Loading [MathJax]/jax/output/HTML-CSS/jax.js

Thursday, March 28, 2013

Abelian Classification Theorem (5.5.24)

MathJax TeX Test Page Dummit and Foote Abstract Algebra, section 5.5, exercise 24:

Let pα11pα22...pαrr be the prime decomposition of n. Prove: (|G|=nG is abelian)i,j{1,...,r}(αi{1,2}pipαjj1).

Proof: Recall that a direct product is abelian if and only if all of its factors are abelian (5.1.1).

() (1) Assume there is a prime factor pαkk in the prime decomposition of n such that αk3. Obtain a nonabelian group P of order p3j as described in the material and we have P×Zn/p3j is a nonabelian group of order n, contrary to assumption.

(2) Assume pipαjj1 for some i,j. If αj=1, then obtain the nonabelian group P of order pipj also as presented in the material and we have P×Zn/(pipj) is a nonabelian group of order n, also contrary to assumption. If αj=2, then we have pip2j1=(pj1)(pj+1). Now, |Aut(Zpj×Zpj)|=pj(pj1)2(pj+1), so it contains a cyclic group of order pi by Cauchy, to which φ shall nontrivially map Zpi. We obtain H=(Zpj×Zpj)φZpi is a nonabelian group of order pip2j, so that H×Zn/(pip2j) is a nonabelian group of order n, a contradiction.

() Assume G is a nonabelian group of order n. Inductively, every proper subgroup of G is abelian (since any proper subgroup divides G's order, and any of the factors whose exponents are reduced from 2 to 1 in the process are insured since pq21=(q1)(q+1)pq1), so that by 4.5.56 G is solvable. Let N be the normal subgroup such that G/N is abelian, and in particular N is not trivial due to the nonabelianness of G. If N has a prime decompositional width greater than 1 (i.e. is divisible by two distinct primes), then two distinct Sylow subgroups A and B of N are normal in N due to N's proper-subgroup abelianness, and are thus characteristic in N, leading to them being normal in G. Since G/A and G/B are of smaller order than G and holding to the a priori qualities of G, and thus abelian, we have by 5.4.15 that G/(AB)=G/1 is abelian, so N is of width 1, and of order pi or p2i.

If G is of width 1, then by the abelianness of groups of order p or p2 we know that G is abelian, so assume G is of width greater than 1. By Cauchy, there is a subgroup ˉHˉG (normal since the latter is abelian) with an order of pk - a prime distinct from pi - so that HG is of order pipk or p2ipk. If G=H of order pipk, then G is abelian by previous material. If G=H of order p2ipk, then since pipk1 we have pk1modpi, leading to the Sylow pi-subgroup C being normal so that G=CD for a pk-subgroup D, but Aut(C) has order either pi(pi1)2(pi+1) or pi(pi1) and pkp2i1=(pi1)(pi+1), so we have it is the trivial homomorphism in the semidirect product and G is abelian. Therefore H is a proper subgroup and thus abelian so that its Sylow pk-subgroup Pk is characteristic in H and thus normal in G, so that G/Pk is also abelian and now G/(NPk)=G/1G is abelian.

No comments:

Post a Comment