Home Mathematics Group varieties not closed under cellular covers and localizations
Article Publicly Available

Group varieties not closed under cellular covers and localizations

  • Daniel Herden EMAIL logo , Montakarn Petapirak and José L. Rodríguez
Published/Copyright: August 17, 2017

Abstract

A group homomorphism e:HG is a cellular cover of G if for every homomorphism φ:HG there is a unique homomorphism φ¯:HH such that φ¯e=φ. Group localizations are defined dually. The main purpose of this paper is to establish 20 varieties of groups which are not closed under taking cellular covers. This will use the existence of a special Burnside group of exponent p for a sufficiently large prime p as a key witness. This answers a question raised by Göbel in [12]. Moreover, by using a similar witness argument, we can prove the existence of 20 varieties not closed under localizations. Finally, the existence of 20 varieties of groups neither closed under cellular covers nor under localizations is presented as well.

1 Introduction

In recent years, there has been an increasing interest in cellular covers of groups. These were first treated categorically by the third named author and Scherer in [28] to compute the fundamental group of cellular approximations of spaces in the sense of Bousfield and Dror Farjoun. Many initially significant results were later published in [8] by Dror Farjoun, Göbel and Segev. They mainly studied cellular covers of nilpotent groups and finite groups, and also some properties of groups inherited by their cellular covers. It was observed that cellular covers of nilpotent groups of class n are nilpotent of class n as well, in particular, cellular covers of abelian groups are abelian. This shows that the variety 𝔑n of nilpotent groups of class n is closed under taking cellular covers as well as the variety 𝔄n of abelian groups of exponent n. Hence, we obtain examples of countably many distinct varieties which are closed under taking cellular covers. Some more studies on cellular covers of particular groups and of groups with specific additional properties have been conducted and can be found in a considerable amount of literature; see e.g. [6, 7, 9, 10, 11, 12, 26, 29]. At the moment, this area of research is still active with many interesting open questions.

In 2010, Göbel [12] examined cellular covers of R-modules and of varieties of groups, and established an example of countably many distinct varieties which are not closed under taking cellular covers by considering the Burnside variety 𝔅p of exponent p for primes p>1075:

Theorem 1.1.

It is clear that the only abelian subvarieties of Bp are the trivial variety {1} and the variety Ap of abelian groups of exponent p. But the variety

cell𝔅p:=HH is a cellular cover of G𝔅p

contains (for any prime p>1075) all abelian groups. Thus Bp is not cellular closed.

However, the question of whether or not there exist uncountably many such varieties was left open, and it is natural to ask:

Question 1.

Are there 20 varieties of groups which are not closed under cellular covers?

This constitutes the main motivation of this paper, where we will establish 20 pairwise distinct varieties of groups which are not cellular closed in Theorem 3.4. This result combines Ol’shanskii’s 1970 proof (see [23]) of the existence of a continuum of distinct varieties with his construction of a special Burnside group p of exponent p for any sufficiently large prime p. In addition, in Section 2 we provide the construction of countably many distinct varieties neither cellular closed nor finitely based with the help of an explicit system of relations. The existence of 20 pairwise distinct varieties of groups not closed under localizations, which acts as a dual result to Theorem 3.4, is presented in Section 4. In Section 5, the existence of 20 varieties of groups which are neither closed under cellular covers nor under localizations is established.

Our notation is standard, as found in [15, 17, 21, 25]. We write homomorphisms on the right-hand side.

2 Countably many varieties neither cellular closed nor finitely based

As in [12], we deal with varieties of groups, i.e., classes of groups which are closed under taking subgroups, quotients, and direct products, and consider the additional operation of taking cellular covers.

Definition 2.1.

A homomorphism e:HG is a cellular cover of G if every homomorphism φ:HG factors uniquely through e, i.e., there is a unique homomorphism φ¯:HH such that φ¯e=φ:

In a slight abuse of notation, we will often identify e:HG with the group H and therefore call H a cellular cover of G. With this in mind, a variety 𝔙 of groups is closed under cellular covers if it contains all cellular covers of groups in 𝔙. By cell𝔙 we will denote the smallest variety containing 𝔙 which is closed under taking cellular covers.

In general, if e:HG is a cellular cover, then e:HIme is a cellular cover too, and 1KereHIme1 is a central extension (Kere𝔷H). Moreover, if e:HG is a surjective cellular cover, then H/𝔷HG/𝔷G.

For any variety of groups 𝔙, let 𝔙 be the variety generated by the quotients G/𝔷G of groups G𝔙, and let 𝔙c denote the variety of central extensions of groups in 𝔙. Then cell𝔙𝔙c holds (see [12, Theorem 6.3]). As the variety 𝔑n satisfies 𝔑n=(𝔑n)c, for any integer n, it follows that these varieties are closed under cellular covers; for more details see [12].

The Burnside variety 𝔅n of exponent n is the variety defined by the law xn=1. That is, for any group G𝔅n, gn=1 for all gG. For convenience, we call any group in a Burnside variety a Burnside group.

In 1902, Burnside [3] asked if it is the case that every finitely generated group in which every element is of finite order must necessarily be finite. In 1968 Novikov and Adjan constructed a counterexample in [22], which established a finitely generated but infinite group for odd exponents n4381. Later, Ol’shanskii shortened their lengthy construction for odd n>1010 in [24]. By applying geometric ideas to defining relations in groups, Ol’shanskii gave in [25] for primes p>1075 an inductive construction for a particular group p which is infinite such that every nontrivial proper subgroup is cyclic of order p. This particular group is a so-called Tarski monster group which turns out to be a negative answer to Burnside’s question as well. We recall some essential properties of p.

  1. p is infinite.

  2. Every nontrivial proper subgroup of p is cyclic of order p.

  3. p is generated by two elements.

Any such group must be trivially simple.

Lemma 2.2.

If the group B satisfies conditions (i) and (ii), then B is simple.

Proof.

To show that B is simple, suppose that NB is a nontrivial normal subgroup of B. It follows that N is a proper subgroup of order p. Let g be any element in BN. Because N is normal and g is of finite order, g,N=gN is a proper finite subgroup of order >p, a contradiction. ∎

We recall one more useful result, cf. [25, Theorem 31.1].

Theorem 2.3.

If Bp=F/N is a free presentation of Bp, then N/[F,N] is a free abelian group of countable rank.

Remark 2.4.

Let X={x1,x2} be a set of two generators and F=F(X) the associated free group. We have an aspherical representation p=X| for a suitable set of relators. In particular, we have p=F/N with N=F the normal closure of the set of relators . It is clear that N/[F,N] is abelian, however, the asphericity of the presentation implies that {r=1r} is an independent set of relations and, consequently, that {r[F,N]r} is a basis of N/[F,N].

Let F/N be a free presentation of the Burnside group p. Since the Schur multiplier H2(p,)=(FN)/[F,N] is a subgroup of N/[F,N], H2(p,) is free abelian as well. More precisely, it is well known that the Schur multiplier H2(p,) is free abelian of infinite countable rank, cf. [25, Corollary 31.2].

For the next step, we need some more tools: A group G is said to be perfect if it is equal to its own commutator subgroup, i.e., G=G=[G,G]; see e.g. [27]. Following [2], an epimorphism f:GQ is called a perfect cover of Q if G is perfect and Kerf𝔷G. We note that [17, Theorem 2.10.3] transfers immediately from the case of finite perfect groups to the infinite perfect group p.

Lemma 2.5.

The group Bp=F/N has a central extension

1KA𝑒p1,

where K:=(FN)/[F,N], A:=F/[F,N], and

e:F/[F,N]F/N(f[F,N]fN)

is a perfect cover.

Remark 2.6.

This central extension is also known as the universal perfect cover of p and has a number of important properties; see for example [2] and [17]. In a similar fashion, a universal perfect cover is possible for any perfect group. Observe here that p is clearly perfect as it is a non-abelian simple group.

The perfect cover e is also a cellular cover of p. The following argument is due to Göbel [12].

Lemma 2.7.

The group Bp has a cellular cover e:ABp, where Bp=F/N, A=F/[F,N], and

e:F/[F,N]F/N(f[F,N]fN).

The group A contains a copy of Z as subgroup.

Proof.

We have to show that e:Ap induces a bijection

e*:Hom(A,A)Hom(A,p)(φ¯φ¯e).

First, we claim that the map e* is surjective: Let φHom(A,p). We must find φ¯Hom(A,A) such that φ¯e=φ. If φ=0, then we choose φ¯=0. So φ¯e=φ clearly holds. Next, we assume that φ0, then 1Aφp. If Aφp, then, by the choice of p, Aφ is cyclic hence abelian. Since A is perfect, we have that

Aφ=[A,A]φ=[Aφ,Aφ],

i.e., Aφ is perfect as well. Consequently, we obtain that Aφ=[Aφ,Aφ]=1 which contradicts the assumption that φ0. Thus, we now have that Aφ=p, i.e., φ is surjective. As p is simple, we know that 𝔷p=1. Moreover, we can see that (𝔷A)φ𝔷p=1 because φ is surjective. From the above central extension, we obtain that K𝔷AKerφ and φ=eβ follows for some βHom(p,p). Next, we claim that βAut(p). Since φ0 and p is simple, we have that Kerβ=1, which shows that β is injective. We can see that β is also surjective by using the fact that φ is surjective. Thus, we finally have that βAut(p), hence Kerφ=Kere=K.

We next consider the following diagram:

where the map λ is an epimorphism defined by sending f[F,N]F/[F,N] to fNF/N. By the help of [17, Lemma 2.4.1], we obtain a homomorphism α:F/[F,N]A such that the above diagram commutes, i.e., αe=λβ. Recall that A=F/[F,N]F/[F,N]. Hence, α:F/[F,N]A can be restricted to φ¯:=αA. Then φ¯Hom(A,A). But e=λA, and hence φ=eβ=φ¯e.

Next, we will show that e* is injective: Let φ1,φ2Hom(A,A) such that φ1e=φ2e. Define

ψ:AA(a(aφ1)(a-1φ2)).

It is easy to see that AψKere=K. For a,bA, we obtain that

(ab)ψ=(ab)φ1(ab)-1φ2=aφ1(bφ1b-1φ2)a-1φ2=aφ1bψa-1φ2.

Using the fact that K=𝔷A, since bψK and a-1φ2A, we then obtain

(ab)ψ=aφ1a-1φ2bψ=aψbψ,

i.e., ψHom(A,K). Moreover, as A is perfect and AψK=𝔷A is abelian, we obtain Aψ=[A,A]ψ=[Aψ,Aψ]=1. As a result, ψ=0 and then

1=aφ1a-1φ2=aφ1(aφ2)-1

for all aA. This shows that φ1=φ2. Thus e* is injective, as desired.

Therefore, we have that e:Ap is a cellular cover of p. Note that A contains K=H2(p,) which is free abelian of infinite countable rank. ∎

Remark 2.8.

In fact, Lemma 2.5 provides the universal central extension of the perfect group p. In this context, Lemma 2.7 gives a more direct proof of the general result [8, Lemma 3.10] that every universal central extension corresponds to a surjective cellular cover.

For our countably many varieties neither cellular closed nor finitely based, let 𝔙p be the variety defined by the following infinite system of group laws:

(2.1)[x1,x2]p=1,([x1,x2][x3,x4])p=1,([x1,x2][x2k-1,x2k])p=1,

Ol’shanskii proved in [25] for all primes p>1010 that 𝔙p is not finitely based. In fact, [25, Lemma 31.4] provides a system of groups witnessing the non-finiteness of (2.1) within 𝔅p2. Thus the infinite system of group laws

(2.2)

x1p2=1,
[x1,x2]p=1,
([x1,x2][x2k-1,x2k])p=1,

defines a variety 𝔚p𝔅p2 which is not finitely based as well. In addition, for primes p>1075, this variety 𝔚p will contain the group p but clearly not the group of integers. By Lemma 2.7, the cellular cover A of p cannot belong to 𝔚p. This shows that this variety is not closed under taking cellular covers, which establishes the following result.

Theorem 2.9.

There exist countably many pairwise distinct varieties of groups which are neither cellular closed nor finitely based.

3 Uncountably many varieties not closed under cellular covers

As every variety can be described by a suitable system of group laws and the set of possible laws has cardinality 0, this immediately implies that there exist not more than 20 varieties, and the question whether there indeed exist 20 pairwise distinct varieties was raised by B. H. Neumann [20] in 1937. One might expect that a positive answer to this question can be given as follows: Start with a countable system of laws which is not equivalent to any finite system of laws. It should then be possible to construct 20 pairwise distinct varieties by choosing suitable subsets of this system of laws. This idea was realized in the constructions of 20 varieties of groups by Adjan [1] and Vaughan-Lee [30]. Unfortunately, not every infinite system of laws qualifies for this approach and, in particular, the system of laws (2.1) is unsuitable as every law implies all its predecessors (by replacing some variables by the identity). Here we will abstain from this approach and utilize instead the ideas of a construction presented by Ol’shanskii [23] in 1970.

We denote the variety of abelian groups of exponent n by 𝔄n. Applying some properties of locally finite varieties and monolithic groups, Ol’shanskii’s paper [23] proved the following result.

Lemma 3.1.

Let e be a positive integer, and assume that there exists an infinite series of finite groups Ti,iω, such that for each i,

  1. Ti is from a fixed locally finite variety 𝔙𝔅e,

  2. Ti is monolithic, and

  3. Ti is not isomorphic to any factor of Tj for ij.

If p is a prime not dividing e, then the product variety ApV has 20 pairwise distinct subvarieties.

Remark 3.2.

Note that 𝔄p𝔙 as product of locally finite varieties is again locally finite, cf. [27, Theorem 14.3.1].

The main implication from [23] is the following corollary.

Corollary 3.3.

The following statements hold.

  1. For any e=8p1, where p1 is an odd prime, the conditions of Lemma 3.1 are satisfied by a suitable series Ti, iω, of finite solvable groups of length 4.

  2. For any distinct odd primes p1 and p2 there exist 20 pairwise distinct subvarieties of the locally finite variety of length 5 solvable groups of exponent 8p1p2.

We apply these results to obtain the existence of 20 pairwise distinct varieties of groups which are not closed under cellular covers. Moreover, this answers a question raised in [12].

Theorem 3.4.

There exist 20 pairwise distinct varieties of groups which are not closed under cellular covers.

Proof.

Let p3>1075 be a prime, and let 𝔅p3 be the associated Burnside variety and p3𝔅p3 the Burnside group mentioned in Section 2. Next, we apply the idea of enlarging two given varieties 𝔘 and 𝔙 by considering the product variety 𝔘𝔙, which consists of all extensions of a group in 𝔘 by a group in 𝔙.

Since 𝔅p3 is not the variety of all groups, by applying the right cancellation law for product varieties, we have that 𝔙𝔅p3𝔙𝔅p3 for any varieties 𝔙𝔙. See [21] for more details on varieties of groups.

Thus, there remain 20 pairwise distinct product varieties 𝔙𝔅p3, where 𝔙 is a variety obtained from Corollary 3.3 (b). It is clear that 𝔙𝔅p3 contains the group p3, as p3𝔅p3𝔙𝔅p3. Since 𝔙 is of exponent 8p1p2 and 𝔅p3 is of exponent p3, the product variety 𝔙𝔅p3 is of exponent 8p1p2p3. But, as shown in Lemma 2.7, there exists a cellular cover e:Ap3 such that A. Therefore, A cannot be of finite exponent and obviously is not in the product variety 𝔙𝔅p3. This shows the existence of 20 pairwise distinct varieties which are not closed under cellular covers. ∎

The next result actually strengthens Theorem 2.9.

Corollary 3.5.

There exist 20 pairwise distinct varieties of groups which are neither cellular closed nor finitely based.

Proof.

From Theorem 3.4 we obtain 20 varieties which are not cellular closed. As there are only countably many finitely based varieties, there actually must exist 20 pairwise distinct varieties which are neither cellular closed nor finitely based. ∎

Remark 3.6.

Observe that the existence of 20 pairwise distinct not finitely based varieties in Corollary 3.5 is derived only implicitly by set-theoretic considerations, while in Theorem 2.9 countably many such varieties are explicitly given with the help of a suitable system of group laws (2.2).

4 Uncountably many varieties not closed under localizations

In this section, we will present a result dual to Theorem 3.4 for localizations. Cellular covers are also known as co-localizations. We can see the resemblance between cellular covers and localizations in the following definition which can also be found in [13] and [14] (cf. [4, Lemma 2.1]).

Definition 4.1.

A group homomorphism e:GH is a localization of G if every homomorphism φ:GH factors uniquely through e, i.e., there is a unique homomorphism φ¯:HH such that eφ¯=φ. It is equivalent to saying that e:GH is a localization of G if the following diagram commutes:

In a slight abuse of notation, we will often identify e:GH with the group H and therefore call H a localization of G.

As with cellular covers, there has been research into which properties of groups transfer to their localizations; see [4] or [5] for motivating examples. Libman showed in [19] that localizations fail to preserve finiteness of groups, i.e., a localization of a finite group need not be finite. In particular, he proved for all even integers n10 that the canonical irreducible representation ρ:AnSO(n-1,) is a localization of An, where An is the alternating group on n letters and SO(n,) is the special orthogonal group of orthogonal n×n matrices with determinant 1 over the field of real numbers. Thus, there exists a localization of infinite order of the finite group An. So it is reasonable to ask about the existence of an upper bound for the cardinality of localizations of An. The answer to this question is given by the following result, keeping in mind that An is non-abelian and simple for all n5. A proof of this result can be found in [13] under the set-theoretic assumption ZFC + GCH and, more generally, without any set-theoretic restrictions in [14].

For more details, we quote the main theorem from [14] in Theorem 4.3 which needs the following definition.

Definition 4.2.

We say that a group G is suitable if the following conditions hold:

  1. G1 is finite with trivial center.

  2. Aut(G) is complete.

  3. If G¯Aut(G) and G¯G, then G¯=Inn(G).

Recall that a group H is said to be complete if its center 𝔷H and outer automorphism group Out(H):=Aut(H)/Inn(H) are both trivial; cf. [27, p. 412].

Theorem 4.3.

Let A be a family of suitable groups and μ an infinite cardinal such that μ0=μ. Then we can find a group H of cardinality λ=μ+ such that the following holds:

  1. The group H is simple. Moreover, if 1hH, then any element of H is a product of at most four conjugates of h.

  2. Any G𝒜 is a subgroup of H and any two different groups in 𝒜 have trivial intersection 1 when considered as subgroups of H. If 𝒜 is not empty, then H[𝒜]=H, where the A-socleH[𝒜] is the subgroup of H generated by all copies of groups G𝒜 in H.

  3. Any monomorphism φ:GH, G𝒜, is induced by some hH, i.e., there is an element hH such that φ=h*G.

  4. If G¯H is an isomorphic copy of some G𝒜, then the centralizer CH(G¯) is trivial.

  5. Any monomorphism φ:HH is an inner automorphism, i.e., φ=h* for some hH.

  6. The group H contains a free subgroup of rank μ.

Note that condition (vi) is an extra property of the group H derived from [14, Construction 3.4 (i)]. In particular, H is complete with conditions (i) and (v).

Next, consider a non-abelian finite simple group G. We see that condition (i) of Definition 4.2 holds trivially, and condition (ii) is an immediate consequence of [27, Theorem 13.5.10]. For condition (iii), let now G¯Aut(G) with G¯G. If G¯Inn(G), then G¯=Inn(G) by cardinality. Hence, assume G¯Inn(G). In this case, the simple group G¯ embeds canonically as a subgroup of Aut(G)/Inn(G). According to Schreier’s conjecture [27, p. 403], Aut(G)/Inn(G) is solvable. Thus, G is both non-abelian finite simple and solvable, a contradiction. Therefore, one obtains the following (cf. [13, Proposition 2 (a)]):

Lemma 4.4.

Any non-abelian finite simple group is suitable.

As an immediate consequence of Lemma 4.4 and Theorem 4.3 we have the following

Proposition 4.5.

The following statements hold.

  1. For any even n10 there exists a localization of An with a free subgroup of countable rank.

  2. If G is a non-abelian finite simple group and κ an infinite cardinal, then there exists a localization of G with a free subgroup of rank κ.

Proof.

(a) This is an immediate consequence of (b) as An is simple for all n5. Alternatively, we may use Libman’s localization ρ:AnSO(n-1,): We know that SO(3,), and hence any SO(n,) for n3, contains a free subgroup of rank 2; cf. [16, pp. 469–472]. But it is well known that a free group of rank 2 has a free subgroup of countable rank; cf. [18, Vol. II].

(b) All non-abelian finite simple groups are suitable, by Lemma 4.4. Thus, applying Theorem 4.3 for the choice 𝒜={G} and μ=κ0, there exists a group H containing G with the following properties:

  1. The group H is simple.

  2. Any monomorphism from G to H is a restriction of an inner automorphism of H.

  3. The centralizer of G in H is trivial.

  4. Any monomorphism from H to H is an inner automorphism of H.

  5. The group H contains a free subgroup of rank κ0κ.

It remains to show that the inclusion e:GH is a localization of G, i.e., we will show that for any φ:GH, there exists a unique φ¯:HH such that eφ¯=φ.

The proof taken from [13, Section 1] is the following. If φ=0, then choose φ¯=0. We obtain Geφ¯=Gφ¯=1=Gφ. Since H is simple and 1GKerφ¯, the zero map is the only possibility for φ¯.

If φ0, then we obtain that Kerφ=1 because KerφG and G is simple. Thus, φ is a monomorphism. By (iii), φ is a restriction of an inner automorphism of H, say φ=h¯*G, where h¯H. Choose φ¯=h¯*. Then eφ¯=φ follows.

Next, assume that φ:HH is another homomorphism such that eφ=φ. Consequently, φ0, and hence φ is a monomorphism of H, by (i). Thus, by (v), there exists hH such that φ=h*. We then obtain that

gh*=gφ=geφ=gφ=geφ¯=gφ¯=gh¯*

for all gG. Hence, h*=h¯* on G, and h-1gh=h¯-1gh¯ for all gG. This implies that h¯h-1CH(G). Therefore, (iv) implies h¯=h, hence φ¯=φ. This shows the uniqueness of φ¯ and completes the proof. ∎

Proposition 4.5(b) directly implies:

Corollary 4.6.

Any non-abelian finite simple group has a localization which contains a free subgroup of arbitrarily large cardinality.

This observation can be utilized to give an explicit example of 20 varieties neither closed under localizations nor finitely based.

Theorem 4.7.

There exist 20 pairwise distinct varieties of groups which are neither closed under localizations nor finitely based.

Proof.

Consider the result by Adjan [1] which shows that, for all odd n4381, the set of relations

(4.1)(xrnyrnx-rny-rn)n=1,

where r is a prime, is irreducible. In particular, any infinite set of such relations automatically generates a variety which is not finitely based. Let 𝔙 be any such variety. It is easy to see that every group of exponent n satisfies all the laws (4.1), more precisely, 𝔅n𝔙.

Let now n4381 be odd and divisible by 15 (e.g., n=4395). Note that A5 is a non-abelian finite simple group with exponent 30. For any element xA5, it follows that (xrn)2=x2rn=1 holds because 302n. Thus,

xrn=x-rn

which implies that the relation

(xrnyrnx-rny-rn)n=(xrnyrnxrnyrn)n=(xrnyrn)2n=1

is satisfied in A5. This shows that A5 is a member of 𝔙. But, by Corollary 4.6, this gives us that 𝔙 is not closed under localizations as otherwise 𝔙 would contain free groups of arbitrary rank, hence all groups, which is a contradiction. This establishes 20 pairwise distinct varieties of groups neither closed under localizations nor finitely based. ∎

Remark 4.8.

It is not apparent whether or not Adjan’s varieties 𝔙 are closed under cellular covers. In particular, our argumentation from Theorem 2.9 fails as 𝔄𝔙 for the variety 𝔄 of all abelian groups, a problem which seems difficult to bypass.

5 Uncountably many varieties neither closed under cellular covers nor under localizations

In this section, we will discuss some further consequences of Corollary 3.3(b).

Theorem 5.1.

There exist 20 pairwise distinct locally finite varieties of groups which are not closed under localizations.

Proof.

Let S be any non-abelian finite simple group and set 𝔚:=var{S}. Then 𝔚 is locally finite, cf. [21, Theorem 15.71]. Applying the same argument as in Theorem 3.4, we have 𝔙𝔚𝔙𝔚 for any varieties 𝔙𝔙. Thus, there exist 20 distinct product varieties 𝔙𝔚, where 𝔙 is a locally finite variety obtained from Corollary 3.3(b). It is clear that each product variety 𝔙𝔚 contains the group S, as S𝔚𝔙𝔚. Furthermore, 𝔙𝔚 is locally finite as it is a product variety of locally finite varieties. The same argument as in Theorem 4.7 applies to show that none of these 20 varieties can be closed under localizations. ∎

Remark 5.2.

We make the following remarks.

  1. Note that Adjan’s varieties from Theorem 4.7 do contain the variety 𝔄 of all abelian groups and thus are not locally finite.

  2. Instead of the product variety 𝔙𝔚, we might use 𝔙𝔚, the minimal variety containing both 𝔙 and 𝔚. If p is a prime with gcd(p,|S|)=1 and 𝔙 a variety of exponent pk for some k, then 𝔙𝔚 will be the collection of all direct products of a group from 𝔙 and a group from 𝔚. As 𝔙 is uniquely determined by 𝔙𝔚 via 𝔙=(𝔙𝔚)𝔅pk, we have 𝔙𝔚𝔙𝔚 for any varieties 𝔙𝔙. Thus, the statement would follow from the existence of 20 pairwise distinct locally finite varieties of exponent pk.

Once again, by set-theoretic considerations, we obtain the following result which is dual to Corollary 3.5.

Corollary 5.3.

There exist 20 pairwise distinct locally finite varieties of groups which are neither closed under localizations nor finitely based.

It is not surprising that, combining the ideas from Theorem 3.4 and Theorem 5.1, we obtain the following result as an extra gift.

Theorem 5.4.

There exist 20 pairwise distinct varieties of groups which are neither closed under cellular covers nor under localizations.

Proof.

Recall from Theorem 3.4 that we obtain 20 pairwise distinct product varieties 𝔙𝔅p3 of exponent 8p1p2p3, where 𝔙 is a variety obtained from Corollary 3.3 (b). Let S be any non-abelian finite simple group and set 𝔚:=var{S}. Since S is finite, 𝔚 is a locally finite variety of exponent at most the order of the group S. By multiplying the variety 𝔚 on the right-hand side of each product variety 𝔙𝔅p3, we obtain product varieties of the form 𝔙𝔅p3𝔚 for each variety 𝔙 obtained from Corollary 3.3(b). Because 𝔚 is not the variety of all groups and the 20 varieties 𝔙𝔅p3 are all pairwise distinct, the right cancellation law for product varieties can be applied to assure that there still remain 20 pairwise distinct product varieties of the form 𝔙𝔅p3𝔚.

However, we see that the two groups p3 and S are members of each variety 𝔙𝔅p3𝔚 as p3𝔅p3𝔙𝔅p3𝔙𝔅p3𝔚 and S𝔚𝔙𝔅p3𝔚. From Section 2 we know that p3 has cellular covers of infinite exponent and from Section 4 we know that S has localizations of infinite exponent while each variety 𝔙𝔅p3𝔚 is of finite exponent (at most 8p1p2p3q where q is the exponent of 𝔚). It follows that each product variety 𝔙𝔅p3𝔚 is neither closed under taking cellular covers nor closed under taking localizations. ∎


Dedicated to the memory of Rüdiger Göbel



Communicated by Alexander Olshanskii


Award Identifier / Grant number: MTM2013-42178-P

Award Identifier / Grant number: MTM2016-76453-C2-2-P

Award Identifier / Grant number: FQM-213

Funding statement: The third named author was supported by project MTM2013-42178-P, funded by the Spanish Ministry of Economy and Competitiveness, A EI/FEDER grant MTM2016-76453-C2-2-P, and by the Junta de Andalucía Grant FQM-213.

Acknowledgements

The authors would like to thank the editor for his insightful comments on the paper.

References

[1] S. I. Adjan, Infinite irreducible systems of group identities, Izv. Akad. Nauk SSSR Ser. Mat. 34 (1970), 715–734. 10.1070/IM1970v004n04ABEH000928Search in Google Scholar

[2] F. R. Beyl and J. Tappe, Group Extensions, Representations, and the Schur Multiplicator, Lecture Notes in Math. 958, Springer, Berlin, 1982. 10.1007/BFb0067022Search in Google Scholar

[3] W. Burnside, On an unsettled question in the theory of discontinuous groups, Quart. J. Pure Appl. Math. 33 (1902), 230–238. Search in Google Scholar

[4] C. Casacuberta, On structures preserved by idempotent transformations of groups and homotopy types, Crystallographic Groups and Their Generalizations (Kortrijk 1999), Contemp. Math. 262, American Mathematical Society, Providence (2000), 39–68. 10.1090/conm/262/04167Search in Google Scholar

[5] C. Casacuberta, J. L. Rodríguez and J.-Y. Tai, Localizations of abelian Eilenberg–Mac Lane spaces of finite type, Algebr. Geom. Topol. 16 (2016), no. 4, 2379–2420. 10.2140/agt.2016.16.2379Search in Google Scholar

[6] W. Chachólski, E. Damian, E. D. Farjoun and Y. Segev, The A-core and A-cover of a group, J. Algebra 321 (2009), no. 2, 631–666. 10.1016/j.jalgebra.2008.09.022Search in Google Scholar

[7] W. Chachólski, E. D. Farjoun, R. Göbel and Y. Segev, Cellular covers of divisible abelian groups, Alpine Perspectives on Algebraic Topology, Contemp. Math. 504, American Mathematical Society, Providence (2009), 77–97. 10.1090/conm/504/09876Search in Google Scholar

[8] E. D. Farjoun, R. Göbel and Y. Segev, Cellular covers of groups, J. Pure Appl. Algebra 208 (2007), no. 1, 61–76. 10.1016/j.jpaa.2005.11.004Search in Google Scholar

[9] E. D. Farjoun, R. Göbel, Y. Segev and S. Shelah, On kernels of cellular covers, Groups Geom. Dyn. 1 (2007), no. 4, 409–419. 10.4171/GGD/20Search in Google Scholar

[10] L. Fuchs, Cellular covers of totally ordered abelian groups, Math. Slovaca 61 (2011), no. 3, 429–438. 10.2478/s12175-011-0021-6Search in Google Scholar

[11] L. Fuchs and R. Göbel, Cellular covers of abelian groups, Results Math. 53 (2009), no. 1–2, 59–76. 10.1007/s00025-008-0327-zSearch in Google Scholar

[12] R. Göbel, Cellular covers for R-modules and varieties of groups, Forum Math. 24 (2012), no. 2, 317–337. 10.1515/form.2011.064Search in Google Scholar

[13] R. Göbel, J. L. Rodriguez and S. Shelah, Large localizations of finite simple groups, J. Reine Angew. Math. 550 (2002), 1–24. 10.1515/crll.2002.072Search in Google Scholar

[14] R. Göbel and S. Shelah, Constructing simple groups for localizations, Comm. Algebra 30 (2002), no. 2, 809–837. 10.1081/AGB-120013184Search in Google Scholar

[15] R. Göbel and J. Trlifaj, Approximations and Endomorphism Algebras of Modules. Volume 1: Approximations, extended ed., De Gruyter Exp. Math. 41, Walter de Gruyter, Berlin, 2012. 10.1515/9783110218114Search in Google Scholar

[16] F. Hausdorff, Grundlagen der Mengenlehre, Leipzig, 1914. Search in Google Scholar

[17] G. Karpilovsky, The Schur Multiplier, London Math. Soc. Monogr. Ser. New Ser. 2, The Clarendon Press, New York, 1987. Search in Google Scholar

[18] A. G. Kurosh, The Theory of Groups, 2nd ed., Chelsea Publishing, New York, 1960. Search in Google Scholar

[19] A. Libman, A note on the localization of finite groups, J. Pure Appl. Algebra 148 (2000), no. 3, 271–274. 10.1016/S0022-4049(00)00019-0Search in Google Scholar

[20] B. H. Neumann, Identical relations in groups. I, Math. Ann. 114 (1937), no. 1, 506–525. 10.1007/BF01594191Search in Google Scholar

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

[22] P. S. Novikov and S. I. Adjan, Infinite periodic groups. III, Izv. Akad. Nauk SSSR Ser. Mat. 32 (1968), 709–731. 10.1070/IM1968v002n03ABEH000653Search in Google Scholar

[23] A. J. Ol’šanskiĭ, The finite basis problem for identities in groups, Izv. Akad. Nauk SSSR Ser. Mat. 34 (1970), 376–384. 10.1070/IM1970v004n02ABEH000911Search in Google Scholar

[24] A. Y. Ol’shanskii, The Novikov–Adjan theorem, Mat. Sb. 118 (1982), 203–235. Search in Google Scholar

[25] A. Y. Ol’shanskii, Geometry of Defining Relations in Groups, Kluwer Academic Publishers, Dordrecht, 1991. 10.1007/978-94-011-3618-1Search in Google Scholar

[26] M. Petapirak, Varieties of groups and cellular covers, PhD thesis, University of Duisburg-Essen, Essen, 2014. Search in Google Scholar

[27] D. J. S. Robinson, A Course in the Theory of Groups, 2nd ed., Grad. Texts in Math. 80, Springer, New York, 1996. 10.1007/978-1-4419-8594-1Search in Google Scholar

[28] J. L. Rodríguez and J. Scherer, Cellular approximations using Moore spaces, Cohomological Methods in Homotopy Theory (Bellaterra 1998), Progr. Math. 196, Birkhäuser, Basel (2001), 357–374. 10.1007/978-3-0348-8312-2_22Search in Google Scholar

[29] J. L. Rodríguez and L. Strüngmann, On cellular covers with free kernels, Mediterr. J. Math. 9 (2012), no. 2, 295–304. 10.1007/s00009-010-0109-1Search in Google Scholar

[30] M. R. Vaughan-Lee, Uncountably many varieties of groups, Bull. Lond. Math. Soc. 2 (1970), 280–286. 10.1112/blms/2.3.280Search in Google Scholar

Received: 2017-2-21
Published Online: 2017-8-17
Published in Print: 2017-11-1

© 2017 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 30.1.2026 from https://www.degruyterbrill.com/document/doi/10.1515/jgth-2017-0021/html?lang=en
Scroll to top button