Description

SMALL SKEW FIELDS CÉDRIC MILLIET Abstract A division ring of positive characteristic with countably many pure types is a field Wedderburn showed in 1905 that finite fields are commutative As for infinite

Information

Category:
## Devices & Hardware

Publish on:

Views: 26 | Pages: 6

Extension: PDF | Download: 0

Share

Transcript

SMALL SKEW FIELDS CÉDRIC MILLIET Abstract A division ring of positive characteristic with countably many pure types is a field Wedderburn showed in 1905 that finite fields are commutative As for infinite fields, we know that superstable [1, Cherlin, Shelah] and supersimple [4, Pillay, Scanlon, Wagner] ones are commutative In their proof, Cherlin and Shelah use the fact that a superstable field is algebraically closed Wagner showed that a small field is algebraically closed [5], and asked whether a small field should be commutative We shall answer this question positively in non-zero characteristic 1 Preliminaries Definition 11 A theory is small if it has countably many n-types without parameters for all integer n A structure is small if its theory is so We shall denote dcl(a) the definable closure of a set A Note that if K is a field and A a subset of K, then dcl(a) is a field too Smallness is clearly preserved under interpretation and addition of finitely many parameters Let D, D 1, D 2 be A-definable sets in some structure M with A M We define the Cantor-Bendixson rank CB A (D) and degree dcb A (D) of D over A Definition 12 By induction, we define CB A (D) 0 if D is not empty CB A (D) α+1 if there is an infinite family of disjoint A-definable subsets D i of D, such that CB A (D i ) α for all i ω CB A (D) β for a limit ordinal β if CB A (D) α for all α β Definition 13 dcb A (D) is the greatest integer d such that D can be divided into d disjoint A-definable sets, with same rank over A as D Proposition 14 If M is small and A is a finite set, (i) The rank CB A (M) is ordinal (ii) The degree dcb A is well defined (iii) If D 1 D 2, then CB A (D 1 ) CB A (D 2 ) (iv) CB A (D 1 D 2 ) = max{cb A (D 1 ), CB A (D 2 )} (v) CB A and dcb A are preserved under A-definable bijections If A is empty, we shall write CB and dcb rather than CB or dcb 2000 Mathematics Subject Classification 03C15, 03C50, 03C60, 12E15 Key words and phrases Smallness, skew fields 1 SMALL SKEW FIELDS 2 Remark 15 Let H G be A-definable small groups with H dcl(a) G dcl(a) Then, either CB A (H) CB A (G), or CB A (H) = CB A (G) and dcb A (H) dcb A (G) Corollary 16 A small integral domain with unity is a field Proof Let R be this ring If r is not invertible, then 1 / rr hence rr dcl(r) R dcl(r) Apply Remark 15, but R and rr have same rank and degree over r Note that R need not have a unity (see Corollary 110) More generally, if ϕ is a definable bijection between two definable groups A B in a small structure, then A equals B Proposition 17 (Descending Chain Condition) Let G be a small group and g a finite tuple in G Let H be a subgroup of dcl(g) In H, there is no strictly decreasing infinite chain of subgroups of the form G 0 H G 1 H G 2 H , where the G i are H-definable subgroups of G Proof By Remark 5, either the rank or the degree decreases at each step Corollary 18 Let G be a small group, g a finite tuple, H subgroup of dcl(g), and G i a family of g-definable subgroups of G indexed by I There is a finite subset I 0 of I such that G i H = G i H i I 0 Another chain condition on images of endomorphisms : i I Proposition 19 Let G be a small group and h a group homomorphism of G There exists some integer n such that Imh n equals Imh n+1 Proof Suppose that the chain (Imh n ) n 1 be strictly decreasing Consider the following tree G(x) Imh Imh 2 h(x)imh 2 Imh 3 h 2 (x)imh 3 h(x)imh 3 h(x)h 2 (x)imh 3 Consider the partial type {x / h n Imh n+1, n 1} We call it Φ(x) The sequence (h n Imh n+1 ) n 1 is increasing, and each set G \ h n Imh n+1 is non-empty, so Φ is finitely consistent Let b be a realization of Φ in a saturated model The graph G(b) has 2 ω consistent branches, whence S 1 (b) has cardinal 2 ω, a contradiction with G being small Corollary 110 Let G be a small group and h a group homomorphism of G There exists some integer n such that G equals Kerh n Imh n SMALL SKEW FIELDS 3 Proof Take n as in Proposition 19, and write f for h n We have Imf 2 = Imf, so for all g G there exists some element g such that f(g) = f 2 (g ) Hence f(gf(g ) 1 ) = 1 and gf(g ) 1 Kerf, that is g Kerf Imf It was shown in [6] that a definable injective homomorphism of a small group is surjective Note that this follows again from Corollary 110 Recall a result proved in [5] : 2 Small skew fields Fact 21 An infinite small field is algebraically closed From now on, consider an infinite small skew field D elements of finite order We begin by analysing Lemma 22 Let a D an element of order n ω Then a is central in D Proof Either D has zero characteristic, so Z(D) is infinite, hence algebraically closed But Z(D)(a) is an extension of Z(D) of degree d n, whence a Z(D) Or D has positive characteristic Suppose that a is not central, then [3, Lemma 311] there exists x in D such that xax 1 = a i a If x has finite order, then all elements in the multiplicative group x, a have finite order Hence x, a is commutative [3, Lemma 313], contradicting xax 1 a So x has infinite order Conjugating m times by x, we get x m ax m = a im But a and a i have same order n, with gcd(n, i) = 1 Put m = φ(n) By Fermat s Theorem, i m 1[n], so x m and a commute Then L = Z(C D (a, x m )) is a definable infinite commutative subfield of D which contains a Let L x be {l L, x 1 lx = l} This is a proper subfield of L Moreover 1 [L x (a) : L x ] n But L x is infinite as it contains x By Fact 21, it is algebraically closed and cannot have a proper extension of finite degree Proposition 23 Every element of D has a n th root for each integer n Proof Let a D If a has infinite order, Z(C D (a)) is an infinite commutative definable subfield of D Hence it is algebraically closed, and a has an n th root in Z(C D (a)) Otherwise a has finite order According to Lemma 22 it is central in D Let x D have infinite order Then a Z(C D (a, x)), a commutative, infinite, definable, and thus algebraically closed field Remark 24 Note that since D is divisible, it has elements of arbitrary large finite order, which are central by Lemma 22 Taking D omega-saturated, we can suppose Z(D) infinite Let us now show that a small skew field is connected, that is to say, has no definable proper subgroup of finite index Proposition 25 D is connected Proof Multiplicatively : By Proposition 23, D is divisible so has no subgroup of finite index Additively : Let H be a definable subgroup of D + of finite index n In zero characteristic, D + is divisible, so n = 1 In general, let k be an infinite finitely generated subfield of D Consider a finite intersection G = i I d ih of left SMALL SKEW FIELDS 4 translates of H by elements in k such that G k is minimal ; this exists by the chain condition By minimality, G k = d k dh k, so G k is a left ideal of k Furthermore, G is a finite intersection of subgroups of finite index in D + ; it has therefore finite index in D Thus G k has finite index in D k = k, and cannot be trivial, so G k = k = H k This holds for every infinite finitely generated k, whence H = D Now we look at elements of infinite order Lemma 26 a D have infinite order Then C D (a) = C D (a n ) for all n 0 Proof Clearly C D (a) C D (a n ) Consider L = Z(C D (a n )) It is algebraically closed by Fact 21, but L(a) is a finite commutative extension of L, whence a L and C D (a n ) C D (a) Now suppose that D is not commutative We shall look for a commutative centralizer C and show that the dimension [D : C] is finite This will yield a contradiction Lemma 27 Let a D, t D \ im(x ax xa) and ϕ : x t 1 (ax xa) Then D = imϕ kerϕ Moreover, if k = dcl(a, t, x), where x is a finite tuple, then k = imϕ k kerϕ k Proof Let K = kerϕ = C D (a) Put I = imϕ; this is a right K vector space, so I K = {0}, since 1 K I is impossible by the choice of t Consider the morphism D ϕ : + /K D + /K x ϕ(x) ϕ is an embedding, and D + /K is small ; by Corollary 10, ϕ is surjective and D/K = ϕ(d/k), hence D = I K Now, let k = dcl(a, t, x) where x is a finite tuple of parameters in D I and K are k definable For all α k there exists a unique couple (α 1, α 2 ) I K such that α = α 1 + α 2, so α 1 and α 2 belong to dcl(α, a, t) k, that is to say k = I k K k Lemma 28 For every a / Z(D), the map ϕ a : x ax xa is onto Proof Suppose ϕ a not surjective Let t / imϕ a, and k = dcl(t, a, x) be a non commutative subfield of D for some finite tuple x Consider the morphism ϕ : D + D + x t 1 (ax xa) Set H = imϕ, and K = C D (a) = kerϕ By Lemma 27 we get k = H k K k Let N = I a ih be a finite intersection of left-translates of H by elements in k, such that N k be minimal We have N k = i I a i H k = d k dh k, so N k is a left ideal Moreover, H k is a right K k vector-space of codimension 1 Then N k has codimension at most n = I If N k = k, then H k = k, whence K k = {0}, a contradiction So N k is trivial and, k is a K k-vector space of dimension at most n By [2, Corollary 2 p49] we get [k : K k] = [Z(k)(a) : Z(k)] But Z(k) = Z(C D (k)) k with Z(C D (k)) algebraically closed Note that every SMALL SKEW FIELDS 5 element of k commutes with Z(C D (k)), so a Z(k), which is absurd if we add b / C D (a) in k Theorem 29 A small field in non-zero characteristic is commutative Proof Let a D be non-central, and let us show that x ax xa is not surjective Otherwise there exists x such that ax xa = a, hence axa 1 = x + 1 We would then have a p xa p = x + p = x, and x C D (a p ) \ C D (a), a contradiction with Lemma 22 3 Open problems 31 Zero characteristic Note that we just use characteristic p in proof of theorem 19 to show that there exist a / Z(D) such that x ax xa is not surjective Thus questions 1 and 2 are equivalent : Question 1 Is a small skew field D of zero characteristic commutative? Question 2 Is every x ax xa surjective onto D for a / Z(D)? 32 Weakly small fields Weakly small structures have been introduced to give a common generalization of small and minimal structures Minimal fields are known to be commutative Definition 31 A structure M is weakly small if for all finite set of parameters A in M, there are only countably many 1-types over A Question 3 Is a weakly small field algebraically closed? Question 4 Is a weakly small skew field commutative? Note that a positive answer to question 3 implies a positive answer to question 4, as all the proves given still hold In general, one can prove divisibility and connectivity of an infinite weakly small field Proposition 32 Every element in an infinite weakly small field D has a n th root for all n ω Proof Let a D In zero characteristic, Z(C D (a)) is an infinite definable commutative subfield of D, hence weakly small According to [5, Proposition 9], every element in Z(C D (a)) has a n th root In positive characteristic, we can reason as in the proof of Lemma 12, and find y with infinite order which commutes with a Apply one more time [5, Proposition 9] to Z(C D (a, y)) So D is divisible and the proof of Proposition 25 still holds Proposition 33 An infinite weakly small field is connected SMALL SKEW FIELDS 6 References [1] Gregory Cherlin and Saharon Shelah, Superstable fields and groups, Annals of Mathematical Logic 18, 1980 [2] Paul M Cohn, Skew fields constructions, Cambridge University Press, 1977 [3] Israel N Herstein, Noncommutative Rings, The Mathematical Association of America, fourth edition, 1996 [4] Anand Pillay, Thomas Scanlon and Frank O Wagner, Supersimple fields and division rings, Mathematical Research Letters 5, , 1998 [5] Frank O Wagner, Small fields, The Journal of Symbolic Logic 63, 3, 1998 [6] Frank O Wagner, Small stable groups and generics, The Journal of Symbolic Logic 56, 1991 Current address, Cédric Milliet: Université de Lyon, Université Lyon 1, Institut Camille Jordan UMR 5208 CNRS, 43 boulevard du 11 novembre 1918, Villeurbanne Cedex, France address, Cédric Milliet:

Related Search

Similar documents

We Need Your Support

Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks