proof, it is easy to check that such an XX is also symmetric. Hence we have the following.

Lemma 2.3.

Let GG be a finite group. If GG is soluble then there exists a commutator-closed and symmetric subset XX of GG such that G=XG=\langle X\rangle and every element of XX has prime power order.

Next results give sufficient conditions for a group to be generated by certain ww-values of prime power order.

Lemma 2.4.

Let GG be a finite soluble group. Then for every k2k\geq 2 the group γk(G)\gamma_{k}(G) is generated by γk\gamma_{k}-values of prime power order.

Proof.

By Lemma 2.3 there exists a commutator-closed and symmetric subset XX of GG such that G=XG=\langle X\rangle and every element of XX has prime power order. Hence by [10, Lemma 3.6 item (c)] we get γk(G)=[x1,,xt]xiXX1,tk\gamma_{k}(G)=\langle[x_{1},\dots,x_{t}]\mid x_{i}\in X\cup X^{-1},t\geq k\rangle. Note that XX1=XX\cup X^{-1}=X, and [x1,,xt][x_{1},\dots,x_{t}] is a γk\gamma_{k}-value for every tkt\geq k. Also, since XX is commutator-closed, we get that [x1,,xt]X[x_{1},\dots,x_{t}]\in X has prime power order. Thus, γk(G)\gamma_{k}(G) is generated by γk\gamma_{k}-values of prime power order. ∎

Lemma 2.5.

[4, Lemma 4] Let kk be a positive integer and GG a finite group such that G=GG=G^{\prime}. If pp is a prime dividing the order of GG, then GG is generated by γk\gamma_{k}-values of qq-power order for primes qpq\neq p.

Lemma 2.6.

[8, Lemma 2.5] Let GG be a finite soluble group and let QQ be a Sylow qq-subgroup of GG. Then for every i1i\geq 1 the group QG(i)Q\cap G^{(i)} can be generated by δi\delta_{i}-values lying in QQ.

The following relation between the subgroups γk(G)\gamma_{k}(G) and G(k)G^{(k)} of GG is an immediate consequence of [10, Lemma 3.25].

Lemma 2.7.

If GG is a group, then for every k0k\geq 0 we have G(k)γk+1(G)G^{(k)}\leq\gamma_{k+1}(G).

We conclude this section with a direct application of [11, Theorem 1], which says that in almost every finite quasisimple group all elements are commutators.

Proposition 2.8.

Let GG be a finite quasisimple group and suppose that Z(G)Z(G) is a pp-group. Then every element of GG is a commutator, with the following exceptions:

  1. (1)

    p=3p=3, G/Z(G)𝐀6G/Z(G)\cong\mathbf{A}_{6}, Z(G)C3Z(G)\cong C_{3} and the non-central elements of GG that are not commutators have order 1212;

  2. (2)

    p=2p=2, G/Z(G)PSL(3,4)G/Z(G)\cong\mathrm{PSL}(3,4), Z(G)C2×C4Z(G)\cong C_{2}\times C_{4} and the non-central elements of GG that are not commutators have order 66;

  3. (3)

    p=2p=2, G/Z(G)PSL(3,4)G/Z(G)\cong\mathrm{PSL}(3,4), Z(G)C4×C4Z(G)\cong C_{4}\times C_{4} and the non-central elements of GG that are not commutators have order 1212.

3. Groups with Property P(w,p)P(w,p)

In this section ww is any group-word. We study the properties of groups satisfying property P(w,p)P(w,p) but being minimal such that w(G)w(G) is not pp-nilpotent.

Definition 3.1.

We say that a group GG is a minimal P(w,p)P(w,p)-exception if

  • GG satisfies P(w,p)P(w,p),

  • w(G)w(G) is not pp-nilpotent,

  • whenever HH is a proper subgroup of GG, the group w(H)w(H) is pp-nilpotent;

  • whenever G/NG/N is a proper quotient of GG satisfying P(w,p)P(w,p), the group w(G/N)w(G/N) is pp-nilpotent.

Next lemma shows that property P(w,p)P(w,p) is closed with respect to forming subgroups and certain images.

Lemma 3.2.

Let GG be a finite group satisfying P(w,p)P(w,p).

  • If HGH\leq G is a subgroup of GG then HH satisfies P(w,p)P(w,p).

  • If NGN\mathrel{\unlhd}G is a normal subgroup of GG of pp^{\prime}-order, then G/NG/N satisfies P(w,p)P(w,p).

Proof.

If HGH\leq G then for every hHwh\in H_{w} we have hGwh\in G_{w}. Thus HH satisfies P(w,p)P(w,p) if GG does.

Assume that NN is a normal subgroup of GG of pp^{\prime}-order and consider G¯=G/N\overline{G}=G/N. Let xNG¯wxN\in\overline{G}_{w} be a ww-value of pp^{\prime}-order and let 1yNG¯w1\neq yN\in\overline{G}_{w} be a ww-value of order divisible by pp. Then we can assume that x,yGwx,y\in G_{w}. Also, pp divides the order of yy and, since NN has pp^{\prime}-order, pp does not divide the order of xx. Thus by P(w,p)P(w,p) we deduce that pp divides the order of xyxy. In particular xyNxy\notin N and pp divides the order of xyN=xNyNxyN=xN\cdot yN. This shows that G¯\overline{G} satisfies property P(w,p)P(w,p). ∎

Lemma 3.3.

If GG is a minimal P(w,p)P(w,p)-exception then Op(G)=1O_{p^{\prime}}(G)=1 and Fitp(G)=Op(G)Fit_{p}(G)=O_{p}(G).

Proof.

Set N=Op(G)N=O_{p^{\prime}}(G). Aiming for a contradiction, suppose N1N\neq 1 and consider G¯=G/N\overline{G}=G/N. Then |G¯|<|G||\overline{G}|<|G| and by Lemma 3.2 the group G¯\overline{G} satisfies P(w,p)P(w,p). Since GG is a minimal P(w,p)P(w,p)-exception we deduce that w(G¯)=w(G)N/Nw(\overline{G})=w(G)N/N is pp-nilpotent. Then w(G)N/N=S/NH/Nw(G)N/N=S/N\cdot H/N where S/NSylp(w(G)N/N)S/N\in\hbox{\rm Syl}_{p}(w(G)N/N) and H/N=Op(w(G)N/N)H/N=O_{p^{\prime}}(w(G)N/N). Set K=Hw(G)K=H\cap w(G). Since Hw(G)NH\mathrel{\unlhd}w(G)N, we deduce that Kw(G)K\mathrel{\unlhd}w(G). Also, |K||K| divides |H|=[H:N]|N||H|=[H\colon N]|N| and so |K||K| is prime to pp. Finally, [w(G):K]=[w(G)H:H][w(G)\colon K]=[w(G)H\colon H] divides [w(G)N:H][w(G)N\colon H], that is a power of pp. We deduce that KK is a normal pp-complement of w(G)w(G) and w(G)w(G) is pp-nilpotent, a contradiction. Therefore we must have Op(G)=1O_{p^{\prime}}(G)=1. The second part of the statement now follows from Lemma 2.1.

Lemma 3.4.

Suppose GG is a minimal P(w,p)P(w,p)-exception, G>GG>G^{\prime} and G/w(G)G^{\prime}/w(G^{\prime}) is nilpotent. Then GG^{\prime} has a normal Sylow pp-subgroup.

Proof.

Note that GG^{\prime} satisfies P(w,p)P(w,p) by Lemma 3.2. Since GG is a minimal P(w,p)P(w,p)-exception and |G|<|G||G^{\prime}|<|G| we deduce that w(G)w(G^{\prime}) is pp-nilpotent. Note that w(G)w(G^{\prime}) is a normal subgroup of GG^{\prime}, so Op(w(G))Op(G)Op(G)=1O_{p^{\prime}}(w(G^{\prime}))\leq O_{p^{\prime}}(G^{\prime})\leq O_{p^{\prime}}(G)=1 by Lemma 3.3. Hence Op(w(G))=1O_{p^{\prime}}(w(G^{\prime}))=1 and since w(G)w(G^{\prime}) is pp-nilpotent we deduce that w(G)w(G^{\prime}) is a pp-group. In particular w(G)w(G^{\prime}) is nilpotent. By assumption G/w(G)G^{\prime}/w(G^{\prime}) is nilpotent. Hence the group GG^{\prime} is metanilpotent. Let F(G)F(G^{\prime}) denote the Fitting subgroup of GG^{\prime}. Then F(G)Fitp(G)Fitp(G)=Op(G)F(G^{\prime})\leq Fit_{p}(G^{\prime})\leq Fit_{p}(G)=O_{p}(G) (again by Lemma 3.3), so F(G)F(G^{\prime}) is a pp-group. Let PSylp(G)P\in\hbox{\rm Syl}_{p}(G^{\prime}) be a Sylow pp-subgroup. Then F(G)PF(G^{\prime})\leq P. On the other hand, [P,Op(F(G))]=[P,1]=1[P,O_{p^{\prime}}(F(G^{\prime}))]=[P,1]=1 and by Lemma 2.2 we get PF(G)P\leq F(G^{\prime}). Therefore P=F(G)P=F(G^{\prime}) is normal in GG^{\prime}. ∎

Lemma 3.5.

Suppose GG is a minimal P(w,p)P(w,p)-exception and soluble. If HH is a normal subgroup of GG and PSylp(H)P\in\hbox{\rm Syl}_{p}(H) then CH(P)PC_{H}(P)\leq P.

Proof.

By Lemma 3.3 we have Op(G)=1O_{p^{\prime}}(G)=1. Since HH is normal in GG we have Op(H)=1O_{p^{\prime}}(H)=1 and so Fitp(H)=Op(H)PFit_{p}(H)=O_{p}(H)\leq P (with the same proof used to show that Fitp(G)=Op(G)Fit_{p}(G)=O_{p}(G)). Then F(H)Fitp(H)PF(H)\leq Fit_{p}(H)\leq P and so CH(P)CH(F(H))C_{H}(P)\leq C_{H}(F(H)). Since GG is soluble, HH is soluble as well and by [9, Theorem 6.1.3] we get

CH(P)CH(F(H))F(H)Fitp(H)P.C_{H}(P)\leq C_{H}(F(H))\leq F(H)\leq Fit_{p}(H)\leq P.

Therefore CH(P)PC_{H}(P)\leq P. ∎

4. Proof of Theorem C

In this section we consider the word w=γkw=\gamma_{k} and we prove Theorem C.

Note that if a finite group GG satisfies property P(γ2,p)P(\gamma_{2},p) then it satisfies property P(γk,p)P(\gamma_{k},p) for every k2k\geq 2. However, in general the converse is not true. As an example, consider the group GG generated by g1,g2,g3,h1,h2g_{1},g_{2},g_{3},h_{1},h_{2} subject to the following relations

gi2=hj3=1,[g2,g1]=g3,[h1,g1]=[h1,g3]=h1,[h2,g3]=h2,h1g2=h2,h2g2=h1.g_{i}^{2}=h_{j}^{3}=1,\ \ [g_{2},g_{1}]=g_{3},\ \ [h_{1},g_{1}]=[h_{1},g_{3}]=h_{1},\ \ [h_{2},g_{3}]=h_{2},h_{1}^{g_{2}}=h_{2},\ \ h_{2}^{g_{2}}=h_{1}.

Then GG is a group of order 7272 with γ3(G)C3×C3\gamma_{3}(G)\cong C_{3}\times C_{3}, thus GG satisfies property P(γ3,3)P(\gamma_{3},3). However h2=[h2,g3]h_{2}=[h_{2},g_{3}] has order 33, g3=[g2,g1]g_{3}=[g_{2},g_{1}] has order 2 and h2g3h_{2}g_{3} has order 22. Thus GG does not satisfy property P(γ2,3)P(\gamma_{2},3).

Lemma 4.1.

Let GG be a quasisimple group. Then Gγ2=GγkG_{\gamma_{2}}=G_{\gamma_{k}} for every k2k\geq 2. In particular if GG is quasisimple, pp is a prime and k2k\geq 2, then GG satisfies property P(γk,p)P(\gamma_{k},p) if and only if it satisfies property P(γ2,p)P(\gamma_{2},p).

Proof.

It is enough to prove that Gγ2=Gγ3G_{\gamma_{2}}=G_{\gamma_{3}}, then the result will follow by induction on kk. Clearly every γ3\gamma_{3}-value is a commutator, so Gγ3Gγ2G_{\gamma_{3}}\subset G_{\gamma_{2}}.

Now, let g=[a,b]Gγ2g=[a,b]\in G_{\gamma_{2}} be a commutator. Using Ore’s conjecture and the fact that GG is quasisimple, we can write a=xza=xz where xx is a commutator and zZ(G)z\in Z(G). So g=[xz,b]=[x,b]Gγ3g=[xz,b]=[x,b]\in G_{\gamma_{3}}. Therefore Gγ2Gγ3G_{\gamma_{2}}\subset G_{\gamma_{3}}. This completes the proof. ∎

Lemma 4.2.

Let GG be a finite group, pp a prime and k2k\geq 2. Suppose that GG satisfies P(γk,p)P(\gamma_{k},p) and let xGγkx\in G_{\gamma_{k}} be a γk\gamma_{k}-value of pp^{\prime}-order. Then for every element gGg\in G, the element [g,k1x][g,_{k-1}x] has pp^{\prime}-order.

Proof.

Note that for every gGg\in G we have

[g,k1x]x1=x[g,k2x].[g,_{k-1}x]\cdot x^{-1}=x^{-[g,_{k-2}x]}.

Aiming for a contradiction, suppose there exists gGg\in G such that pp divides the order of [g,k1x][g,_{k-1}x]. Then [g,k1x]1[g,_{k-1}x]\neq 1 and by P(γk,p)P(\gamma_{k},p) we deduce that pp divides the order of [g,k1x]x1[g,_{k-1}x]\cdot x^{-1}. Thus pp divides the order of x[g,k2x]x^{-[g,_{k-2}x]}, that coincides with the order of xx, a contradiction. This proves the statement. ∎

Lemma 4.3.

Let GG be a finite group satisfying P(γk,p)P(\gamma_{k},p). If PP is a pp-subgroup of GG and xNG(P)x\in N_{G}(P) is such that xGγkx\in G_{\gamma_{k}} has pp^{\prime}-order, then [P,x]=1[P,x]=1.

Proof.

Let gPg\in P be an element. By Lemma 4.2 the element [g,k1x][g,_{k-1}x] has pp^{\prime}-order. On the other hand, since xNG(P)x\in N_{G}(P) we have that [g,k1x]P[g,_{k-1}x]\in P. Therefore the only possibility is [g,k1x]=1[g,_{k-1}x]=1. This shows that [P,k1x]=1[P,_{k-1}x]=1. Therefore by [9, Theorem 5.3.6] we get [P,x]=[P,k1x]=1[P,x]=[P,_{k-1}x]=1. ∎

Lemma 4.4.

Let GG be a finite group and pp a prime. Suppose G=GG=G^{\prime}, GG satisfies P(γk,p)P(\gamma_{k},p) and assume that GG is minimal (with respect to the order) such that γk(G)\gamma_{k}(G) is not pp-nilpotent. Then Fitp(G)=Op(G)=Z(G)Fit_{p}(G)=O_{p}(G)=Z(G) and GG is quasisimple.

Proof.

Note that GG is a minimal P(γk,p)P(\gamma_{k},p)-exception. Hence by Lemma 3.3 we have Fitp(G)=Op(G)Fit_{p}(G)=O_{p}(G). By Lemma 2.5 the group GG is generated by γk\gamma_{k}-values of pp^{\prime}-order. Since Fitp(G)=Op(G)Fit_{p}(G)=O_{p}(G), Lemma 4.2 implies that Fitp(G)Z(G)Fit_{p}(G)\leq Z(G). On the other hand, Z(G)Z(G) is abelian and so pp-nilpotent and it is normal in GG, so Z(G)Fitp(G)Z(G)\leq Fit_{p}(G). Thus Fitp(G)=Z(G)Fit_{p}(G)=Z(G).

It remains to prove that GG is quasisimple. Since G=GG=G^{\prime}, it is enough to show that G/Z(G)G/Z(G) is simple. Let N/Z(G)G/Z(G)N/Z(G)\mathrel{\unlhd}G/Z(G) be a proper normal subgroup of G/Z(G)G/Z(G). Then NGN\mathrel{\unlhd}G is a proper subgroup and NN has property P(γk,p)P(\gamma_{k},p) by Lemma 3.2. Hence γk(N)\gamma_{k}(N) is pp-nilpotent by minimality of GG. Therefore γk(N)Fitp(G)=Z(G)\gamma_{k}(N)\leq Fit_{p}(G)=Z(G). In particular γk+1(N)=[γk(N),N]=1\gamma_{k+1}(N)=[\gamma_{k}(N),N]=1 and so NN is nilpotent and NFitp(G)=Z(G)N\leq Fit_{p}(G)=Z(G). Thus N/Z(G)=1N/Z(G)=1. This shows that G/Z(G)G/Z(G) is simple and completes the proof. ∎

Lemma 4.5.

Let GG be a finite quasisimple group, pp a prime and k2k\geq 2. If GG satisfies P(γk,p)P(\gamma_{k},p) and Z(G)Z(G) is a pp-group, then every element of GG is a γk\gamma_{k}-value.

Proof.

If every element of GG is a commutator, then by induction on kk we deduce that every element of GG is a γk\gamma_{k}-value. Suppose GG contains elements that are not commutators. Then by Proposition 2.8 one of the following holds:

  1. (1)

    p=3p=3, G/Z(G)𝐀6G/Z(G)\cong\mathbf{A}_{6} and Z(G)C3Z(G)\cong C_{3};

  2. (2)

    p=2p=2, G/Z(G)PSL(3,4)G/Z(G)\cong\mathrm{PSL}(3,4) and Z(G)C2×C4Z(G)\cong C_{2}\times C_{4};

  3. (3)

    p=2p=2, G/Z(G)PSL(3,4)G/Z(G)\cong\mathrm{PSL}(3,4) and Z(G)C4×C4Z(G)\cong C_{4}\times C_{4}.

Using GAP we can check that none of the groups in the above list satisfies property P(γ2,p)P(\gamma_{2},p) (where p=3p=3 in the first case and p=2p=2 in the others). Hence by Lemma 4.1 the groups in the list do not satisfy property P(γk,p)P(\gamma_{k},p), a contradiction. ∎

Lemma 4.6.

Let GG be a finite group, pp a prime and k2k\geq 2. Suppose GG satisfies P(γk,p)P(\gamma_{k},p) and is minimal (with respect to the order) such that γk(G)\gamma_{k}(G) is not pp-nilpotent. Then GG is soluble and GG^{\prime} has a normal Sylow pp-subgroup.

Proof.

We first show that G>GG>G^{\prime}. Aiming for a contradiction, suppose G=GG=G^{\prime}. Then by Lemma 4.4 we deduce that GG is quasisimple and Z(G)=Op(G)Z(G)=O_{p}(G) is a pp-group. By Lemma 4.5 every element of GG is a γk\gamma_{k}-value. Let PGP\leq G be a pp-subgroup and let yNG(P)y\in\mathrm{N}_{G}(P) be an element of pp^{\prime}-order. Then yy is a γk\gamma_{k}-value of pp^{\prime}-order and by Lemma 4.3 we deduce that [P,y]=1[P,y]=1. Hence NG(P)/CG(P)N_{G}(P)/C_{G}(P) is a pp-group. By Frobenius criterion we conclude that GG is pp-nilpotent, a contradiction. Hence G>GG>G^{\prime}. The fact that GG^{\prime} has a normal Sylow pp-subgroup follows from Lemma 3.4.

It remains to show that GG is soluble. Since GG is a minimal P(γk,p)P(\gamma_{k},p)-exception and G>GG>G^{\prime}, γk(G)\gamma_{k}(G^{\prime}) is a pp-nilpotent normal subgroup of GG. Hence γk(G)Fitp(G)\gamma_{k}(G^{\prime})\leq Fit_{p}(G) is a pp-group by Lemma 3.3. By Lemma 2.7 applied to GG^{\prime} we get that G(k)=(G)(k1)γk(G)G^{(k)}=(G^{\prime})^{(k-1)}\leq\gamma_{k}(G^{\prime}). Thus G(k)G^{(k)} is a pp-group and GG is soluble. ∎

Proof of Theorem C.

If γk(G)\gamma_{k}(G) is pp-nilpotent then GG satisfies property P(γk,p)P(\gamma_{k},p) by Corollary B. Suppose that GG has property P(γk,p)P(\gamma_{k},p). Aiming for a contradiction, suppose GG is minimal (with respect to the order) such that γk(G)\gamma_{k}(G) is not pp-nilpotent. Then by Lemma 4.6 we get that GG is soluble and GG^{\prime} has a normal Sylow pp-subgroup TT. Set P=Tγk(G)Sylp(γk(G))P=T\cap\gamma_{k}(G)\in\hbox{\rm Syl}_{p}(\gamma_{k}(G)). Note that GG is a minimal P(γk,p)P(\gamma_{k},p)-exception, so by Lemma 3.5 we deduce that Cγk(G)(P)PC_{\gamma_{k}(G)}(P)\leq P. Since Pγk(G)P\mathrel{\unlhd}\gamma_{k}(G), Lemma 4.3 implies that every γk\gamma_{k}-value of GG has order divisible by pp. Using Lemma 2.4 we conclude that every γk\gamma_{k}-value of GG has pp-power order. Therefore γk(G)P\gamma_{k}(G)\leq P is a pp-group and so it is pp-nilpotent, a contradiction. This completes the proof. ∎

5. Proof of Theorem D

In this section we consider the word w=δkw=\delta_{k} and we prove Theorem D.

Lemma 5.1.

Let GG be a finite group satisfying P(δk,p)P(\delta_{k},p) and let xGδkx\in G_{\delta_{k}} be a δk\delta_{k}-value of pp^{\prime}-order. Then for every element gGg\in G, the element [g,x,x][g,x,x] has pp^{\prime}-order.

Proof.

Let gGg\in G be an element and let xGδkx\in G_{\delta_{k}} be a δk\delta_{k}-value of pp^{\prime}-order. Note that

[g,x,x]=[xg,x]x[g,x,x]=[x^{-g},x]^{x}

is a δk\delta_{k}-value of GG. Also,

[g,x,x]x1=x[g,x].[g,x,x]\cdot x^{-1}=x^{-[g,x]}.

Now, x1x^{-1} and x[g,x]x^{-[g,x]} are δk\delta_{k}-values of GG of pp^{\prime}-order (their order is equal to the one of xx). Since GG satisfies P(δk,p)P(\delta_{k},p), we deduce that [g,x,x][g,x,x] has pp^{\prime}-order. ∎

Lemma 5.2.

Let GG be a finite group satisfying P(δk,p)P(\delta_{k},p). If PP is a pp-subgroup of GG and xNG(P)x\in N_{G}(P) is such that xGδkx\in G_{\delta_{k}} has pp^{\prime}-order, then [P,x]=1[P,x]=1.

Proof.

Let gPg\in P be an element. By Lemma 5.1 the element [g,x,x][g,x,x] has pp^{\prime}-order. On the other hand, since xNG(P)x\in N_{G}(P) we have that [g,x,x]P[g,x,x]\in P. Therefore the only possibility is [g,x,x]=1[g,x,x]=1. This shows that [P,x,x]=1[P,x,x]=1. Hence by [9, Theorem 5.3.6] we get [P,x]=[P,x,x]=1[P,x]=[P,x,x]=1. ∎

Proof of Theorem D.

If G(k)G^{(k)} is pp-nilpotent then GG satisfies property P(δk,p)P(\delta_{k},p) by Corollary B.

Suppose GG is soluble and satisfies P(δk,p)P(\delta_{k},p) but is minimal such that G(k)G^{(k)} is not pp-nilpotent. Note that every proper subgroup and every proper quotient of GG is soluble, so GG is a minimal P(δk,p)P(\delta_{k},p)-exception. By Lemma 3.2 the group GG^{\prime} satisfies P(δk,p)P(\delta_{k},p). Since GG is soluble, we have G>GG>G^{\prime} and so G(k+1)=δk(G)G^{(k+1)}=\delta_{k}(G^{\prime}) is pp-nilpotent. In particular G(k+1)G^{(k+1)} is contained in Fitp(G)Fit_{p}(G), that is a pp-group by Lemma 3.3. Let PSylp(G(k))P\in Syl_{p}(G^{(k)}). Then G(k+1)PG^{(k+1)}\leq P and

[P,G(k)][G(k),G(k)]=G(k+1)P.[P,G^{(k)}]\leq[G^{(k)},G^{(k)}]=G^{(k+1)}\leq P.

Thus PP is normal in G(k)G^{(k)}. By the Schur-Zassenhaus theorem [9, Theorem 6.2.1] there exists a subgroup HH of G(k)G^{(k)} of pp^{\prime}-order such that G(k)=PHG^{(k)}=PH. Let q1,q2,,qnq_{1},q_{2},\dots,q_{n} be all prime numbers dividing the order of G(k)G^{(k)} and distinct from pp, with qiqjq_{i}\neq q_{j} for every iji\neq j. Let QiQ_{i} be a Sylow qiq_{i}-subgroup of GG and set Qi^=QiG(k)Sylqi(G(k))\hat{Q_{i}}=Q_{i}\cap G^{(k)}\in\hbox{\rm Syl}_{q_{i}}(G^{(k)}). Since GG is soluble, by Lemma 2.6 the group Qi^\hat{Q_{i}} is generated by δk\delta_{k}-values of GG lying in QiQ_{i}. Note that H=Q1^,,Qn^H=\langle\hat{Q_{1}},\dots,\hat{Q_{n}}\rangle, so HH is generated by δk\delta_{k}-values of GG of pp^{\prime}-order. Since PG(k)P\mathrel{\unlhd}G^{(k)}, we deduce that [P,H]=1[P,H]=1 by Lemma 5.2. Thus HH is a normal pp-complement of G(k)G^{(k)} and G(k)G^{(k)} is pp-nilpotent, a contradiction. This completes the proof. ∎

References

  • [1] M. Asaad, On pp-nilpotence of finite groups, J. Algebra 277 (2004), 157–164.
  • [2] A. Ballester-Bolinches and G. Xiuyun, Some results on p-nilpotence and solubility of finite groups, J. Algebra 228 (2000), 491–496.
  • [3] R. Bastos and C. Monetta, Coprime commutators in finite groups, Comm. Algebra 47 (2019), no. 10, 4137–4147.
  • [4] R. Bastos, C. Monetta, and P. Shumyatsky, A criterion for metanilpotency of a finite group, J. Group Theory 21 (2018), no. 4, 713–718.
  • [5] R. Bastos and P. Shumyatsky, A sufficient condition for nilpotency of the commutator subgroup, Sib. Math. J. 57 (2016), no. 5, 762–763.
  • [6] B. Baumslag and J. Wiegold, A sufficient condition for nilpotency in a finite group, preprint available at arXiv:1411.2877v1 [math.GR] (2014).
  • [7] A. Beltrán and A. Sáez, Existence of normal Hall subgroups by means of orders of products, Math. Nachr. 292 (2019), 720–723.
  • [8] J. da Silva Alves and P. Shumyatsky, On nilpotency of higher commutator subgroups of a finite soluble group, Arch. Math. (Basel) 116 (2021), no. 1, 1–6.
  • [9] D. Gorenstein, Finite groups, second ed., Chelsea Publishing Co., New York, 1980.
  • [10] E. I. Khukhro, p-automorphisms of finite p-groups, London Mathematical Society Lecture Note Series. 246, 1998.
  • [11] M. W. Liebeck, E. A. O’Brien, A. Shalev, and P. H. Tiep, Commutators in finite quasisimple groups, Bull. Lond. Math. Soc. 43 (2011), no. 6, 1079–1092.
  • [12] C. Monetta and A. Tortora, A nilpotency criterion for some verbal subgroups, Bull. Aust. Math. Soc. 100 (2019), no. 2, 281–289.