Autocommuting probability of a finite group relative to its subgroups
Parama Dutta and Rajat Kanti Nath111Corresponding author
Abstract
Let H⊆K be two subgroups of a finite group G and Aut(K) the automorphism group of K. The autocommuting probability of G relative to its subgroups H and K, denoted by Pr(H,Aut(K)), is the probability that the autocommutator of a randomly chosen pair of elements, one from H and the other from Aut(K), is equal to the identity element of G. In this paper, we study Pr(H,Aut(K)) through a generalization.
*Department of Mathematical Sciences, Tezpur University,
Napaam-784028, Sonitpur, Assam, India.
Emails: [email protected] and [email protected]
Key words: Automorphism group, Autocommuting probability, Autoisoclinism.
2010 Mathematics Subject Classification: 20D60, 20P05, 20F28.
1 Introduction
Let G be a finite group acting on a set Ω. Let Pr(G,Ω) denote the probability that a randomly chosen element of Ω fixes a randomly chosen element of G. In 1975, Sherman [11] initiated the study of Pr(G,Ω) considering G to be an abelian group and Ω=Aut(G), the automorphism group of G. Note that
[TABLE]
where [x,α] is the autocommutator of x and α defined as x−1α(x). The ratio Pr(G,Aut(G)) is called autocommuting probability of G.
Let H and K be two subgroups of a finite group G such that H⊆K.
Motivated by the works in [2, 6], we define
[TABLE]
where g∈K.
That is, Prg(H,Aut(K)) is the probability that the autocommutator of a randomly chosen pair of elements, one from H and the other from Aut(K), is equal to a given element g∈K. The ratio Prg(H,Aut(K)) is called generalized
autocommuting probability of G relative to its subgroups H and K. Clearly, if H=G and g=1 then Prg(H,Aut(K))=Pr(G,Aut(G)). We would like to mention here that the case when H=G is considered in [3]. In this paper, we study Prg(H,Aut(K)) extensively. In particular, we obtain some computing formulae, various bounds and a few characterizations of G through a subgroup. We conclude the paper describing an invariance property of Prg(H,Aut(K)).
We write S(H,Aut(K)) to denote the set {[x,α]:x∈H and α∈Aut(K)} and [H,Aut(K)]:=⟨S(H,Aut(K))⟩. We also write
L(H,Aut(K)):={x∈H:[x,α]=1 for all α∈Aut(K)} and L(G):=L(G,Aut(G)), the absolute center of G (see [5]). Note that L(H,Aut(K)) is a normal subgroup of H contained in H∩Z(K). Further, L(H,Aut(K))=α∈Aut(K)∩CH(α), where CH(α)={x∈H:[x,α]=1} is a subgroup of H.
Let CAut(K)(x):={α∈Aut(K):α(x)=x} for x∈H and CAut(K)(H)={α∈Aut(K):α(x)=x for all x∈H}. Then CAut(K)(x) is a subgroup of Aut(K) and CAut(K)(H)=x∈H∩CAut(K)(x).
Clearly, Prg(H,Aut(K))=1 if and only if [H,Aut(K)]={1} and g=1 if and only if H=L(H,Aut(K)) and g=1. Also, Prg(H,Aut(K))=0 if and only if g∈/S(H,Aut(K)). Therefore, we consider H=L(H,Aut(K)) and g∈S(H,Aut(K)) throughout the paper.
2 Some computing formulae
For any x∈H, let us define the set Tx,g(H,K)={α∈Aut(K):[x,α]=g}, where g is a fixed element of K. Note that Tx,1(H,K)=CAut(K)(x). The following two lemmas play a crucial role in obtaining computing formula for Prg(H,Aut(K)).
Lemma 2.1**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If Tx,g(H,K)=ϕ then Tx,g(H,K)=σCAut(K)(x) for some σ∈Tx,g(H,K) and hence ∣Tx,g(H,K)∣=∣CAut(G)(x)∣.
Proof.
Let σ∈Tx,g(H,K) and β∈σCAut(K)(x). Then β=σα for some α∈CAut(K)(x). We have
[TABLE]
Therefore, β∈Tx,g(H,K) and so σCAut(K)(x)⊆Tx,g(H,K). Again, let γ∈Tx,g(H,K) then γ(x)=xg. We have σ−1γ(x)=σ−1(xg)=x and so σ−1γ∈CAut(K)(x). Therefore, γ∈σCAut(K)(x) which gives Tx,g(H,K)⊆σCAut(K)(x). Hence, the result follows.
∎
Consider the action of Aut(K) on K given by (α,x)↦α(x) where α∈Aut(K) and x∈K. Let orbK(x):={α(x):α∈Aut(K)} be the orbit of x∈K. Then by orbit-stabilizer theorem, we have
[TABLE]
Lemma 2.2**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Then Tx,g(H,K)=ϕ if and only if xg∈orbK(x).
Proof.
The result follows from the fact that α∈Tx,g(H,K) if and only if xg∈orbK(x).
∎
The following theorem gives two computing formulae for Prg(H,Aut(K)).
Theorem 2.3**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If g∈K then
[TABLE]
Proof.
We have {(x,α)∈H×Aut(K):[x,α]=g}=x∈H⊔({x}×Tx,g(H,K)), where ⊔ represents the union of disjoint sets. Therefore, by (1.1), we have
[TABLE]
Hence, the result follows from Lemma 2.1, Lemma 2.2 and (2.1).
∎
Considering g=1 in Theorem 2.3, we get the following computing formulae for Pr(H,Aut(K)).
Corollary 2.4**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Then
[TABLE]
where orbK(H)={orbK(x):x∈H}.
Corollary 2.5**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If CAut(K)(x)={I} for all x∈H∖{1}, where I is the identity element of Aut(K), then
[TABLE]
Proof.
By Corollary 2.4, we have
[TABLE]
Hence, the result follows.
∎
We also have
∣{(x,α)∈H×Aut(K):[x,α]=1}∣=α∈Aut(K)∑∣CH(α)∣ and hence
[TABLE]
We conclude this section with the following two results.
Proposition 2.6**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If g∈K then
[TABLE]
Proof.
Let
[TABLE]
Then (x,α)↦(α(x),α−1) gives a bijection between A and B. Therefore ∣A∣=∣B∣. Hence, the result follows from (1.1).
∎
Proposition 2.7**.**
Let G1 and G2 be two finite groups. Let H1,K1 and H2,K2 be subgroups of G1 and G2 respectively such that H1⊆K1, H2⊆K2 and gcd(∣K1∣,∣K2∣)=1. If (g1,g2)∈K1×K2 then
[TABLE]
Proof.
Let
[TABLE]
Since gcd(∣K1∣,∣K2∣)=1, by Lemma 2.1 of [1], we have Aut(K1×K2)=Aut(K1) ×Aut(K2). Therefore, for every αK1×K2∈Aut(K1×K2) there exist unique αK1∈Aut(K1) and αK2∈Aut(K2) such that αK1×K2=αK1×αK2, where αK1×αK2((x,y))=(αK1(x),αK2(y)) for all (x,y)∈H1×H2. Also, for all (x,y)∈H1×H2, we have [(x,y),αK1×K2]=(g1,g2) if and only if [x,αK1]=g1 and [y,αK2]=g2. These leads to show that X=Y×Z. Therefore
[TABLE]
Hence, the result follows from (1.1).
∎
3 Various bounds
In this section, we obtain various bounds for Prg(H,Aut(K)). We begin with the following lower bounds.
Proposition 3.1**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Then, for g∈K, we have
- (a)
Prg(H,Aut(K))≥∣H∣∣L(H,Aut(K))∣+∣H∣∣Aut(K)∣∣CAut(K)(H)∣(∣H∣−∣L(H,Aut(K))∣)* if g=1.*
2. (b)
Prg(H,Aut(K))≥∣H∣∣Aut(K)∣∣L(H,Aut(K))∣∣CAut(K)(H)∣* if g=1.*
Proof.
Let C denote the set {(x,α)∈H×Aut(K):[x,α]=g}.
(a) We have (L(H,Aut(K))×Aut(K))∪(H×CAut(K)(H)) is a subset of C and
∣(L(H,Aut(K))×Aut(K))∪(H×CAut(K)(H))∣=∣L(H,Aut(K))∣∣Aut(K)∣+∣CAut(K)(H)∣∣H∣−∣L(H,Aut(K))∣∣CAut(K)(H)∣. Hence, the result follows from (1.1).
(b) Since g∈S(H,Aut(K)) we have C is non-empty. Let (y,β)∈C then (y,β)∈/L(H,Aut(K))×CAut(K)(H) otherwise [y,β]=1. It is easy to see that the coset (y,β)(L(H,Aut(K))×CAut(K)(H)) is a subset of C having order ∣L(H,Aut(K))∣∣CAut(K)(H)∣. Hence, the result follows from (1.1).
∎
Proposition 3.2**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If g∈K then
[TABLE]
The equality holds if and only if g=1.
Proof.
By Theorem 2.3, we have
[TABLE]
The equality holds if and only if xg∈orbK(x) for all x∈H if and only if g=1.
∎
Proposition 3.3**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Let g∈K and p the smallest prime dividing ∣Aut(K)∣. If g=1 then
[TABLE]
Proof.
By Theorem 2.3, we have
[TABLE]
noting that for x∈L(H,Aut(K)) we have xg∈/orbK(x). Also, for x∈H∖L(H,Aut(K)) and xg∈orbK(x) we have ∣orbK(x)∣>1. Since ∣orbK(x)∣ is a divisor of ∣Aut(K)∣ we have ∣orbK(x)∣≥p. Hence, the result follows from (3.1).
∎
Proposition 3.4**.**
Let H1, H2 and K be subgroups of a finite group G such that H1⊆H2⊆K. Then
[TABLE]
The equality holds if and only if xg∈/orbK(x) for all x∈H2∖H1.
Proof.
By Theorem 2.3, we have
[TABLE]
Hence, the result follows.
∎
Proposition 3.5**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If g∈K then
[TABLE]
with equality if and only if g=1 and H=K.
Proof.
By Proposition 3.2, we have
[TABLE]
Hence, the result follows from Corollary 2.4.
∎
Note that if we replace Aut(K) by Inn(K), the inner automorphism group of K, in (1.1) then Prg(H,Inn(K))=Prg(H,K) where
[TABLE]
A detailed study on Prg(H,K) can be found in [2]. The following proposition gives a relation between Prg(H,Aut(K)) and Prg(H,K) for g=1.
Proposition 3.6**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If g=1 then
[TABLE]
Proof.
If g=1 then by [2, Theorem 2.3], we have
[TABLE]
where clK(x)={α(x):α∈Inn(K)}. Since clK(x)⊆orbK(x) for all x∈H, the result follows from (3.2) and Theorem 2.3.
∎
Theorem 3.7**.**
Let H and K be two subgroups of a finite group G such that H⊆K and p the smallest prime dividing ∣Aut(K)∣. Then
[TABLE]
and
[TABLE]
where XH={x∈H:CAut(K)(x)={I}}.
Proof.
We have XH∩L(H,Aut(K))=ϕ. Therefore
[TABLE]
For x∈H∖(XH∪L(H,Aut(K))) we have {I}=CAut(K)(x)=Aut(K) which implies p≤∣CAut(K)(x)∣≤p∣Aut(K)∣. Therefore
[TABLE]
and
[TABLE]
Hence, the result follows from Corollary 2.4, (3) and (3).
∎
We have the following two corollaries.
Corollary 3.8**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If p and q are the smallest primes dividing ∣Aut(K)∣ and ∣H∣ respectively then
[TABLE]
In particular, if p=q then Pr(H,Aut(K))≤p22p−1≤43.
Proof.
Since H=L(H,Aut(K)) we have ∣H:L(H,Aut(K))∣≥q. Therefore, by Theorem 3.7, we have
[TABLE]
∎
Corollary 3.9**.**
Let H and K be two subgroups of a finite group G such that H⊆K and p, q be the smallest primes dividing ∣Aut(K)∣ and ∣H∣ respectively. If H is non-abelian then
[TABLE]
In particular, if p=q then Pr(H,Aut(K))≤p3p2+p−1≤85.
Proof.
Since H is non-abelian we have ∣H:L(H,Aut(K))∣≥q2. Therefore, by Theorem 3.7, we have
[TABLE]
∎
Now we obtain two lower bounds analogous to the lower bounds obtained in [9, Theorem A] and [8, Theorem 1].
Theorem 3.10**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Then
[TABLE]
The equality holds if and only if orbK(x)=xS(H,Aut(K))
for all x∈H∖L(H,Aut(K)).
Proof.
For all x∈H∖L(H,Aut(K)) we have α(x)=x[x,α]∈xS(H,Aut(K)). Therefore orbK(x)⊆xS(H,Aut(K)) and so ∣orbK(x)∣≤∣S(H,Aut(K))∣
for all x∈H∖L(H,Aut(K)). Now, by Corollary 2.4, we have
[TABLE]
Hence, the result follows.
∎
Lemma 3.11**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Then, for any two integers m≥n, we have
[TABLE]
If L(H,Aut(K))=H then equality holds if and only if m=n.
Proof.
The proof is an easy exercise.
∎
Corollary 3.12**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Then
[TABLE]
If H=L(H,Aut(K)) then the equality holds if and only if [H,Aut(K)]=S(H,Aut(K)) and orbK(x)=x[H,Aut(K)]
for all x∈H∖L(H,Aut(K)).
Proof.
Since ∣[H,Aut(K)]∣≥∣S(H,Aut(K))∣, the result follows from Theorem 3.10 and Lemma 3.11.
Note that the equality holds if and only if equality holds
in Theorem 3.10 and Lemma 3.11.
∎
It is worth mentioning that Theorem 3.10 gives better lower bound than the lower bound given by Corollary 3.12. Also
[TABLE]
Hence, Theorem 3.10 gives better lower bound than the lower bound given by Theorem 3.7.
4 A few Characterizations
In this section, we obtain some characterizations of a subgroup H of G if equality holds in Corollary 3.8 and Corollary 3.9. We begin with the following result.
Theorem 4.1**.**
Let H and K be two subgroups of a finite group G such that H⊆K. If Pr(H,Aut(K))=pqp+q−1 for some primes p and q. Then pq divides ∣H∣∣Aut(K)∣. Further, if p and q are the smallest primes dividing ∣Aut(K)∣ and ∣H∣ respectively, then
[TABLE]
In particular, if H and Aut(K) are of even order and Pr(H,Aut(K))=43 then L(H,Aut(K))H≅Z2.
Proof.
By (1.1), we have (p+q−1)∣H∣∣Aut(K)∣=pq∣{(x,α)∈H×Aut(K):[x,α]=1}∣. Therefore, pq divides ∣H∣∣Aut(K)∣.
If p and q are the smallest primes dividing ∣Aut(K)∣ and ∣H∣ respectively then, by Theorem 3.7, we have
[TABLE]
which gives ∣H:L(H,Aut(K))∣≤q. Hence, L(H,Aut(K))H≅Zq.
∎
Theorem 4.2**.**
Let H⊆K be two subgroups of a finite group G such that H is non-abelian and Pr(H,Aut(K))=pq2q2+p−1 for some primes p and q. Then pq divides ∣H∣∣Aut(K)∣. Further, if p and q are the smallest primes dividing ∣Aut(K)∣ and ∣H∣ respectively then
[TABLE]
In particular, if H and Aut(K) are of even order and Pr(H,Aut(K))=85 then L(H,Aut(K))H≅Z2×Z2.
Proof.
By (1.1), we have (q2+p−1)∣H∣∣Aut(K)∣=pq2∣{(x,α)∈H×Aut(K):[x,α]=1}∣. Therefore, pq divides ∣H∣∣Aut(K)∣.
If p and q are the smallest primes dividing ∣Aut(K)∣ and ∣H∣ respectively then, by Theorem 3.7, we have
[TABLE]
which gives ∣H:L(H,Aut(K))∣≤q2. Since H is non-abelian we have ∣H:L(H,Aut(K))∣=1,q. Hence, L(H,Aut(K))H≅Zq×Zq.
∎
The following two results give partial converses of Theorem 4.1 and 4.2 respectively.
Proposition 4.3**.**
Let H and K be two subgroups of a finite group G such that H⊆K. Let p,q be the smallest prime divisors of ∣Aut(K)∣, ∣H∣ respectively and ∣Aut(K):CAut(K)(x)∣=p for all x∈H∖L(H,Aut(K)).
- (a)
If L(H,Aut(K))H≅Zq then Pr(H,Aut(K))=pqp+q−1.
2. (b)
If L(H,Aut(K))H≅Zq×Zq then Pr(H,Aut(K))=pq2q2+p−1.
Proof.
Since ∣Aut(K):CAut(K)(x)∣=p for all x∈H∖L(H,Aut(K)) we have ∣CAut(K)(x)∣=p∣Aut(K)∣ for all x∈H∖L(H,Aut(K)). Therefore, by Corollary 2.4, we have
[TABLE]
Thus
[TABLE]
(a) If L(H,Aut(K))H≅Zq then (4.1) gives Pr(H,Aut(K))=pqp+q−1.
(b) If L(H,Aut(K))H≅Zq×Zq then (4.1) gives Pr(H,Aut(K))=pq2q2+p−1.
∎
5 Autoisoclinic pairs
In the year 1940, Hall [4] introduced the concept of isoclinism between two groups.
Following Hall, Moghaddam et al. [7] have defined autoisoclinism between two groups, in the year 2013. Recall that two groups G1 and G2 are said to be autoisoclinic if there exist isomorphisms ψ:L(G1)G1→L(G2)G2, β:[G1,Aut(G1)]→[G2,Aut(G2)] and γ:Aut(G1)→Aut(G2) such that the following diagram commutes
\begin{CD}\frac{G_{1}}{L(G_{1})}\times\operatorname{Aut}(G_{1})@>{\psi\times\gamma}>{}>\frac{G_{2}}{L(G_{2})}\times\operatorname{Aut}(G_{2})\\
@V{}V{a_{(G_{1},\operatorname{Aut}(G_{1}))}}V@V{}V{a_{(G_{2},\operatorname{Aut}(G_{2}))}}V\\
[G_{1},\operatorname{Aut}(G_{1})]@>{\beta}>{}>[G_{2},\operatorname{Aut}(G_{2})]\end{CD}
where the maps a(Gi,Aut(Gi)):L(Gi)Gi×Aut(Gi)→[Gi,Aut(Gi)], for i=1,2,
are given by
[TABLE]
Such a pair (ψ×γ,β) is called an autoisoclinism between the groups G1 and G2.
We generalize the notion of autoisoclinism in the following way:
Let H1,K1 and H2,K2 be subgroups of the groups G1 and G2 respectively. The pairs of subgroups (H1,K1) and (H2,K2) such that H1⊆K1 and H2⊆K2 are said to be autoisoclinic if there exist isomorphisms ψ:L(H1,AutK1)H1→L(H2,Aut(K2))H2, β:[H1,Aut(K1)]→[H2,Aut(K2)] and γ:Aut(K1)→Aut(K2) such that the following diagram commutes
\begin{CD}\frac{H_{1}}{L(H_{1},\operatorname{Aut}(K_{1}))}\times\operatorname{Aut}(K_{1})@>{\psi\times\gamma}>{}>\frac{H_{2}}{L(H_{2},\operatorname{Aut}(K_{2}))}\times\operatorname{Aut}(K_{2})\\
@V{}V{a_{(H_{1},\operatorname{Aut}(K_{1}))}}V@V{}V{a_{(H_{2},\operatorname{Aut}(K_{2}))}}V\\
[H_{1},\operatorname{Aut}(K_{1})]@>{\beta}>{}>[H_{2},\operatorname{Aut}(K_{2})]\end{CD}
where the maps a(Hi,Aut(Ki)):L(Hi,Aut(Ki))Hi×Aut(Ki)→(Hi,Aut(Ki)), for i=1,2, are given by
[TABLE]
Such a pair (ψ×γ,β) is said to be an autoisoclinism between the pairs of groups (H1,K1)
and (H2,K2). We conclude this paper with the following generalization of [3, Theorem 5.1] and [10, Lemma 2.5].
Theorem 5.1**.**
Let G1 and G2 be two finite groups with subgroups H1,K1 and H2,K2 respectively such that H1⊆K1 and H2⊆K2. If (ψ×γ,β) is an autoisoclinism between the pairs (H1,K1) and
(H2,K2) then, for g∈K1,
[TABLE]
Proof.
Let us consider the sets Sg={(x1L(H1,Aut(K1)),α1)∈L(H1,Aut(K1))H1×Aut(K1):[x1L(H1,Aut(K1)),α1]=g} and Tβ(g)={(x2L(H2,Aut(K2)),α2)∈L(H2,Aut(K2))H2×Aut(K2):[x2L(H2,Aut(K2)),α2]=β(g)}. Since (H1,K1) is
autoisoclinic to (H2,K2) we have ∣Sg∣=∣Tβ(g)∣. Again, it is clear that
[TABLE]
and
[TABLE]
Hence, the result follows from (1.1), (5.1) and (5.2).
∎