The fundamental group of the $p$-subgroup complex
Elias Gabriel Minian, Kevin Ivan Piterman

TL;DR
This paper investigates the fundamental group of the $p$-subgroup complex in finite groups, providing the first example of a non-free fundamental group and reducing the problem to the almost simple case.
Contribution
It demonstrates that the fundamental group can be non-free and establishes a reduction of the problem to almost simple groups under a conjecture.
Findings
First concrete example of a non-free fundamental group in a $p$-subgroup complex
Reduction of the fundamental group study to almost simple groups
Identification of families with free fundamental groups
Abstract
We study the fundamental group of the -subgroup complex of a finite group . We show first that is not a free group (here is the alternating group on letters). This is the first concrete example in the literature of a -subgroup complex with non-free fundamental group. We prove that, modulo a well-known conjecture of M. Aschbacher, , where is a free group and is free if is not almost simple. Here . This result essentially reduces the study of the fundamental group of -subgroup complexes to the almost simple case. We also exhibit various families of almost simple groups whose -subgroup complexes have free fundamental group.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
The fundamental group of the -subgroup complex
Elías Gabriel Minian
and
Kevin Iván Piterman
Departamento de Matemática
IMAS-CONICET
FCEyN, Universidad de Buenos Aires. Buenos Aires, Argentina.
[email protected] ; [email protected]
Abstract.
We study the fundamental group of the -subgroup complex of a finite group . We show first that is not a free group (here is the alternating group on letters). This is the first concrete example in the literature of a -subgroup complex with non-free fundamental group. We prove that, modulo a well-known conjecture of M. Aschbacher, , where is a free group and is free if is not almost simple. Here . This result essentially reduces the study of the fundamental group of -subgroup complexes to the almost simple case. We also exhibit various families of almost simple groups whose -subgroup complexes have free fundamental group.
Key words and phrases:
-subgroups, posets, finite groups, fundamental group.
2010 Mathematics Subject Classification:
20J99, 20J05, 20D20, 20D30, 05E18, 06A11.
Partially supported by grants UBACyT 20020160100081BA
1. Introduction
Let be a finite group and a prime number dividing its order. The posets and consist of, respectively, the non-trivial -subgroups of and the non-trivial elementary abelian -subgroups of , ordered by inclusion. In [Qui78] D. Quillen studied homotopical properties of the poset by means of its order complex (which is now called the Quillen complex of at ). Recall that the order complex of a finite poset is the simplicial complex with simplices the non-empty chains of . Quillen showed that the inclusion is a homotopy equivalence at the level of complexes and conjectured that has a non-trivial normal -subgroup when (or equivalently, ) is contractible. Although Quillen’s conjecture remains open, significant progress has been made. In the early nineties Aschbacher and Smith obtained the most relevant partial confirmation of the conjecture so far [AS93].
The homotopy type of the -subgroup complex is, in general, not known. Quillen showed that has the homotopy type of a bouquet of spheres for a group of Lie type in characteristic or for , where is a solvable -group and is an elementary abelian -group acting on (see [Qui78]). In [PW00] Pulkus and Welker obtained a wedge decomposition for when has a solvable normal -subgroup. There is a question in [PW00], attributed to Thévenaz, whether the -subgroup complex always has the homotopy type of a bouquet of spheres (of possibly different dimensions). In 2004 Shareshian gave the first example of a group whose -subgroup complex is not homotopy equivalent to a bouquet of spheres. Concretely, he showed that there is torsion in the second homology group of (see [Sha04]). Here, denotes the symmetric group on letters.
The fundamental group of the -subgroup complex was first studied by M. Aschbacher. He provided some algebraic conditions for to be simply connected, modulo a well-known conjecture for which there is considerable evidence (see [Asc93, Theorems 1 & 2]). Later, K. Das studied the simple connectivity of the -subgroups complexes of some groups of Lie type (see [Das95, Das98, Das00]). In [Kso03, Kso04], Ksontini investigated the fundamental group of and proved that it is free except possibly for or . In [Sha04], Shareshian extended Ksontini’s results and showed that the fundamental group of is also free for . All these results could suggest that the fundamental group of is always free. We will show that this holds for solvable groups (see Corollary 1.2 below) and, modulo Aschbacher’s conjecture, for -solvable groups (see Corollary 1.1 below). In fact, there are only few known examples of -subgroup complexes which are not homotopy equivalent to a bouquet of spheres, and Shareshian’s counterexample fails in the second homology group but it does have free fundamental group. Surprisingly we found that the fundamental group of is not free (here is the alternating group on letters). Its fundamental group is isomorphic to a free product of the free group on generators and a non-free group whose abelianization is . This is the smallest group with non-free for some . Note that the integral homology of () is free abelian (cf. [Sha04, p.306]), so in this case the obstruction to being a bouquet of spheres relies on the fundamental group and not on the homology.
We will show that -subgroup complexes with non-free fundamental group are rather exceptional. The first of our main results asserts that, modulo Aschbacher’s conjecture, the study of freeness of reduces to the almost simple case. Concretely, we prove the following.
Theorem 5.1.
Let be a finite group and a prime dividing . Assume that Aschbacher’s conjecture holds. Then there is an isomorphism , where is a free group. Moreover, is a free group (and therefore is free) except possibly if is almost simple.
Here . Recall that, for a fixed prime , , is the largest normal -subgroup of and is the largest normal -subgroup of (i.e. of order prime to ). We always can assume that is connected. Note that, in that case, is also connected since the induced map is surjective.
In fact, in Theorem 5.1 we only need Aschbacher’s conjecture to hold for the -simple groups involved in . Here, a group is involved in if for some . Moreover, we only need the conjecture to hold for -rank (see Proposition 4.7 below). Recall that the -rank of a group is .
We recall now Aschbacher’s conjecture [Asc93].
Aschbacher’s Conjecture**.**
Let be a finite group such that , where is an elementary abelian -subgroup of rank and is the direct product of the -conjugates of a simple component of of order prime to . Then is simply connected.
Here denotes the generalized Fitting subgroup of (see Section 3 or [Asc00]). Aschbacher proved the conjecture for all simple groups except for Lie type groups with Lie rank 1 and the sporadic groups which are not Mathieu groups (see [Asc93, Theorem 3]).
Here below we list some immediate consequences of Theorem 5.1.
If is -solvable, , so is not almost simple. In fact, is contractible by Proposition 3.5 below. Let be the unique normal subgroup of containing such that .
Corollary 1.1**.**
Assume that Aschbacher’s conjecture holds. If , then is free. In particular, this holds for -solvable groups.
If we take an abelian simple group in the hypotheses of Aschbacher’s conjecture, then the conjecture holds for by [Qui78, Theorem 11.2]. Since there are no non-abelian simple groups involved in a solvable group G, Aschbacher’s conjecture does not need to be assumed for solvable groups.
Corollary 1.2**.**
If is solvable then is a free group.
By Feit-Thompson, if then there are no non-abelian -simple group. In consequence, Aschbacher’s conjecture does not need to be assumed and we get the following corollary.
Corollary 1.3**.**
There is an isomorphism , where is a free group. Moreover, is a free group (and therefore is free) except possibly if is almost simple.
In Section 4, we use Pulkus-Welker’s wedge decomposition [PW00] to restrict Aschbacher’s conjecture to the -rank case.
Proposition 4.7.
If Aschbacher’s conjecture holds for -rank , then it holds for any -rank . Moreover, if the conjecture holds in -rank for a -simple group then it holds in any -rank for .
In Section 6 we study freeness for some particular cases of almost simple groups. We do not need to assume Aschbacher’s conjecture for these cases. In the following theorem we collect the results of Section 6. We use the notations of the finite simple groups of [GLS98]. In general, by a simple group we will mean a non-abelian simple group.
Theorem 1.4**.**
Suppose that , with a simple group. Then is a free group in the following cases:
- (1)
, 2. (2)
* is disconnected,* 3. (3)
* is simply connected,* 4. (4)
* is simple of Lie type in characteristic and when has Lie rank ,* 5. (5)
* and has abelian Sylow -subgroups,* 6. (6)
* and (the alternating group),* 7. (7)
* is a Mathieu group,* 8. (8)
* or ,* 9. (9)
* and , , .*
From our base example (with ) and Theorem 5.1, one can easily construct an infinite number of examples of finite groups with non-free , by taking extensions of -groups whose -simple groups involved satisfy Aschbacher’s conjecture, by . However is the unique known example so far of a simple group with non-free fundamental group. We do not know whether is non-free for . It would be interesting to find new examples of simple groups (other than the alternating groups) with non-free. Besides the works of Aschbacher, Das, Ksontini and Shareshian mentioned above, we refer the reader to S. Smith’s book [Smi11, Section 9.3] for more details on the fundamental groups of Quillen complexes and applications to group theory, such as uniqueness proofs. Also a recent work of J. Grodal [Gro16] relates the fundamental group of the -subgroup complexes to modular representation theory of finite groups via the exact sequence
[TABLE]
(when is connected). Here denotes the transport category, whose objects are the non-trivial -subgroups of a (fixed) Sylow -subgroup , with . In [Gro16, Remark 2.2] it is shown that the geometric realization of is homotopy equivalent to the Borel construction , and the exact sequence follows from the fibration sequence . Recall that, by Brown’s ampleness theorem, the mod- cohomology of is isomorphic to the mod- cohomology of (see [Bro94, Smi11]). We hope that the results of this article can shed more light on the topology of these objects.
In this paper we study the posets of -subgroups topologically by means of their order complexes. We will say, for instance, that is contractible if its order complex is. Also, the homology groups and the fundamental group of the posets are those of their associated complexes. Note that this is not the convention that we adopted in our previous articles [MP18, Pit19]. In those papers, we handled the posets of -subgroups as finite topological spaces, with an intrinsic topology, where the notion of homotopy equivalence is strictly stronger than in the context of simplicial complexes. However in this article we adopt the more usual convention (as, for example, in [Qui78, AS93, Asc93, PW00, Sha04, Smi11]).
Acknowledgements. We would like to thank Volkmar Welker, John Shareshian and Jesper Grodal for helpful comments.
2. A non-free fundamental group
The fundamental group of the Quillen complex was first investigated by Aschbacher, who analyzed simple connectivity [Asc93]. K. Das studied simple connectivity of the -subgroups complexes of groups of Lie type (see [Das95, Das98, Das00]). In [Kso03, Kso04], Ksontini investigated the fundamental group of the Quillen complex of symmetric groups. Below we recall Ksontini’s results. These results will be used in Proposition 6.11.
Theorem 2.1** ([Kso03, Kso04]).**
Let and let be a prime.
- (1)
If is odd, then . In this case, is simply connected if and only if or . If , then is free unless and . If then and is free. 2. (2)
If , then is simply connected if and only if or . In other cases, is a free group by direct computation.
In [Sha04] Shareshian extended Ksontini’s results and showed that the fundamental group of is also free for .
Theorem 2.2** ([Sha04]).**
* is free when .*
In [Sha04] Shareshian gave the first example of a group whose -subgroup complex is not homotopy equivalent to a bouquet of spheres: he showed that there is torsion in the second homology group of . However its fundamental group is free. Surprisingly we found that the fundamental group of is not free. This is the first concrete known example of a Quillen complex with non-free fundamental group. In fact, is, so far, the unique known example of a simple group whose Quillen complex has non-free fundamental group.
To compute we used the Bouc poset of non-trivial -radical subgroups. Recall that is said to be -radical if . Denote by the poset of non-trival -radical subgroups of . In [Bou84, TW91] it is shown that is a homotopy equivalence at the level of complexes. In particular, .
We calculated using GAP (see [GAP]). We wrote two functions that can be used to compute the order complex of the Bouc poset of a given finite group at a given prime , and its fundamental group. The codes are shown in the Appendix.
We found that is a free product of the free group on generators and a non-free group on generators and relators whose abelianization is . It does not have torsion elements but it has commuting relations. Note that the integral homology of is free abelian (cf. [Sha04, p.306]). As a consequence of Theorem 5.1, one can construct an infinite number of examples of finite groups with non-free , by taking extensions of -groups whose -simple groups involved satisfy Aschbacher’s conjecture, by . As we mentioned in the introduction, it would be interesting to find other examples of simple groups with non-free fundamental group.
We were able to verify that is the smallest group with a Quillen complex with non-free . Note that, by Theorem 5.1, we only need to verify freeness in almost simple groups (note also that Aschbacher’s conjecture holds for groups of order less than the order of ). On the other hand, Theorem 1.4 allowed us to discard many potential counterexamples. The remaining almost simple groups which are smaller than were checked by computer calculations.
3. Notations and preliminary algebraic and topological results
We fix notations and recall some basic definitions. We refer the reader to Aschbacher’s book [Asc00] for more details on finite group theory.
For a finite group and a fixed prime number divining its order , denote by . The -rank of , denoted by , is the maximal integer such that there exists an elementary abelian -subgroup of of order . Recall that is the largest normal -subgroup of and is the largest normal -subgroup of . If , then and are, respectively, the centralizer and the normalizer of in . Here, .
, , , denote, respectively, the generalized Fitting subgroup, the center, the derived subgroup and the Frattini subgroup of . By definition, , where is the Fitting subgroup of and is the layer of , i.e. the subgroup generated by the components of (the subnormal quasisimple subgroups). Recall that , , and that if are distinct components of . By a simple group we will mean a non-abelian simple group. We write , with a simple group, for an almost simple group with .
Denote by the dihedral group of order , the alternating group on letters, the symmetric group on letters and the cyclic group of order . The symbol denotes a split extension of by . We write if and are isomorphic groups.
By the classification of the finite simple groups, a simple group is either an alternating group , a group of Lie type or one of the sporadic group. Throughout this paper, we will follow the notation of [GLS98] for the simple groups.
As we mentioned in the introduction, denotes the order complex of a finite poset . We denote by the geometric realization of . Any poset map (i.e. order preserving) induces a simplicial map and thus, a continuous map between the geometric realizations. If are poset maps with (i.e. for all ) or , then and are homotopic (see [Qui78]). We use the symbol to denote a homotopy equivalence between topological spaces (or posets).
If is a poset and , set . Analogously we define , , . Denote by (resp. ) the set of maximal (resp. minimal) elements of . If are sets, then denotes the complement of in .
The following result follows immediately from [BM12, Corollary 4.10]. We include here an alternative proof.
Proposition 3.1**.**
Let be a finite connected poset and let be a subposet such that is an anti-chain (i.e. , and are not comparable). If is simply connected, then is free.
Proof.
Since the inclusion is a cofibration and is simply connected, by van Kampen theorem there is an isomorphism induced by the quotient map. Since is an anti-chain, the space has the homotopy type of a wedge of suspensions. Therefore, is a free group. ∎
Recall that a topological space is -connected if is trivial for . By convention, is -connected if and only if it is non-empty. A map between topological spaces is an -equivalence if is an isomorphism for and an epimorphism for . A finite CW-complex of dimension is said to be -spherical if it is -connected. Equivalently, it has the homotopy type of a bouquet of spheres of dimension (see [Qui78, Section 8]).
Proposition 3.2** ([Bjo03, Theorem 2], see also [Bar11b]).**
Let be a map between posets. Assume that is -connected for all . Then is an -equivalence.
Denote by the subposet of consisting of elements such that the complex has dimension at most . Note that and . Recall that is a homotopy equivalence by [Qui78, Proposition 2.1].
Proposition 3.3**.**
The inclusions and are -equivalences. In particular, the inclusions and induce isomorphisms between the fundamental groups.
Proof.
We show that the inclusion is an -equivalence by using the previous proposition. Let . Note that . If , then and is contractible (see Proposition 3.5 below). In particular, it is -connected. Suppose . If is elementary abelian, then is the poset of proper subspaces of , which is a wedge of spheres of dimension by the classical Solomon-Tits result. If is not elementary abelian, , and , where is the Frattini subgroup of . Then, induces a homotopy between the identity map and the constant map inside , and therefore is contractible. This shows that is an -equivalence. A similar proof works for . ∎
Remark 3.4*.*
By Proposition 3.3, in order to study the fundamental group of the Quillen complex, we only need to deal with the subposet . In fact we will not use the posets or . Note that we could have deduced the isomorphism without need of Propositions 3.2 and 3.3: it follows from van Kampen theorem and the fact that for any , is a wedge of spheres of dimension greater than or equal to . We decided to include Proposition 3.3 for future references.
The following result was proved by Quillen [Qui78, Proposition 2.4].
Proposition 3.5**.**
If then and are contractible.
The converse of this proposition is Quillen’s conjecture [Qui78, Conjecture 2.9].
Remark 3.6*.*
For any subgroup , consider the subposet . Note that the inclusion is a strong deformation retract via .
Lemma 3.7**.**
Let and let . Then defined by is a strong deformation retract.
Proof.
The result is clear if is empty. If it is not empty, let be the map . Then by modular law, and . ∎
We will use the following lemma of [Asc93].
Lemma 3.8** ([Asc93, (6.9)]).**
Let and suppose is simply connected. If is connected for each subgroup of order , then is simply connected.
If and are posets, their join is the poset whose underlying set is the disjoint union and with the following order relation . We keep the given order in and , and set for all and . It is easy to see that and therefore (see [Qui78, Proposition 1.9]).
Proposition 3.9** ([Bar11, Lemma 6.2.4]).**
If and are finite non-empty posets, then is a free group of rank .
Proposition 3.10** ([Qui78, Proposition 2.6]).**
If , then . In particular, if , is connected and is free. Moreover, is simply connected if and only if is connected for some .
Recall that is disconnected if and only if has a strongly -embedded subgroup (see [Qui78, Proposition 5.2]). The groups with this property are classified and we will use this classification later.
Theorem 3.11** ([Asc93, (6.1)]).**
The poset is disconnected if and only if either and , or is one of the following groups:
- (1)
Simple of Lie type of Lie rank and characteristic , 2. (2)
* with ,* 3. (3)
, or with , 4. (4)
, , , or with , 5. (5)
* with .*
Remark 3.12*.*
The simple groups of Lie type and Lie rank are the groups , , and . In characteristic , these are , and and they are the unique simple groups with a strongly -embedded subgroup. There are no simple groups of -rank (see [GLS98]).
Remark 3.13*.*
Suppose is disconnected and let be a connected component. Let be the stabilizer of under the conjugation of on the connected components of . It can be shown that and that is a strongly -embedded subgroup of (see for example [Asc00, Section 46] and [Qui78, Section 5]). Moreover, since permutes transitively the connected components of , they have isomorphic fundamental groups. This allows us to define as (for any connected component ). Therefore the study of the fundamental group of the Quillen complexes can be restricted to the connected case.
4. Reduction to ,
In this section, we reduce the study of the fundamental group of to the case and . We assume that is connected.
The reduction is clear since . If , then is contractible by Proposition 3.5 and in particular simply connected. Therefore, we may assume .
The reduction relies on the wedge lemma of homotopy colimits. We will use Pulkus-Welker’s result [PW00, Theorem 1.1] but for instead of . Recall that by Proposition 3.3.
Note that and that is connected when is connected since the induced map is surjective. The following lemma is a slight variation of Pulkus-Welker’s result [PW00, Theorem 1.1]. Note that we have replaced the hypothesis of solvability of the normal -subgroup in [PW00, Theorem 1.1] by simple connectivity of for of -rank .
Lemma 4.1**.**
Let be a normal -subgroup of such that is simply connected for each elementary abelian -subgroup of -rank . Then
[TABLE]
In particular, for a suitable base point,
[TABLE]
Proof.
We essentially follow the proof of Pulkus-Welker [PW00, Theorem 1.1]. Let be a normal -subgroup of . Write and let be the map induced by taking quotients. Note that it is well defined and surjective. We will use [PW00, Corollary 2.4]. For this, we have to verify that the inclusions are homotopic to constant maps. Note that and .
By hypothesis and Remark 4.2 below we deduce that and are spherical of the corresponding dimension for each of -rank at most . For instance, if has -rank , then is [math]-spherical and is -spherical.
The result now follows from the fact that the inclusion of a sphere of dimension into a sphere of dimension is homotopic to a constant map, and and are spherical. ∎
Remark 4.2*.*
Let be an elementary abelian -group of -rank at least acting on a -group . Then is connected. To show this, assume otherwise and take a minimal counterexample . Thus, and by minimality. Therefore, and is one of the groups in the list of Theorem 3.11. But none of the groups in that list is elementary abelian of -rank at least . Therefore, is connected.
Lemma 4.3**.**
Let be a normal -subgroup of and let of -rank at most . Assume Aschbacher’s conjecture for -rank . Then, the fundamental group of is free if or and trivial if .
Proof.
We can suppose . We examine the possible ranks of . If , is a disjoint union of points while is a non-empty graph. Thus, their join is homotopic to a wedge of -spheres and -spheres.
If , is a connected non-empty graph by the above remark. Note that may be either empty, if is maximal, or discrete. Thus, their join is homotopic to a wedge of -spheres or -spheres.
It remains the case . Here, is empty. Suppose is not simply connected and take a minimal counterexample. We will show that the group satisfies the hypotheses of Aschbacher’s conjecture (and this leads to a contradiction). We may assume that , so acts faithfully on . Suppose that has a non-trivial proper normal subgroup which is also -invariant. We apply Lemma 4.1 with and as the normal -subgroup since is simply connected for of -rank by minimality of . Therefore,
[TABLE]
By minimality, is simply connected. Now we show that is also simply connected for every .
If , then and it is simply connected by induction.
If , then is a join of a connected space (see Remark 4.2) with a non-empty space. Thus, it is simply connected. Note that there are no maximal elements of order in since is a Sylow -subgroup of .
If , then is non-empty and is connected by [Asc93, Theorem 2]. Therefore, is simply connected.
Since all the spaces in the wedge of Equation (1) are simply connected, we deduce that is simply connected, which is a contradiction. Therefore, has no -invariant non-trivial proper normal subgroup. In particular, must be characteristicaly simple and therefore is a direct product of isomorphic simple groups on which acts transitively. Consequently, is in the hypotheses of Aschbacher’s conjecture and is simply connected. ∎
Remark 4.4*.*
Note that in the proof of Lemma 4.3, Aschbacher’s conjecture only needs to be assumed on the -simple groups involved in .
Now we apply these results to reduce to the case . Assume and Aschbacher’s conjecture for -rank . Then, by Lemmas 4.1 and 4.3 and Remark 4.2,
[TABLE]
All the groups are free by Lemma 4.3. Therefore, is free whenever is. Let .
Corollary 4.5**.**
Assume Aschbacher’s conjecture for -rank . Then there is an isomorphism , where is a free group. In particular, is free if is free.
Remark 4.6*.*
In Corollary 4.5, we only need Aschbacher’s conjecture to hold on the -simple groups involved in .
We finish this section with some remarks concerning Aschbacher’s conjecture. Recall the statement of the conjecture.
Aschbacher’s Conjecture**.**
Let be a finite group such that , where is an elementary abelian -subgroup of rank and is the direct product of the -conjugates of a simple component of of order prime to . Then is simply connected.
Aschbacher showed that the conjecture holds for a wide class of simple groups : the alternating groups, the groups of Lie type and Lie rank at least , the Mathieu sporadic groups and the groups with even (see [Asc93, Theorem 3]). The case of the Lyons sporadic group is proved in [AS92], and Segev dealt with many of the groups of Lie type and Lie rank in [Seg94].
In the following proposition we reduce the study of Aschbacher’s conjecture to the -rank case.
Proposition 4.7**.**
If Aschbacher’s conjecture holds for -rank , then it holds for any -rank . Moreover, if the conjecture holds in -rank for a -simple group then it holds in any -rank for .
Proof.
We use again Pulkus-Welker’s decomposition restricted to , as in Lemma 4.1.
Let be as in the statement of the conjecture, where . Assume that the conjecture holds for -rank . Set where the are the components of , all isomorphic to . By Lemma 4.1,
[TABLE]
Since , we may replace by and take without loss of generality. The poset is contractible, and in particular simply connected.
Note that is isomorphic to . Therefore, it is [math]-connected (resp. -connected) for (resp. ).
On the other hand, is -connected and [math]-connected for and , respectively (see Remark 4.2). Thus, is simply connected when or .
It remains to show that is simply connected if . We may assume , so acts faithfully on . If acts transitively on the set , we are in the hypotheses of Aschbacher’s conjecture for -rank and is simply connected. If the action of on is not transitive, then we may write where each is a -invariant direct product of some of the . Then, inductively and by hypothesis, and are simply connected for each of -rank . By Lemma 4.1, we conclude that is simply connected. ∎
5. Reduction to the almost simple case
In this section, we reduce the study of freeness of the fundamental group to the almost simple case. The main result of this section is the following.
Theorem 5.1**.**
Let be a finite group and a prime dividing . Assume that Aschbacher’s conjecture holds. Then there is an isomorphism , where is a free group. Moreover, is a free group (and therefore is free) except possibly if is almost simple.
Note that by Corollary 4.5, we only need to prove the moreover part. If the theorem does not hold, we can take a minimal counterexample , and then we can assume that satisfies the following conditions:
- (C1)
and is connected, 2. (C2)
(since otherwise is contractible by Proposition 3.5), 3. (C3)
is not a free group. In particular, is not simply connected and , 4. (C4)
(by minimality and Corollary 4.5), 5. (C5)
(by Proposition 3.10).
Remark 5.2*.*
From conditions (C2) and (C4) we deduce that , , and is the direct product of simple components of , each one of order divisible by . In particular , so .
Remark 5.3*.*
If satisfies the above conditions, by Remark 5.2, . Therefore has free fundamental group if , and it is simply connected for (see Proposition 3.10). If , is almost simple. We deal with the cases and separately (see Theorems 5.7 and 5.8 below).
In what follows, we do not need to assume Aschbacher’s conjecture. In [Asc93, Sections 7 & 8], Aschbacher characterized the groups for which some link , with of order , is disconnected. The following proposition deals with the case of connected links. Concretely, [Asc93, Theorem 1] asserts that if and the links are connected for all of order , then either is simply connected, is almost simple and and are not simply connected, or else has certain particular structure. We prove that in the later case, the fundamental group is free.
Proposition 5.4**.**
Suppose satisfies conditions (C1)…(C5). If the links are connected for all of order , then is almost simple and is not simply connected.
Proof.
We use [Asc93, Theorem 1]. By conditions (C1)…(C5), corresponds either to case (3) or case (4) of [Asc93, Theorem 1]. Case (4) implies that is almost simple and is not simply connected. If is in case (3) of [Asc93, Theorem 1], then is free (which is a contradiction by (C3)). This is deduced from the proof of [Asc93, (10.3)], since under these hypotheses and are homotopy equivalent, and is free by Proposition 3.10. ∎
Remark 5.5*.*
In [Asc93, Theorem 1], Aschbacher’s conjecture is required. However, since we are assuming , we do not need to assume the conjecture in the above proposition.
For the rest of this section we will assume that is not almost simple, so with . We deal with the cases and separately.
Remark 5.6*.*
Let be a simple group with a strongly -embedded subgroup, i.e. such that is disconnected. Then is a simple group of Lie type and Lie rank in characteristic and it is isomorphic to one of the simple groups or by Theorem 3.11. In any case, the Sylow -subgroups of have the trivial intersection property. That is, if (see [Sei82, Theorem 7]). Therefore, has connected components. If is a connected component of , for some and thus, is contractible by Proposition 3.5. In particular, the components of are simply connected.
Theorem 5.7**.**
Under conditions (C1)…(C5), if is a direct product of two simple groups, then , (the standard wreath product), with a simple group of Lie type and Lie rank in characteristic , and is a free group with generators.
Proof.
Note that is homotopy equivalent to , which is simply connected if and only if or is connected (see Proposition 3.10).
Assume is simply connected. Since is not simply connected, by Lemma 3.8 there exists some subgroup of order such that is disconnected. Since , by simple connectivity. By [Asc93, (10.5)] acts regularly on the set of components of and each has a strongly -embedded subgroup. In particular and . Since is simply connected, then is connected for some by Proposition 3.10, so does not have a strongly -embedded subgroup, which is a contradiction.
Now suppose is not simply connected. Then, is a free group by Proposition 3.10, and and are simple groups with strongly -embedded subgroups. We use [Asc93, (10.3)]. By the above hypotheses, corresponds to either case (2) or case (3) of [Asc93, (10.3)]. In case (3), as we mentioned in the proof of Proposition 5.4, and are homotopy equivalent (which contradicts the conditions on ). Therefore, is in case (2) of [Asc93, (10.3)], and for some of order . Then, for an involution , and is a group of Lie type and Lie rank in characteristic by Remark 5.6.
We prove now that is free. Let and . Let and let be the complement of in . If then , then . Therefore, consists of some minimal elements of the poset, and we have
[TABLE]
For each , by Lemma 3.7. Note that since . By Remark 5.6, has connected components and each component is simply connected. Since is contractible, then, by the non-connected version of van Kampen theorem (see [Bro06, Section 9.1]), , where is the free group of rank . Since for each , then recursively we have , where is the free group of rank .
By Remark 3.6, . Therefore is a free group on generators by Proposition 3.9.
Finally we compute . Let be the number of distinct involutions in . Therefore, , where is the number of involutions not contained in . Note that . If is an involution, then with and . The condition implies with and . Since , and , i.e. . Therefore, and .
In conclusion, is a free group with generators. ∎
Now we deal with the case .
Theorem 5.8**.**
Under conditions (C1)…(C5), if is a direct product of simple groups with , then is odd, , each has a strongly -embedded subgroup, is permuted regularly by some subgroup of order of , and is a free group.
Proof.
The hypotheses imply that is simply connected by Proposition 3.10. Then there exists some subgroup of order such that is disconnected by Lemma 3.8. We apply [Asc93, (10.5)]. The hypotheses imply that we are in case (5) of [Asc93, (10.5)]. Then permutes regularly the components and each has a strongly -embedded subgroup. In particular, is odd and for all .
Set and let . Then . If , then , so . In particular,
[TABLE]
is simply connected. Therefore, by Lemma 3.8, is simply connected, hence is also simply connected by Remark 3.6. Consider the complement . If then . Thus, where permutes regularly the components and . Since , we conclude that and , i.e. . Therefore, is an anti-chain and, by Proposition 3.1, is free. ∎
Proof of Theorem 5.1.
We may assume that , is connected and . By Corollary 4.5, we only need to prove that is free when and is not almost simple. We may assume conditions (C1)…(C5), and the proof now follows from Remark 5.3 and Theorems 5.7 and 5.8. ∎
6. Freeness in some almost simple cases
In this section we prove that is free when is an almost simple group with some extra hypothesis. We will use the structure of the outer automorphism group of a simple group. We refer the reader to sections 7 and 9 of [GL83] and Chapters 2 to 5 of [GLS98]. For the -rank of simple groups we will use the results of section 10 of [GL83] and in particular [GL83, (10-6)].
Consider a finite group such that , where is a simple group of order divisible by . We may suppose that , and is connected.
Theorem 6.1**.**
Let and be as above. Then is free if is disconnected or simply connected.
Proof.
We prove first that is free when is disconnected. In this case, has a strongly -embedded subgroup. We deal with each case of the list of Theorem 3.11.
- •
If , then is odd and by [GL83, (7-13)].
- •
If is a simple group of Lie type and Lie rank in characteristic , then the Sylow -subgroups of have the trivial intersection property, i.e. if and (see [Sei82, Theorem 7]). The proof is similar to the proofs of Theorems 5.7 and 5.8. Let and let . Since , consists of subgroups of order . By Remarks 3.6 and 5.6, has simply connected components. If , then by Lemma 3.7, and the Sylow -subgroups of intersect trivially, so has simply connected components. Then is free by van Kampen theorem.
- •
or since these groups are not simple.
- •
In the remaining cases, by [GL83, (10-6)] or by direct computation.
Now we prove that is free when is simply connected. Note that since otherwise , contradicting that is simple. By Lemma 3.8, we may assume that is disconnected for some of order . Therefore, we are dealing with one of the cases (1), (2), (3) or (4) of [Asc93, (10.5)]. We deal with each one of them.
- (1)
If is of Lie type and Lie rank in characteristic , then is disconnected, contradicting the hypothesis. 2. (2)
If , is even and , or , then is of Lie type and Lie rank at most . In any case, is not simply connected since it has the homotopy type of a wedge of spheres of dimension equal to the Lie rank of minus (see [Qui78, Theorem 3.1]). If , then . Hence, is free by Proposition 3.1 applied to . Note that is simply connected since by Remark 3.6. 3. (3)
If and with even, then has Lie rank , and thus is a wedge of -spheres, contradicting the hypothesis (see [Qui78, Theorem 3.1]). 4. (4)
If , and , then by [GL83, (9-3)]. Therefore, is free by Proposition 3.1 applied to .
∎
Corollary 6.2**.**
If is a Lie type group in characteristic and when has Lie rank , then is free.
Proof.
Since is homotopy equivalent to the building associated to (see [Qui78, Theorem 3.1]), it is a bouquet of spheres of dimension , where is the Lie rank of . If , is in the hypotheses of Theorem 6.1. If , since . In either case, is a free group. ∎
Remark 6.3*.*
Corollary 6.2 does not give information in the case that has Lie rank and . By [Qui78, Theorem 3.1], is a free group but if then it may happen that . For example, take and . Note that . We have computed for all possible groups , with , and they turned out to be free. If , is simply connected, while is a non-trivial free group.
Recall the classification of J. Walter of simple groups with abelian Sylow -subgroup [Wal69].
Theorem 6.4** ([Wal69]).**
Let be a a simple group with abelian Sylow -subgroup . Then is isomorphic to one of the following groups:
- (1)
, (and is elementary abelian of order ), 2. (2)
, (and is elementary abelian of order ), 3. (3)
, odd (and is elementary abelian of order ), 4. (4)
* (and is elementary abelian of order ).*
In the proof of the next result, we will work with Bouc poset of non-trivial -radical subgroups instead of (see Section 2).
Theorem 6.5**.**
Suppose is almost simple, and has abelian Sylow -subgroups. Then is free.
Proof.
Let . By Walter’s Theorem 6.4, is one of the groups (1)…(4).
The case (2) follows from the disconnected case of Theorem 6.1.
In case (4), and is free on generators by computer calculation.
In case (3), , with odd. Then has odd order and . It suffices to prove that has dimension . Note that . Let . By [GLS98, Theorem 6.5.5], the normalizers of the non-trivial -subgroups have the following forms: for involutions and for four-subgroups. If is an involution, . If is a four-subgroup of then since . For a Sylow -subgroup of we have that . Therefore, the poset contains the subgroups generated by one involution and the Sylow -subgroups. In consequence, is a -dimensional simplicial complex (and therefore it has free fundamental group).
In case (1), , then is odd and it is not a square. Therefore, has odd order and thus we may assume . In any case, by [GLS98, Theorem 4.10.5(b)]. ∎
Now we compute the fundamental group of for some particular sporadic groups . Note that if .
Example 6.6**.**
By computer calculations, is free for or . Note that and . Note that, if is odd, for or (see [GL83, (10-6)]).
Example 6.7**.**
If or and , then is free. By [Kot97, Proposition 3.1.4] and [UY02, Section 6.1], there are only two conjugacy classes of non-trivial -radical subgroups of . Therefore, has dimension . For , , so is free.
Example 6.8**.**
If and , then is free. By computer calculations, if is a Sylow -subgroup of , there exist three subgroups of (up to conjugacy) which are -radical subgroups of : , and . Their orders are , , . Moreover, and . Then for any such that , and therefore is -dimensional. For , , so is free.
Proposition 6.9**.**
Assume is a Mathieu sporadic group. If is odd and , then has free fundamental group. If , is simply connected except for , in which case is a non-trivial free group.
Proof.
Let be a one of the Mathieu groups , , , or . In all cases, if is odd. Therefore, we may assume that . Recall that for , and , and for and .
Note that . For or we checked with GAP that is simply connected.
If , then . Let . Recall that by Remark 3.6 and therefore it is simply connected. Any is generated by an involution acting by outer automorphism on . By [GLS98, Table 5.3c], its centralizer in has a non-trivial normal -subgroup. That is, is contractible by Proposition 3.5. Then for any , is simply connected by van Kampen theorem and, recursively, is simply connected. A similar reasoning shows that is simply connected for (see [GLS98, Table 5.3b]).
By [Smi11, p.295] is homotopy equivalent to its -local geometry, which is simply connected.
It remains to determine the fundamental group of . For this we use the classification of the maximal subgroups of and . First note that is a maximal subgroup of of odd index . In particular, any elementary abelian -subgroup of is contained in some conjugate of . Therefore, is a cover of by subcomplexes. We have computed the intersections between different conjugates of with GAP. All the intersections , with , form a subgroup of of order . All the maximal subgroups of have order less than except for the maximal subgroup (and all its conjugates) which have order exactly . Thus, and, by van Kampen theorem, each element of is simply connected. The triple intersections of different conjugates of are all isomorphic to , and the quadruple intersections of different conjugates of are all isomorphic to . This shows that double and triple intersections of elements of are connected. In consequence, by van Kampen theorem, is simply connected. ∎
We investigate now the fundamental group of the Quillen complex of alternating groups at . We use Ksontini’s results on (see Section 2).
Remark 6.10*.*
By Theorem 6.1 the poset is disconnected if and only if , since, for , the unique isomorphism of an alternating group with a group of the list is (see [GL83, (3-3)]).
Proposition 6.11**.**
Let . The fundamental group of is simply connected for and . For , each component of is simply connected, for it is free of rank , and for it is free of rank .
Proof.
If , then and is contractible. If , , which has trivial intersections of Sylow -subgroups by [Sei82, Theorem 7]. Therefore, its connected components are simply connected. The cases can be obtained directly by computer calculations.
We prove the case . We proceed similarly as before. Take . By Remark 3.6, , and let be its complement in . Note that consists of the subgroups of order of generated by involutions which can be written with an odd number of disjoint transpositions. Note that for any , . By Ksontini’s Theorem 2.1, . Therefore, by van Kampen theorem, in order to prove that is trivial, we only need to show that the intersections are simply connected for all .
We appeal now to the characterization of the centralizers of involutions in to show that is simply connected if . Let be an involution acting as the product of disjoint transpositions and with fixed points. By [GLS98, Proposition 5.2.8], , where has index , and . Here, the wreath product is taken with respect to the natural permutation of on the set . Moreover, where is the subgroup . If then . If , then , and . In any case, since and . Therefore, if , and is contractible. In particular it is simply connected. If , , and therefore is simply connected by Theorem 2.1 (note that since ). ∎
Combining Proposition 6.11 with Theorem 2.1 we deduce the following corollary.
Corollary 6.12**.**
If , then is a free group.
Proof.
If , then or . In any case, is free by Proposition 6.11 and Theorem 2.1. For , and . If , then or . In either case, is free by the above results or by computer calculations. ∎
Appendix
In order to compute fundamental groups of Quillen complexes, we used GAP (see [GAP]), together with the GAP package HAP ([HAP]). We wrote two functions that compute the order complex of the Bouc poset of a given finite group at a given prime , and its fundamental group.
For example, to compute the fundamental group of we execute the following code in GAP.
gap> G:=AlternatingGroup(10);;
gap> BpG:=allRadicalSubgroups(G,3);;
gap> pi1BpG:=pi1(BpG,G,3);;
gap> pi1BpG;
Here below we exhibit the codes of these functions.
allRadicalSubgroups:=function(G,p)
**local** BpS, BpG, H, N, a, b, i, g, add, numCC,
normalizers, reducedList, subgroups, transversals;
*# We calculate first the non-trivial p-radical subgroups*
*# contained in a fixed Sylow p-subgroup.*
S:=SylowSubgroup(G,p);
subgroups:=Set(Filtered(SubgroupsSolvableGroup(S), l-> Order(l) > 1));
BpS:=Filtered(subgroups, l-> Size(l) = Size(PCore(Normalizer(G,l),p)));
*# The list reducedList will contain exactly one representative for*
*# each conjugacy class of each element in BpS.*
reducedList:=[];
**for** a **in** BpS **do**
add:=**true**;
i:=1;
**while** add **and** i <= Size(reducedList) **do**
b:=reducedList[i];
**if** IsConjugate(G,a,b) **then**
add:=**false**;
**fi**;
i:=i+1;
**od**;
**if** add **then**
Add(reducedList,a);
**fi**;
**od**;
*# numCC is the number of conjugacy classes of non-trivial*
*# p-radical subgroups.*
numCC:=Size(reducedList);
*# For each representative of non-trivial p-radical subgroups,*
*# we calculate its normalizer and a transversal of this.*
normalizers:=List(reducedList, H-> Normalizer(G,H));
transversals:=List(normalizers, N->
List(RightTransversal(G,N),i->CanonicalRightCosetElement(N,i)));
*# Now we compute Bp(G).*
BpG:=[];
**for** i **in** [1..numCC] **do**
**for** g **in** transversals[i] **do**
H:=reducedList[i]^g;
Add(BpG, H);
**od**;
**od**;
**return** BpG;
end;;
allMaximalChains:=function(G, S, BpS, elementList, orders)
*# This function returns all maximal chains of non-trivial*
*# p-radical subgroups of G.*
**local** g, i, j, o, t, x, H, Q, N, R, T, lastPoint, partialChain,
properSubgroups, radSubgroupsConj;
N:=Normalizer(G,S);
T:=List(RightTransversal(G,N), x-> CanonicalRightCosetElement(N,x));
R:=[];
o:=PositionSorted(orders, Size(S));
**for** g **in** T **do**
Q:=[[ [o,PositionSorted(elementList[o],S^g)] ]];
t:=1;
radSubgroupsConj:=List(BpS, x-> x^g);
**while** t<= Size(Q) **do**
partialChain:=Q[t];
lastPoint:=partialChain[Size(partialChain)];
i:=lastPoint[1];
j:=lastPoint[2];
H:=elementList[i][j];
properSubgroups:=Filtered(radSubgroupsConj, x->
Size(x) < Size(H) **and** IsSubset(H, GeneratorsOfGroup(x)));
properSubgroups:=List(properSubgroups, x->
[PositionSorted(orders, Size(x)),
PositionSorted(elementList[PositionSorted(orders,Size(x))],x)]);
**for** x **in** properSubgroups **do**
Add(Q, Concatenation(partialChain,[x]));
**od**;
t:=t+1;
**od**;
R:=Concatenation(R,Q);
**od**;
**return** R;
end;;
pi1:=function(BpG,G,p)
**local** i, o, x, R, S, BpS, elementList, orders;
S:=SylowSubgroup(G,p);
BpS:=Filtered(BpG, x-> IsSubset(S, GeneratorsOfGroup(x)));
orders:=Unique(List(BpS, Size));
Sort(orders);
elementList:=List(orders, x-> []);
**for** o **in** orders **do**
i:=PositionSorted(orders, o);
elementList[i]:=Set(Filtered(BpG, x->Size(x) = o));
**od**;
*# We calculate all maximal chains of BpG.*
R:=allMaximalChains(G, S, BpS, elementList, orders);
*# Finally, we compute the fundamental group of the*
*# simplicial complex generated by these maximal chains.*
**return** FundamentalGroup(SimplicialComplex(R));
end;;
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[Asc 93] M. Aschbacher. Simple connectivity of p 𝑝 p -group complexes. Israel J. Math. 82 (1993), no. 1-3, 1-43.
- 2[Asc 00] M. Aschbacher. Finite group theory, second ed. Cambridge Studies in Advanced Mathematics, vol. 10. Cambridge University Press, Cambridge, 2000, xii+304.
- 3[AS 92] M. Aschbacher, Y. Segev. The uniqueness of groups of Lyons type, J. Am. Math. Soc. 5 (1992), 75-98.
- 4[AS 93] M. Aschbacher, S. D. Smith. On Quillen’s conjecture for the p 𝑝 p -groups complex. Ann. of Math. (2) 137 (1993), no. 3, 473-529.
- 5[Bar 11] J. Barmak. Algebraic Topology of Finite Topological Spaces and Applications. Lecture Notes in Math. 2032, Springer, 2011, xviii+170.
- 6[Bar 11b] J. Barmak. On Quillen’s Theorem A for posets. J. Combin. Theory Ser. A 118(2011) pp. 2445-2453.
- 7[BM 12] J. Barmak, E.G. Minian. G-colorings of posets, coverings and presentations of the fundamental group. Preprint available in Ar Xiv. https://arxiv.org/abs/1212.6442
- 8[Bjo 03] A. Björner. Nerves, fibers and homotopy groups. J. Combin. Theory, Ser. A, 102 (2003), 88–93.
