OUTLINE
S ETS OF E LEMENTS THAT P AIRWISE
G ENERATE A G ROUP
Andrea Lucchini
University of Padova, Italy
Joint work with Attila Maroti
GLAUBERMAN CONFERENCE Chicago, 25 March 2008
O
1
A
GRAPH ASSOCIATED TO A2-
GENERATED GROUP2
F
INITE SIMPLE GROUPS3
W
HAT CAN BE PROVED FOR ARBITRARY GROUPS?
4
D
IRECT POWER OF SIMPLE GROUPS5
A S
UDOKUG
AME6
S
OLVABLE GROUPSAGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
THE MAIN DEFINITION
GRAPH THEORETICAL TERMINOLOGY
DEFINITIONS
G a finite (non cyclic) group that can be generated by two elements;
µ(G) the largest positive integer m so that there exists a subset X in G of order m, with the property that any distinct pair of elements of X generates G.
DEFINITIONS
G a finite (non cyclic) group that can be generated by two elements;
µ(G) the largest positive integer m so that there exists a subset X in G of order m, with the property that any distinct pair of elements of X generates G.
EXAMPLE
µ(Alt(5)) = 8. We can consider
X := {(1, 2, 3), (3, 4, 5), (1, 2, 3, 4, 5), (1, 2, 3, 5, 4), (1, 2, 4, 3, 5), (1, 2, 4, 5, 3), (1, 2, 5, 3, 4), (1, 2, 5, 4, 3)}
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
THE MAIN DEFINITION
GRAPH THEORETICAL TERMINOLOGY
IN GRAPH THEORETICAL NOTATIONS:
We define a graphΓ(G)on the set of elements of G, connecting two vertices by an edge if they generate G.
µ(G) is theclique numberof Γ(G) (the maximum size of a complete subgraph).
REMARK
µ(G) ≤ χ(Γ(G)), where χ(Γ(G)) is thechromatic numberof Γ(G), i.e the smallest number of colors needed to color the vertices of Γ(G) so that no two adjacent vertices share the same color.
REMARK
If G is a union of m proper subgroups, then χ(Γ(G)) ≤ m (we assign different colors to the different subgroups and a vertex can receive any of the colors associated to the covering subgroups containing
IN GRAPH THEORETICAL NOTATIONS:
We define a graphΓ(G)on the set of elements of G, connecting two vertices by an edge if they generate G.
µ(G) is theclique numberof Γ(G) (the maximum size of a complete subgraph).
REMARK
µ(G) ≤ χ(Γ(G)), where χ(Γ(G)) is thechromatic numberof Γ(G), i.e the smallest number of colors needed to color the vertices of Γ(G) so that no two adjacent vertices share the same color.
REMARK
If G is a union of m proper subgroups, then χ(Γ(G)) ≤ m. Hence χ(Γ(G)) ≤ σ(G), whereσ(G)is the least integer k such that G is the
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
THE MAIN DEFINITION
GRAPH THEORETICAL TERMINOLOGY
SUMMARIZING
Γ(G) {x, y } ∈ E(Γ(G)) ⇐⇒ hx, y i = G µ(G) maximum size of a complete subgraph
χ(Γ(G)) the smallest number of colors so that no two adjacent vertices share the same color
σ(G) the least integer k such that G is the union of k proper subgroups
µ(G) ≤ χ(Γ(G)) ≤ σ(G)
K NOWN RESULTS FOR SIMPLE GROUPS
Let S be a finite nonbelian simple group and letm(S)be the minimal index of a proper subgroup of S.
(Liebeck and Shalev, 1995)There exists a constant c such that for any S we have c · m(S) ≤ µ(S)(Using probabilistic results on the generation of finite simple groups combined with Turán’s Theorem of extremal graph theory).
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
K NOWN RESULTS FOR SIMPLE GROUPS
Let S be a finite nonbelian simple group and letm(S)be the minimal index of a proper subgroup of S.
(Liebeck and Shalev, 1995)There is a constant c such that c · m(S) ≤ µ(S) for any finite nonabelian simple group S.
(Blackburn, 2006)If S = Alt(n), n is a sufficiently large integer and n ≡ 2 mod 4, then µ(S) = σ(S) = 2n−2.
K NOWN RESULTS FOR SIMPLE GROUPS
Let S be a finite nonbelian simple group and letm(S)be the minimal index of a proper subgroup of S.
(Liebeck and Shalev, 1995)There is a constant c such that c · m(S) ≤ µ(S) for any finite nonabelian simple group S.
(Blackburn, 2006)If S = Alt(n), n is a sufficiently large integer and n ≡ 2 mod 4, then µ(S) = σ(S) = 2n−2.
(Britnell, Evseev, Guralnick, Holmes and Maroti, 2007)Let S = PSL(n, q), n ≥ 12. If q is odd or n 6≡ 2 mod 4, then µ(S) = σ(S). In any case µ(S)σ(S) ≥ 1 −qq−1n−1 =1 − m(S)1 .
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
K NOWN RESULTS FOR SIMPLE GROUPS
Let S be a finite nonbelian simple group and letm(S)be the minimal index of a proper subgroup of S.
(Liebeck and Shalev, 1995)There is a constant c such that c · m(S) ≤ µ(S) for any finite nonabelian simple group S.
(Blackburn, 2006)If S = Alt(n), n is a sufficiently large integer and n ≡ 2 mod 4, then µ(S) = σ(S) = 2n−2.
(Britnell, Evseev, Guralnick, Holmes and Maroti, 2007)Let S = PSL(n, q), n ≥ 12. If q is odd or n 6≡ 2 mod 4, then µ(S) = σ(S). In any case µ(S)σ(S) ≥ 1 −qq−1n−1 =1 − m(S)1 . If S = Sz(q), then µ(S)/σ(S) = 1 − 1/m(S).
In general, it seems interesting to investigate the quotient µ(S)/σ(S) for a finite nonabelian simple group.
QUESTION
Is it true that
µ(S)
σ(S) → 1 as |S| → ∞ ? Does there exist a universal constant c such that
µ(S)
σ(S) ≥ 1 − c m(S) for all nonabelian finite simple groups S?
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
W HAT ABOUT ARBITRARY FINITE GROUPS ?
No result is known about µ(G) when G is an arbitrary finite 2-generated group.
QUESTION
For simple groups the numbers µ(G), χ(Γ(G)) and σ(G) are not too different and coincide in many cases. Does something similar hold for any finite (non cyclic) 2-generated group?
Thecrownsof a group play a relevant role in the discussion of problems about generation.
DEFINITIONS
LetLbe the set of the monolithic primitive epimorphic images L of G.
For L ∈ L and t ∈ N definethe crown based powerof size t:
Lt = (l1, . . . ,lt) ∈Lt
l1≡ · · · ≡ lt mod soc L . LetRbe the set of normal subgroups R of G, minimal with respect to the property G/R ∼= Lt for some L ∈ L and t ∈ N.
PROPOSITION
x1, . . . ,xµpairwise generate G if and only if x1R, . . . , xµR pairwise
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
QUESTION
Assume that L is a 2-generated monolithic primitive group and define τ = τ (L)as the largest t with the property that Lt is 2-generated.
Compare µ(L) and µ(Lτ).How much smaller is µ(Lτ)than µ(L)?
D IRECT POWER OF SIMPLE GROUPS
If L is a finite nonabelian simple group, thenLτ=Lτ whereτis the number of the Aut(L)-orbits on the set of the generating pairs of L.
QUESTION
It is easy to prove that σ(L) = σ(Lτ); what about µ(Lτ)?
REMARK
Let {(xi,yi) |1 ≤ i ≤ τ } be a set of representatives for the orbits of Aut(L) on the set of the generating pairs of L;
define x := (x1, . . . ,xτ) ∈Lτ,y := (y1, . . . ,yτ) ∈Lτ. Lτ= h¯x , ¯y i ⇔ (¯x , ¯y ) = (x , y )γ∃γ ∈ Aut(Lτ).
The non isolated vertices of Γ(Lτ)are the elements of the setΩof
¯
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
µ(Lτ)is the clique number of ¯Γand can be estimated using:
PROPOSITION
If X is a clique and Y a coclique in a vertex-transitive graph on m vertices, then |X ||Y | ≤ m.
If M is a proper subgroup of L, then {(l1, . . . ,lτ) ∈ Ω |l1∈ M} is a coclique in ¯Γ.
THEOREM
Define PMas the conditional probability that (l1,l2) ∈M × L, given that hl1,l2i = L. Then µ(Lτ) ≤1/PM.
COROLLARY
There exists a constant c such that for any nonabelian simple group L we have
REMARK
If n ≡ 2 mod 4 and n is large enough then σ(Alt(n)τ) = σ(Alt(n)) = 2n−2;
µ(Alt(n)τ) ≤cn.
Hence in general µ(Lτ)is much smaller than σ(Lτ).
QUESTIONS
How sharp is our upper bound for µ(Lτ)?
µ(G) ≥ 3 for any 2-generated finite group. Does there exist a non abelian simple group L with µ(Lτ) >3?
What about the chromatic number of Γ(Lτ)?
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
A S UDOKU G AME
µ(Alt(5)) = 8 and σ(Alt(5)) = 10.
G = Alt(5)19is the largest 2-generated direct power of Alt(5).
REMARK
If µ(Alt(5)19) ≥n, then there exists a n × 19 matrix, whose entries are elements of Alt(5), such that
the n elements in each column pairwise generate Alt(5);
if l1 l2
m1 m2
is a 2 × 2 submatrix, then (l1,m1)and (l2,m2)are not Sym(5)-conjugate.
5 3 5 5 5 3 3 2 5 5 5 2 2 3 5 5 5 3 3
5 3 3 5 3 5 5 3 2 5 2 5 5 2 3 5 3 5 5
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
The first two rows of the matrix are known (their columns are a set of representatives for the 19 Sym(5)-conjugate classes of generating pairs). The other rows must contain elements of a given order in the same proportion: 10 element of order 5, 6 elements of order 3, 3 elements of order 2.
A column contains at most two entries of order 3, and the elements of order 3 must be positioned in such a way that no 2 × 2 submatrix contains only elements of order 3. This implies
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
QUESTION
Decide whether µ(Alt(5)19) =3 or µ(Alt(5)19) =4.
µ(Alt(5)19=4 if and only if we can construct a matrix with the requested properties and where the orders of the elements are distributed in one of the two following way:
5 3 5 5 5 3 3 2 5 5 5 2 2 3 5 5 5 3 3
5 3 3 5 3 5 5 3 2 5 2 5 5 2 3 5 3 5 5
5 5 5 3 3 5 3 5 5 3 3 5 3 5 5 2 2 5 2
5 5 3 3 5 3 5 5 3 2 5 3 5 5 2 3 5 2 5
5 3 5 5 5 3 3 2 5 5 5 2 2 3 5 5 5 3 3
5 3 3 5 3 5 5 3 2 5 2 5 5 2 3 5 3 5 5
5 5 5 3 3 5 3 5 5 3 3 5 3 5 5 2 2 5 2
S OLVABLE GROUPS
In the case of solvable groups the precise value of σ(G) is known:
THEOREM(TOMKINSON1997)
Let G be a finite non-cyclic solvable group and let H/K be the smallest chief factor of G having more than one complement in G.
Then σ(G) = |H/K | + 1.
We want to study the relationship between µ(G) and σ(G) when G is a 2-generated solvable group. We start by considering the crown based power.
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
DEFINITIONS AND USEFUL REMARKS
Assume that L is a 2-generated solvable primitive group:
L = V o H, with H an irreducible subgroup of GL(V ).
LetF =EndH(V ),r = |EndH(V )| :
Lt=Vto H is 2-generated if and only if t ≤ r .
Lr =Vr o His the largest 2-generated crown based power of L.
We may identify H with a subgroup of GL(r , F ).
H = hX1,X2i ⇒ rank Ir − X1 Ir− X2 = r .
W ∈ Vr can be viewed as a r × r matrix with coefficients over F . PROPOSITION
Assume W1,W2∈ Vr and H = hX1,X2i:
L =Vr X ,W X i ⇔ detIr − X1 Ir− X2 6= 0.
The study of the behavior of the crown based power is reduced to the following:
ALINEAR ALGEBRA QUESTION
Assume that Z1, . . . ,Zµare matrices in Mr ×r(F ) with the property that rank Zi Zj = r ∀i 6= j.
Can we find Y1, . . . ,Yµ∈ Mr ×r(F ) such that detZi Zj
Yi Yj
6= 0 ∀i 6= j ?
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
AN EASY EXAMPLE
Assume that X1, . . . ,Xµ pairwise generate H and that Xi− Xj is a non-singular matrix whenever i 6= j. Then
det1 − Xi 1 − Xj Ir Ir
=det1 − Xi Xi− Xj
Ir 0
6= 0;
we may take Y1= · · · =Ym=Ir and deduce thatµ(Lr) ≥ µ.
LEMMA
Assume that H is a 2-generated irreducible nilpotent subgroup of GL(V ). If L = V o H and r = dimEndH(V )V , then µ(Lr) = µ(L).
THEOREM
If G is a non cyclic 2-generated solvable group of Fitting length at most 2, then µ(G) = χ(Γ(G)) = σ(G).
QUESTION
Does there exist a 2-generated solvable group with µ(G) 6= σ(G)?
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
G ENERALIZATIONS TO PROFINITE GROUPS
DEFINITIONS
Let G be a profinite group that can be (topologically) generated by d ≥ 2 elements:
we denote byµd(G)the largest integer m with the property that there exists an m-tuple of elements of G such that any d distinct entries together (topologically) generate G.
LEMMA
If the non-cyclic group G can be generated by 2 elements, then (d − 1)µ2(G) ≤ µd(G) ≤ (d − 1)σ(G).
LEMMA
µ2(SL(2, Z)) = σ(SL(2, Z)) = 4;
µd(SL(2, Z)) = 4(d − 1) = µd(SL(2, Z/2Z)).
QUESTION
Is it true that µd(SL(n, Z)) = µd(SL(n, Z/2Z)) for all integers n and d greater than or equal to 2?
THEOREM
If n ≥ 12, then the following three statements are true.
1 µd( \SL(n, Z)) = µd(SL(n, Z/2Z)).
2 σ(SL(n, Z)) = σ( \SL(n, Z)) = σ(SL(n, Z/2Z)).
AGRAPH ASSOCIATED TO A2-GENERATED GROUP FINITE SIMPLE GROUPS WHAT CAN BE PROVED FOR ARBITRARY GROUPS? DIRECT POWER OF SIMPLE GROUPS A SUDOKUGAME SOLVABLE GROUPS PROFINITE GROUPS
REMARK
LetG =SL(n, Z).\ Whenever n ≥ 3, d ≥ 2 and d ≤ k ≤ µd(G), the probability is positive that a randomly chosen k -tuple with entries from G has the property that any d entries will together generate G.
Let G be a profinite group that can be generated by d elements;
let ν be the normalized Haar measure on G (or on some direct power Gt);
for any k ≥ d , let Ω(G, k , d ) be the set of k -tuples of elements of G with the property that every d distinct entries generate G;
let P(G, k , d ) = ν(Ω(G, k , d )) and P(G, d ) = P(G, d , d ).
THEOREM