Focal subgroup theorem: Difference between revisions

From Groupprops
 
(One intermediate revision by the same user not shown)
Line 40: Line 40:


<math>G' = [G,G] = \langle xy^{-1} \mid x,y \in G, \exists g \in G, gxg^{-1} = y \rangle</math>.
<math>G' = [G,G] = \langle xy^{-1} \mid x,y \in G, \exists g \in G, gxg^{-1} = y \rangle</math>.
and:
<math>P' = [P,P] = \langle xy^{-1} \mid x,y \in P, \exists g \in P, gxg^{-1} = y \rangle</math>.


'''To prove''': <math>P_0 = P \cap G'</math>.
'''To prove''': <math>P_0 = P \cap G'</math>.
Line 45: Line 49:
===Initial observations===
===Initial observations===


First, note that <math>P' \subseteq P_0 \subseteq P \cap G'</math>. The first inclusion is because every [[commutator]] of elements in <math>P</math> is contained inside the generating set for <math>P_0</math>, and the second inclusion is because every element in the given generating set for <math>P_0</math> is both in <math>P</math> and in <math>G_0</math>.
{| class="sortable" border="1"
 
! Step no. !! Assertion/construction !! Facts used !! Given data used !! Previous steps used !! Explanation
Since <math>P_0</math> contains <math>P'</math>, <math>P/P_0</math> is an [[abelian group]].
|-
 
| 1 || <math>P' \subseteq P_0 \subseteq P \cap G'</math> || || || || <toggledisplay>The first inclusion is because every [[commutator]] of elements in <math>P</math> is contained inside the generating set for <math>P_0</math>, and the second inclusion is because every element in the given generating set for <math>P_0</math> is both in <math>P</math> and in <math>G'</math>.</toggledisplay>
===Proof using linear representation theory===
|-
 
| 2 || <math>P_0</math> is normal in <math>P</math> and <math>P/P_0</math> is abelian. || || || Step (1) || This follows from the part of Step (1) that asserts that <math>P' \le P_0</math>.
'''Claim''': Given any linear character on <math>P/P_0</math>, there exists a linear character on <math>G</math> extending it.
|-
 
| 3 || Let <math>p_1, p_2, \ldots, p_m</math> be the distinct prime divisors of <math>|G|</math> with <math>p=p_1</math>. For each <math>p_i</math>, pick a <math>p_i</math>-[[Sylow subgroup]] <math>P_i</math> such that <math>P_1 = P</math>. || || || ||
'''Proof''':
|-
 
| 4 || Any <math>g \in G</math> can be expressed as a product <math>g = g_1g_2\ldots g_m</math> where each <math>g_i</math> is conjugate to some <math>h_i \in P_i</math> and the <math>g_i</math> commute pairwise. || Fact (2) || || Step (3) || <toggledisplay>Consider the cyclic group <math>\langle g \rangle</math>. This is a direct product of its <math>p_i</math>-Sylow subgroups, so <math>g</math> can be expressed as a product of commuting elements <math>g_i</math>, one from each Sylow subgroup. <math>g_i</math> is an element of order a power of <math>p_i</math>, so by Fact (2), <math>g_i</math> is conjugate to an element <math>h_i \in P_i</math>.</toggledisplay>
# Let <math>p_1, p_2, \ldots, p_m</math> be the distinct prime divisors of <math>|G|</math> with <math>p=p_1</math>. For each <math>p_i</math>, pick a <math>p_i</math>-[[Sylow subgroup]] <math>P_i</math> such that <math>P_1 = P</math>.
|-
# Any <math>g \in G</math> can be expressed as a product <math>g = g_1g_2\ldots g_m</math> where each <math>g_i</math> is conjugate to some <math>h_i \in P_i</math> and the <math>g_i</math> commute pairwise. If <math>g_1</math> is also conjugate to <math>h_1' \in P_1</math> then <math>h_1h_1'^{-1} \in P_0</math>.  
| 5 || If <math>g_1</math> (continuing notation from Step (4), though it does not really matter) is conjugate to <math>h_1 \in P_1</math> and ''also'' conjugate to <math>h_1' \in P_1</math> then <math>h_1h_1'^{-1} \in P_0</math>. || || || || <toggledisplay>Because being conjugate is an equivalence relation, and the elements <math>h_1</math> and <math>h_1'</math> are both conjugate to <math>g_1</math>, they are conjugate to each other in <math>G</math>. Hence, by the definition of <math>P_0</math>, <math>h_1h_1'^{-1}</math> is in the generating set for <math>P_0</math>, and hence is in <math>P_0</math>.</toggledisplay>
# Let <math>\lambda</math> be a [[linear character]] of <math>P</math> whose kernel contains <math>P_0</math>. We claim that the function <math>\theta(g) = \lambda(h_1)</math>, where <math>h_1</math> is as described above, is well-defined: The reason is that if <math>h_1</math> and <math>h_1'</math> are two possibilities, then <math>\lambda(h_1h_1'^{-1}) = 1</math> and hence <math>\lambda(h_1) = \lambda(h_1')</math>.
|-
# <math>\theta</math> is a linear character of <math>G</math>: Clearly, <math>\theta</math> is a [[class function]] on <math>G</math>. Then by fact (1), we conclude that <math>\theta</math> is a [[linear character]] of <math>G</math>. Further, <math>\theta</math> extends <math>\lambda</math>, viz., the restriction of <math>\theta</math> to <math>P</math> is <math>\lambda</math>.
| 6 || Let <math>\lambda</math> be a linear character (i.e., the character of a one-dimensional representation) of <math>P</math> whose kernel contains <math>P_0</math>. The function <math>\theta(g) = \lambda(h_1)</math>, where <math>h_1</math> is as described in Step (4), is well-defined || || || Steps (4), (5) || <toggledisplay>The reason is that if <math>h_1</math> and <math>h_1'</math> are two possibilities, then by Step (5), <math>h_1h_1^{-1} \in P_0</math>, so <math>\lambda(h_1h_1'^{-1}) = 1</math> and hence <math>\lambda(h_1) = \lambda(h_1')</math> on account of <math>\lambda</math> being multiplicative since it's a linear character.</toggledisplay>
 
|-
'''How the result follows from the claim''': We now show that <math>P_0 = P \cap G'</math>. Suppose <math>P_0</math> is properly contained in <math>P \cap G'</math>. Then, we can choose a linear character <math>\lambda</math> of the abelian group <math>P/P_0</math> that is nontrivial on some element of <math>G'</math>. However, such a character cannot be extended to a linear character of the whole group, since any linear character of the whole group must be trivial on all elements of <math>G'</math>. Thus, <math>P_0</math> cannot be properly contained in <math>P \cap G'</math>.
| 7 || Continuing with the notation of Step (6), <math>\theta</math> is a class function on <math>G</math>. || || || || <toggledisplay>If <math>g,g'</math> are conjugate by <math>y</math>, then <math>g'_1</math> is the conjugate of <math>g'</math> by <math>y</math>. Hence, the element <math>h_1</math> of <math>P_1</math> that we choose as conjugate to <math>g_1</math> is ''also'' conjugate to <math>g'_1</math>, hence the definition of <math>\theta</math> gives the same value to <math>\theta(g)</math> and <math>\theta(g')</math>.</toggledisplay>
|-
| 8 || Continuing with the notation of Step (6), <math>\theta</math> is a linear character of <math>G</math>. In other words, <math>\theta(gg') = \theta(g)\theta(g')</math> for all <math>g,g' \in G</math>. || || || || ''This critical step is unclear''
|-
| 9 || For any linear character <math>\lambda</math> of <math>P</math> whose kernel contains <math>P_0</math>, there is a linear character <math>\theta</math> of <math>G</math> that extends <math>\lambda</math>, i.e., the restriction of <math>\theta</math> to <math>P</math> is <math>\lambda</math>. || || || Steps (6)-(8) || Step-combination direct, plus: <toggledisplay>We note that if <math>g \in P_1 = P</math> to begin with, then <math>g = g_1 = h_1</math>, so <math>\theta(g) = \lambda(h_1) = \lambda(g)</math>. Thus, the restriction of <math>\theta</math> to <math>P</math> is <math>\lambda</math>.</toggledisplay>
|-
| 10 || <math>P \cap G' \le P_0</math>. || || || || <toggledisplay>Suppose not. Then, we can choose a linear character <math>\lambda</math> of <math>P</math> with kernel containing <math>P_0</math> that is nontrivial on an element of <math>P \cap G'</math> that is outside <math>P_0</math>. However, such a character cannot be extended to a linear character of the whole group, since any linear character of the whole group must be trivial on all elements of <math>G'</math>. On the other hand, Step (9) shows that such a character can be extended to the whole group. Our assumption thus leads to a contradiction.</toggledisplay>
|-
| 11 || <math>P_0 = P \cap G'</math>. || || || Steps (1), (10) || Step-combination direct.
|}


===Proof using the transfer homomorphism===
===Proof using the transfer homomorphism===

Latest revision as of 22:46, 11 September 2011

This article gives the statement and possibly, proof, of an implication relation between two subgroup properties. That is, it states that every subgroup satisfying the first subgroup property (i.e., Sylow subgroup) must also satisfy the second subgroup property (i.e., subgroup whose focal subgroup equals its intersection with the commutator subgroup)
View all subgroup property implications | View all subgroup property non-implications
Get more facts about Sylow subgroup|Get more facts about subgroup whose focal subgroup equals its intersection with the commutator subgroup

Statement

Let P be a p-Sylow subgroup of a finite group G and let:

P0=xy1x,yP,gG,gxg1=y.

In other words, P0 is the focal subgroup of P in G.

Then:

PG=P0

In other words, P is a subgroup whose focal subgroup equals its intersection with the commutator subgroup.

Related facts

Facts used

For the proof using linear representation theory

  1. Brauer's induction theorem

For the proof using the transfer homomorphism

  1. Product decomposition for element in terms of transfer homomorphism

Proof

Given: A finite group G, a p-Sylow subgroup P. P0 is the focal subgroup of P, defined by:

P0=xy1x,yP,gG,gxg1=y.

Further, we have:

G=[G,G]=xy1x,yG,gG,gxg1=y.

and:

P=[P,P]=xy1x,yP,gP,gxg1=y.

To prove: P0=PG.

Initial observations

Step no. Assertion/construction Facts used Given data used Previous steps used Explanation
1 PP0PG [SHOW MORE]
2 P0 is normal in P and P/P0 is abelian. Step (1) This follows from the part of Step (1) that asserts that PP0.
3 Let p1,p2,,pm be the distinct prime divisors of |G| with p=p1. For each pi, pick a pi-Sylow subgroup Pi such that P1=P.
4 Any gG can be expressed as a product g=g1g2gm where each gi is conjugate to some hiPi and the gi commute pairwise. Fact (2) Step (3) [SHOW MORE]
5 If g1 (continuing notation from Step (4), though it does not really matter) is conjugate to h1P1 and also conjugate to h1P1 then h1h1'1P0. [SHOW MORE]
6 Let λ be a linear character (i.e., the character of a one-dimensional representation) of P whose kernel contains P0. The function θ(g)=λ(h1), where h1 is as described in Step (4), is well-defined Steps (4), (5) [SHOW MORE]
7 Continuing with the notation of Step (6), θ is a class function on G. [SHOW MORE]
8 Continuing with the notation of Step (6), θ is a linear character of G. In other words, θ(gg)=θ(g)θ(g) for all g,gG. This critical step is unclear
9 For any linear character λ of P whose kernel contains P0, there is a linear character θ of G that extends λ, i.e., the restriction of θ to P is λ. Steps (6)-(8) Step-combination direct, plus: [SHOW MORE]
10 PGP0. [SHOW MORE]
11 P0=PG. Steps (1), (10) Step-combination direct.

Proof using the transfer homomorphism

We have P0PG, with P0 normal in P and P/P0 abelian. In particular, we can construct the transfer homomorphism τ:GP/P0. Let φ:PP/P0 be the quotient map.

Claim: The restriction of τ to P is surjective to P/P0.

Proof: For any xP, we have, by fact (1), a collection of elements x1,x2,,xtG and nonnegative integers r1,r2,,rt such that ri=[G:P], and we have:

xixrixi1H,τ(x)=i=1txixrixi1modP0.

Since we chose xP, and P/P0 is abelian, we can rearrange terms to obtain:

τ(x)=i=1txrii=1txrixixrixi1modP0.

Every term in the second product is of the form xri(xixrixi1), which is of the form ab1 with a,bP conjugate in G (here a=xri). In particular, every term in the second product is in P0, and we obtain:

τ(x)=i=1txrimodP0,

which, since ri=[G:P], reduces to:

τ(x)=x[G:P]modP0.

Note now that since P is p-Sylow, [G:P] is relatively prime to p, and the map xx[G:P] is a bijection from P/P0 to itself. Thus, τ, restricted to P, surjects to P/P0.

Proof using the claim: Let K be the kernel of τ:GP/P0.

Since K is a kernel of a map to an abelian group, [G,G]K. Thus, P0[G,G]PKP. The restriction of τ to P is a bijective map from P/(PK) to P/P0 (surjective by the claim, and injective because we've quotiented by the kernel). Thus, the size of PK equals the size of P0 forcing P0=[G,G]P=KP. In particular, P0=[G,G]P.

References

Journal references

Textbook references

  • Finite Groups by Daniel Gorenstein, ISBN 0821843427, More info, Page 250, Theorem 3.4, Section 7.3 (Transfer and the focal subgroup)