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

Monday, March 18, 2013

Commutators and Groups of Prime-Cubed Order (5.4.7-9)

Dummit and Foote Abstract Algebra, section 5.4, exercises 7-9:

MathJax TeX Test Page 7. Let p be prime and let P be a non-abelian group of order p3. Prove P=Z(P).

8. Let x,yG commute with [x,y]. Prove that for all nZ+, we have (xy)n=xnyn[y,x]n(n1)/2.

9. Let p be an odd prime, φ be the p-power map φ:xxp, and P be a group of order p3. Prove φ is a homomorphism and im φZ(P). If P is not cyclic, show that |ker φ|{p2,p3}. Is φ a homomorphism in non-abelian groups of order 8? Where is the oddness of p invoked in the above proof? [Use exercise 7.]

Proof: (7) By the class equation we have |Z(P)|1 and by the non-abelianness of P we have |Z(P)|p3. Assuming |Z(P)|=p2 leads to P/Z(P) being cyclic and thus P being abelian, so therefore |Z(P)|=p. By previous investigations and/or similar logic, all groups of order p2 are abelian, so that P/Z(P) is abelian and thus PZ(P). This entails either P is trivial or P=Z(P); the former implies P/1P is abelian, so the latter is the only available possibility.

(8) First note that if x[x,y]x1=[x,y], then (x[x,y]x1)1=[x,y]1, so that by exercise 1 the right-hand side is [y,x] and the left-hand side is (x[x,y]x1)1=x[x,y]1x1=x[y,x]x1, so that x,y in this exercise might be assumed to commute with [y,x] as well.

Proceed by induction. The basis case (xy)1=xy=x1y1[y,x](1)(0)/2 holds true, so assume (xy)k=xkyk[y,x]k(k1)/2 for all k<n. We have (xy)n=(xy)n1xy=xn1yn1[y,x](n1)(n2)/2xy=xn1yn1xy[y,x](n1)(n2)/2=xn1yn1xy[x,y]n1[y,x]n1[y,x](n1)(n2)/2=xn1yn1xy[x,y]n1[y,x]n(n1)/2. This implies that the truth of the original proposition hinges on the validity of the new proposition, xn1yn1xy[x,y]n1=xnyn. Rework this to xyn=yn1xy[x,y]n1, and proceed with further induction. The basis case with n=1 clearly holds, so apply the inductive hypothesis and finish the problem: xyn=xyn1y=yn2xy[x,y]n2y=yn2xy[y,x][x,y]n1y=yn2xy[y,x]y[x,y]n1=yn1xy[x,y]n1.

(9) Assume P is abelian. Then φ(xy)=(xy)p=xpyp=φ(x)φ(y) is a homomorphism and since Z(P)=P we clearly have its image contained therein. Therefore assume P is non-abelian. Since by exercise 7, we have P=Z(P), the setting verifies the criterion for exercise 8 that any x,yP commute with [x,y]P. Since p is odd, we have 2p1, so that pp(p1)/2. This implies [y,x]p(p1)/2=1 since P is a group of order p, and thus we have the equation from the previous exercise collapsing to (xy)p=xpyp, and φ is thus a homomorphism.

Assume φ(x)Z(P) for some xP. This implies ˉxp1 in P/Z(P), and since |P/Z(P)|=p2, that ˉx generates this quotient group, so that it is cyclic, and P is abelian by the familiar process. Therefore, im φZ(P).

By Cauchy's theorem we have x of order p so that the kernel is not the identity. If the kernel is of order p, then the image is of order p2. Since the image being cyclic would imply |yp|=p2 so |y|=p3 for some y, we have the image and thus P contains p21 elements of order p, and thus the kernel's order is greater than p.

As for the oddness of p, it was invoked in collapsing the formula in exercise 8. It does indeed map the elements of the only two non-abelian groups of order 8 (D8 and Q8) to their respective centers, but it is not a homomorphism on either, as φ(s)=1 and φ(r)=r2, but φ(rs)=1r2, and φ(i)=φ(j)=1, but φ(ij)=11.

The proof to exercise 8 might not be the shortest, though it was not a random long shot. The first induction simplified the problem by replicating the tail of what it ultimately was aiming for and thus appealed to a simpler inductive problem, in a way exhibiting an inductive process itself.

No comments:

Post a Comment