ebook img

Pseudofinite groups as fixed points in simple groups of finite Morley rank PDF

0.21 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Pseudofinite groups as fixed points in simple groups of finite Morley rank

PSEUDOFINITE GROUPS AS FIXED POINTS IN SIMPLE GROUPS OF FINITE MORLEY RANK 2 1 PINAR UG˘URLU∗,† 0 2 n a J Abstract. Weprovethatifthegroupoffixedpointsofageneric automorphism of a simple group of finite Morley rank is pseu- 0 3 dofinite, then this group is an extension of a (twisted) Chevalley groupoverapseudofinitefield. Onthewaytoobtainthisresult,we ] classify non-abelian definably simple pseudofinite groups of finite R centralizer dimension (using the ideas of John S. Wilson [28]). G . h t a Introduction m A group of finite Morley rank is a group equipped with a rank func- [ tion from the set of non-empty definable subsets to the non-negative 1 v integers. This rank function imitates the Zariski dimension in alge- 7 braic geometry (for details see [5]). Algebraic groups over algebraically 1 closed fields are examples of groups of finite Morley rank in which 3 6 case the Morley rank coincides with the Zariski dimension. Actually, . 1 the only known infinite simple groups of finite Morley rank are alge- 0 braic groups over algebraically closed fields. The Algebraicity Conjec- 2 ture (stated independently by Gregory Cherlin and Boris Zilber in the 1 : 1970’s) says that any infinite simple group of finite Morley rank is an v i algebraic group over an algebraically closed field. Although this con- X jecture is still open, important results have been obtained by adapting r a and generalizing ideas from the classification of finite simple groups. This approach was suggested by Alexandre Borovik. For a detailed in- formation about the Borovik program, we refer the reader to [1]. The current status of the conjecture can be summarized as follows: In the theory of groups of finite Morley rank, the Sylow 2-subgroups are known to be conjugate and each of them is a finite extension of the central product U ∗ T where U is a definable connected 2-group of bounded exponent and T is a divisible abelian 2-group. Therefore, ∗Supported by TU¨BI˙TAK. †Researchpartiallysupportedbythe MarieCurieEarlyStageTrainingNetwork MATHLOGAPS (MEST-CT-2004- 504029). 1 2 P. UG˘URLU depending on the structure of the connected components of Sylow 2- subgroups, the Algebraicity Conjecture breaks up into four cases. (a) Even type: U 6= 1 and T = 1 (identified with Chevalley groups over algebraically closed fields of characteristic 2). (b) Odd type: U = 1 and T 6= 1 (structural results are obtained on potential non-algebraic odd type groups). (c) Mixed type: U 6= 1 and T 6= 1 (no such groups exist). (d) Degenerate type: U = 1 and T = 1 (difficult case). A new approach to the Algebraicity Conjecture originates from re- sults and ideas of Ehud Hrushovski in [11], where some classes of struc- tures, including simple groups of finite Morley rank, with generic au- tomorphisms are considered. Hrushovski proved that the set of fixed points of a generic automorphism is a PAC structure with a definable measure (fordetails, see [11]). In theparticular caseof simple groupsof finite Morley rank, the ultimate aim is to prove that the group of fixed points is pseudofinite, that is, an infinite model of the theory of finite groups. However, unlike pseudofinite fields (which are characterized by James Ax in [2]), it is not known how to characterize pseudofinite groups. As it is mentioned by Hrushovski in[11], theAlgebraicity Conjecture implies the following conjecture. Principal Conjecture. Let G be an infinite simple group of finite Morley rank with a generic automorphism α. Then, the group of fixed points of α is pseudofinite. This paper is a first step to construct a bridge between Algebraicity Conjecture and Principle Conjecture from the other direction. More precisely, we aim to prove that the Principal Conjecture implies the Al- gebraicity Conjecture. This implication can be stated in the following form. Intermediate Conjecture. Let G be an infinite simple group of finite Morley rank with a generic automorphism α. Assume that the group of fixed points of α is pseudofinite. Then, G is isomorphic to a Chevalley group over an algebraically closed field. In this paper, we do not use the full strength of the genericity as- sumption on the automorphism α. We can state the main result as follows. Theorem 3.1. Let G be an infinite simple group of finite Morley rank and α be an automorphism of G such that the definable hull of C (α) G is G. If C (α) is pseudofinite, then there is a definable (in C (α)) G G PSEUDOFINITE GROUPS IN SIMPLE GROUPS OF FMR 3 normal subgroup S of C (α) such that S is isomorphic to a (twisted) G Chevalley group over a pseudofinite field and C (α) embeds in Aut(S). G Here, C (α) denotes the group of fixed points of α in G. Note that G the assumption on the definable hull of C (α) is satisfied by generic G automorphisms of groups of finite Morley rank. Moreover, we will observe in the last section that under the assumptions of this theorem, degenerate types groups can not exist. The first result we obtain on the way to prove Theorem 3.1 is the classification of non-abelian definably simple pseudofinite groups of finite centralizer dimension. For this classification, we first observe that a sizeable part of Wilson’s classification proof for simple pseudofi- nite groups in [28] works for non-abelian definably simple pseudofinite groups. Then, with the help of the finite centralizer dimension prop- erty, we show that such groups are elementarily equivalent to (twisted) Chevalley groups over pseudofinite fields. Moreover, thanks to the results obtained by Mark Ryten [24], we can replace the elementary equivalence by an isomorphism and we obtain the following result. Proposition 2.11. Every non-abelian definably simple pseudofinite group of finite centralizer dimension is isomorphic to a (twisted) Chevalley group over a pseudofinite field. Note that in some cases a non-abelian definably simple group G is automatically simple. For example, this holds if the theory of G is supersimple [27, Proposition 5.4.10]. However, in our case we do not know whether the theory of a pseudofinite group of finite centralizer dimension is supersimple or not. The structure of this paper can be outlined as follows. In the first section, we give necessary background information about basic model theoretic concepts, ultraproducts, pseudofinite groups, groups of finite Morley rank, and we explain our terminology and no- tation. In the second section, firstly, we analyze the structure of abelian definably simple pseudofinite groups. Then, we classify non-abelian definably simple pseudofinite groups of finite centralizer dimension. In the third section, we prove the main result (Theorem 3.1). 1. Preliminaries This section covers some background material which will be neces- sary throughout this paper. We assume that the reader is familiar with the basic notions in model theory such as structure, language, formula and theory. 4 P. UG˘URLU Two structures M and N, in a common language L, are elementarily equivalent if they satisfy the same L-sentences and we write M ≡ N. A theory is complete if all of its models are elementarily equivalent. A definable set in a structure M is a subset X ⊆ Mn (M denotes the underlying set of the structure M) which is the set of realizations of a first order formula in the language of the structure M. We emphasize that, throughout this paper, we stay within the bor- ders of first order logic and when we say definable, we mean definable possibly with parameters. We consider groups (resp. fields) as struc- tures in the pure group (resp. field) language. 1.1. Ultraproducts and pseudofinite groups. We give some back- ground information about ultraproducts and pseudofinite groups and list some related facts. Proofs of these facts can be found in the in- dicated references. We refer the reader to [4] and [7] for a detailed information about ultraproducts. An ultrafilter U on a non-empty set I is a proper subset of the power set of I such that U is closed under finite intersections and taking supersets and for any subset A of I, we have A ∈ U if and only if I\A ∈/ U . For any i ∈ I, the set {X ⊆ I | i ∈ X} forms an ultrafilter which is called the principal ultrafilter generated by i and any ultrafilter containing a finite set turns out to be a principal ultrafilter. Note that the existence of non-principal ultrafilters is guaranteed by Zorn’s lemma. Now let X denote the cartesian product of non-empty struc- i∈I i tures in a common language L and let U be an ultrafilter on I. Con- Q sidering the elements of X as functions from I to X , define i∈I i i∈I i a relation on X as follows. i∈I i Q S x ∼Q y if and only if {i ∈ I | x(i) = y(i)} ∈ U. U It is routine to check that ∼ is an equivalence relation. The quo- U tient of the cartesian product with respect to this equivalence relation is called the ultraproduct and we denote it by X /U. When all i∈I i structures are the same, then this ultraproduct is called ultrapower. Q We denote the equivalence class of x ∈ X with respect to ∼ by i∈I i U [x] . An ultraproduct of L-structures has a natural (coordinatewise) U Q L-structure. TheimportanceoftheultraproductconstructionisexpressedinL o´s’s Theorem [16] which states that a first order formula is satisfied in the ultraproduct if and only if it is satisfied in the structures indexed by a set belonging to the ultrafilter. In particular, first order properties PSEUDOFINITE GROUPS IN SIMPLE GROUPS OF FMR 5 of the ultraproduct are determined by the first order properties of the structuresintheultraproducttogetherwiththechoiceoftheultrafilter. It can be observed that if U is an ultrafilter on a set I which is a dis- joint union of finitely many subsets I ,...,I , then exactly one of I is 1 m j inU. Moreover, whenever J ∈ U then, the set U = {X ∩J | X ∈ U} J ∼ forms an ultrafilter on J and we have X /U = X /U . In i∈I i j∈J j J particular, ultraproducts over principal ultrafilters are isomorphic to Q Q one of the structures in the cartesian product. Therefore, throughout this paper, the ultrafilters in concern will always be non-principal. Remark 1.1. Throughout the text, we will use the well-known prop- erties mentioned in the previous paragraph. Moreover, whenever we ∼ have X /U = X /U , we will abuse the language, and keep i∈I i j∈J j J writing U and I for the ultrafilter and the index set. Q Q Ulrich Felgner introduced in [10] pseudofinite groups as infinite mod- els of the theory of finite groups in accordance with Ax’s characteri- zation of pseudofinite fields [2]. By a suitable choice of an ultrafilter, it can be shown that any pseudofinite group is elementarily equivalent to a non-principal ultraproduct of finite groups (see [29]). As we will see in Section 2, the additive group of rational numbers, (Q,+), is a pseudofinite group. However, the additive group of integers, (Z,+), is not a pseudofinite group, because the first order statement the map x 7→ x+x is one-to-one if and only if it is onto does not hold in (Z,+) while it holds in every finite group. Throughout this paper, a Chevalley group over an arbitrary field K will be denoted by X(K) where X stands for one of the sym- bols A (n > 1),B (n > 2),C (n > 3),D (n > 4),E ,E ,E ,F ,G . n n n n 6 7 8 4 2 These symbols come from the classification of finite dimensional com- plex simple Lie algebras. Chevalley constructed Chevalley groups over arbitrary fields associated to these symbols. We will not go into de- tails about the construction of neither these groups nor the twisted versions of these groups which can be constructed when the field K has some additional properties (see [25]). We denote twisted Cheval- ley groups by X(K) as well, where X denotes one of the symbols 2A (n > 2),2D (n > 4),3D ,2E ,2B ,2F ,2G . It is known that n n 4 6 2 4 2 (twisted) Chevalley groups over arbitrary fields are simple as abstract groups except for 2A (F ),2B (F ),2F (F ) and 2G (F ). We refer the 2 4 2 2 4 2 2 3 reader to [6] and [25] for details about these groups. The following result of Franc¸oise Point shows that any infinite Chevalley group over an ultraproduct of finite fields is an example of a pseudofinite group. 6 P. UG˘URLU Fact 1.2 (Point [22]). Let {X(F ) | i ∈ I} be a family of (twisted) i Chevalley groups of the same type X over finite or pseudofinite fields, and let U be a non-principal ultrafilter on the set I. Then ∼ X(F )/U = X( F /U). i i i∈I i∈I Y Y If Fact 1.2 is combined with Keisler-Shelah’s Ultrapower Theorem [13] then the following result can be obtained. Fact 1.3. (Wilson [28]) Any group G which is elementarily equivalent to a (twisted) Chevalley group over a pseudofinite field is pseudofinite. 1.2. Groups of finite Morley rank. We list some basic properties of groups of finite Morley rank which will be used in the sequel without references. The proofs can be found in [5] as well as in the indicated articles. Fact 1.4 (Macintyre [17]). A group of finite Morley rank has the de- scending chain condition on definable subgroups, that is, every descend- ing chain of definable subgroups stabilizes after finitely many steps. The descending chain condition on definable subgroups is a strong property which allows one to define the notion of definable hull. For any subset X of a group of finite Morley rank G, there is a smallest definable subgroup of G containing X, which is called the definable hull of X and denoted by d(X). The following fact is a corollary of a result by John T. Baldwin and Jan Saxl [3]. Fact 1.5 ([1], Corollary 2.9). For any subset X of a finite Morley rank group G, the centralizerC (X) is a definablesubgroup. Moreover, there G is an integer n such that for any Y ⊆ G there is Y ⊆ Y with |Y | 6 n 0 0 and C (Y) = C (Y ). G G 0 A group with the property as in the moreover part of Fact 1.5 is said to have finite centralizer dimension. More precisely, for any integer k > 0, a group has centralizer dimension k if it has a proper descending centralizer chain of length k and has no such chain of length greater than k. By a proper descending centralizer chain of length k we mean the chain of the form G = C (x ) > C (x ) > C (x ,x ) > ... > C (x ,...,x ) = Z(G). G 0 G 1 G 1 2 G 1 k It is well-known that the class of groups with finite centralizer dimen- sion is closed under taking subgroups and finite direct products [20]. Moreover, for any integer k > 0, having centralizer dimension k is a first order property in the language of groups (see [8] and [15]). PSEUDOFINITE GROUPS IN SIMPLE GROUPS OF FMR 7 2. Definably Simple Pseudofinite Groups The main result of this section is the characterization of non-abelian definably simple pseudofinite groups of finite centralizer dimension as (twisted) Chevalley groups over pseudofinite fields. Firstly, we state the theorem of Wilson about the classification of simple pseudofinite groups and point out some results in the literature strengthening Wil- son’s theorem. Then, we work on definably simple pseudofinite groups andafter analyzing thestructure of theabelianones, weconcentrate on the non-abelian case. We observe that the first part of Wilson’s clas- sification proof works for this case. Then, using the finite centralizer dimension property we obtain our result. There is a weaker version of simplicity of a group which arises in model theory. A group is called definably simple if it has no non-trivial definable proper normal subgroups. In the class of non-abelian groups of finite Morley rank, which includes non-abelian algebraic groups over algebraically closed fields, definably simple groups coincide with the simple ones [23] (note that Wagner’s result mentioned in the intro- duction is much more general). However, in general, definably simple groups need not be simple. Wilson proved that no ultraproduct of al- ternatinggroupsissimple(unless itisfinite)while itisdefinably simple (see [29]). Moreover, while no infinite abelian group is simple, some of them are definably simple (see Fact 2.2). Therefore, the distinction between the notionsof definably simple andsimple becomes important. Felgner conjectured in [10] that simple pseudofinite groups are iso- morphic to (twisted) Chevalley groups over pseudofinite fields. Al- though Felgner obtained important results, it was Wilson who classi- fied the simple pseudofinite groups, however, only up to elementary equivalence. Fact 2.1 (Wilson[28]). Every simple pseudofinite group is elementarily equivalent to a (twisted) Chevalley group over a pseudofinite field. As it is pointed out by Wilson in [28], the elementary equivalence can be replaced by an isomorphism for the untwisted case in view of results obtained by Simon Thomas in his dissertation [26]. Moreover, M.Ryten’srecent results inhisthesis [24]ensure thatWilson’s theorem can be strengthened for both Chevalley and twisted Chevalley groups and hence Felgner’s conjecture is true. Now, weconcentrateondefinablysimplepseudofinitegroupsstarting from the abelian ones. The following fact is a folklore. Fact 2.2. Let A be an infinite abelian group. The following statements are equivalent. 8 P. UG˘URLU (a) A is definably simple. (b) A is torsion-free divisible. (c) A ≡ p∈I Cp/U ≡ (Q,+) where U is a non-principal ultrafilter on the set I of all prime numbers and C is the cyclic group of p Q order p. Proof. Assume that (a) holds. Then, any non-trivial element of finite order generates a non-trivial proper definable normal subgroup of A, so A is torsion-free. Moreover, if A is not divisible, then nA is a proper definable normal subgroup of A for some integer n > 2 which is not possible by our assumption. Hence we get (b). Assume that (b) holds. Since the theory of divisible abelian torsion-free groups is complete and all groups in part (c) are models of this theory, we get the elementary equivalences in (c). As C /U is a definably simple group we have p∈I p the implication (c) ⇒ (a). (cid:3) Q Theclassificationofnon-abeliandefinablysimplepseudofinitegroups will be obtained in two steps. Firstly, we show that every non-abelian definably simple pseudofinite group is elementarily equivalent to an ultraproduct of non-abelian finite simple groups. This result follows from Wilson’s proof of Fact 2.1. However, for a self-contained proof, we include here the results obtained by Wilson and we observe that his arguments work in the case of non-abelian definably simple pseudofi- nite groups. In the second step, we proceed differently by using our assumption on centralizer chains. Definition 2.3. Let σ denote the following first order sentence which was defined by Felgner in [10]: ∀x∀y (x 6= 1∧C (x,y) 6= 1) → (C (x,y)C (C (x,y)))g = 1 . G G G G " # g∈G \ Above, G stands for an arbitrary group, C (x,y) denotes the central- G izer of {x,y} in G and for any subset X ⊆ G, we have Xg = gXg−1. Throughout this paper, Soc(G) denotes the subgroup generated by minimal normal subgroups of a group G, the so-called socle of G. Now, we recall some facts from [28] which will be needed in the sequel. Fact 2.4 (Wilson [28]). There is an integer k such that each element of each finite non-abelian simple group G is a product of k commutators. Remark 2.5. A stronger version of this result is known as Ore’s con- jecture which states that every element of a finite non-abelian simple PSEUDOFINITE GROUPS IN SIMPLE GROUPS OF FMR 9 group is a commutator [21]. This old conjecture was followed by a stronger conjecture, which is attributed to John Thompson, stating that for every finite non-abelian simple group G, there exists a conju- gacy class C such that G = CC. We will refer to Thompson’s con- jecture later in the text. For a detailed information about the status of these conjectures see the survey article [12]. The proof of Ore’s conjecture has been recently completed (see [19]). Fact 2.6 (Wilson [28]). Let σ be the sentence from Definition 2.3. (a) If G is a non-abelian simple group, then G |= σ, that is, σ holds in G. (b) If G is finite and G |= σ, then Soc(G) is a non-abelian simple group. Fact 2.7 (Wilson [28]). Let G be a finite group with a non-abelian simple socle. If G is not simple then G′ 6= G. Moreover, if every element of Soc(G) is a product of k commutators, then every element of G′ is a product of k +3 commutators. Fact 2.8 (Wilson[28]). Every simple pseudofinite group is elementarily equivalent to an ultraproduct of finite simple groups. In the following lemma, we observe that Wilson’s proofs for Fact 2.6(a) and Fact 2.8 work for non-abelian definably simple pseu- dofinite groups. Lemma 2.9. If G is a non-abelian definably simple pseudofinite group and σ is the sentence from Definition 2.3, then the following statements hold. (a) G |= σ. (b) G ≡ G /U where each G is a non-abelian finite simple i∈I i i group and U is a non-principal ultrafilter on a set I. Q Proof. (a) Assume that σ does not hold in G. Then, there are x,y ∈ G such that x 6= 1 and C (x,y) 6= 1 and G N := (C (x,y)C (C (x,y)))g 6= 1. G G G g∈G \ Clearly, N is a definable normal subgroup of G. Since G is definably simple and N is non-trivial by our assumption, we get N = G. As a result we have, C (x,y)C (C (x,y)) = G. Therefore, C (x,y) is G G G G normalized by G. Since C (x,y) is a non-trivial definable subgroup of G G, we have C (x,y) = G. This is a contradiction, since G can not have G central elements as a non-abelian definably simple group. 10 P. UG˘URLU (b) As a non-abelian pseudofinite group, G is elementarily equivalent to G /U where each G is a non-abelian finite group and U is a i∈I i i non-principal ultrafilter on a set I. Moreover, since G |= σ by part (a), Q we may assume that G |= σ for all i ∈ I by L o´s’s Theorem and i Remark 1.1. It follows by Fact 2.6(b) that Soc(G ) is a non-abelian i simple group for each i ∈ I. Now, let ϕ be a formula in the language i of groups defining the set of products of k + 3 commutators where k is the integer given by Fact 2.4. By Fact 2.7, for any non-simple G , i the formula ϕ defines a proper normal subgroup. Therefore, if G is i i not simple for almost all i ∈ I, then G has a proper definable normal subgroup by L o´s’s Theorem. This is not possible as G is definably simple and hence G is a non-abelian finite simple group for almost i all i ∈ I. Again, by referring to Remark 1.1, we can conclude that G ≡ G /U where G is a non-abelian finite simple group for all i∈I i i i ∈ I. (cid:3) Q Corollary 2.10. Every definably simple pseudofinite group is elemen- tarily equivalent to an ultraproduct of finite simple groups. Proof. Follows from Lemma 2.9(b) and Fact 2.2. (cid:3) Now, we can prove our classification result. Proposition 2.11. Let G be a non-abelian definably simple pseudofi- nite group of finite centralizer dimension. Then G is isomorphic to a (twisted) Chevalley group over a pseudofinite field. Proof. By Lemma 2.9(b), we have G ≡ G /U where each G is i∈I i i a non-abelian finite simple group. Since there are three families of Q non-abelian finite simple groups, without loss of generality, we assume that just one family occurs in the ultraproduct (see Remark 1.1). The possibilities are analyzed below. Case 1. Sporadic Groups Since there are finitely many sporadic groups, we may assume that all G ’s are the same sporadic group H, that is, G ≡ HI/U. However, i this forces G to be finite which is not the case. Case 2. Alternating Groups It is well-known that the centralizer dimension of alternating groups increases as the rank increases. More precisely, if we consider the cen- tralizer chain of the form C(1) > C((12)(34)) > ··· > C((12)(34),...,(k −3,k −2)(k −1,k)), we can observe that the centralizer dimension of Alt(n) is greater than n −1. Therefore, the finite centralizer dimension property is satisfied 4

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.