Basic Algebra (Solutions)
by Huah Chu
Exercises (§1.4, p.42)
1. Let A be a monoid, M(A) the monoid of transformations of A into itself, AL the set of left translations aL, and AR the set of right translations aR. Show that AL (respectively AR) is the centralizer of AR (respectively AL) in M(A) and that AL∩ AR = {cR= cL|c ∈ C}, C the center of A.
Proof. (1) We show that AL = CM (A)(AR). The case of AR = CM (A)(AL) is quite similar.
Since (aRbL)(x) = aR(bx) = (bx)a = b(xa) = bL(xa) = (bLaR)(x), hence AL ⊆ CM (A)(AR). Given any ρ ∈ CM (A)(AR). For all a ∈ M, aRρ = ρaR. In particular, aRρ(1) = ρaR(1), ρ(1)a = ρ(1 · a) = ρ(a). This means that ρ = (ρ(1))L ∈ AL.
(2) AL∧ AR= {cR= cL|c ∈ C}:
It is clearly that cR = cL for c ∈ C. Given any ρ ∈ AL∧ AR, ρ = aL for some a.
Since ρ ∈ AR = C(AL), hence, for all b ∈ M, aLbL(1) = bLaL(1). Thus ab = ba and
a ∈ C. So ρ ∈ {cL|c ∈ C}. ¤
2. Show that if n ≥ 3, then the center of Sn is of order 1.
Proof. Given any 1 6= α ∈ Sn, there exists i such that α(i) 6= 1, say α(i) = j. Choose k 6= i, j since n ≥ 3. Take γ be any permutation in Sn such that γ(i) = i and γ(j) = k.
Then γα(i) = γ(j) = k, and αγ(i) = α(i) = j. Hence γα 6= αγ and α /∈ C(Sn). ¤ Remark. For any α ∈ Sn (n ≥ 3), we can find β ∈ Sn such that it has the same cycle decomposition as α and α 6= β (§1.6). Then there exists γ such that γαγ−1 = β (Ex.
4, §1.6) and γα 6= αγ.
3. Show that any group in which every a satisfies a2 = 1 is abelian. What if a3 = 1 for every a?
Proof. (1) Note that the condition a2 = 1 implies a = a−1 for all a ∈ G. For any a, b ∈ G, since (ab)2 = 1 it follows that ab = (ab)−1. But (ab)−1 = b−1a−1= ba. Hence ab = ba.
(2) If a3 = 1 for all a ∈ G, G need not be abelian. We shall use Sylow’s Theorem and group extensions to construct a counterexample.
1
Let G be a finite nonabelian group such that a3 = 1 for all a ∈ G. By Sylow’s Theorem (§1.13), |G| = 3n. If |G| = 9, G is abelian (§1.12, exercise 6). Hence we assume that |G| = 27.
G contains a normal subgroup K of order 9 (exercise 5.2, p.87 in Rotman: The theory of groups, An introduction). K must be elementary abelian, that is, K = ha, b|a3 = b3 = 1, ab = bai. G/K ' H = hhi is a cyclic group of order 3. Then G is an extension of K by H (see the Remark after exercise 9, §1.12).
h induces an automorphism α of K. If we use the additive notation for composition of K, then K can be regard as a 2-dimensional vector space over finite fieldF3 and α can be represented as a matrix in M2(F3). By suitable changing the basis, assume α has the rational form
· 0 1 a b
¸
(§3, 10). Since α3 = 1, it is not difficult to show that the only solution is α =
· 0 1
−1 −1
¸ .
We have known Ext2Z[H](Z, H) = H2(H, K) and H2(H, K) = KH/NK for finite cyclic group H, where KH = {k ∈ K|αk = k}, NK = {(1 + α + α2)k|k ∈ K}. Hence to find all extensions of K by H, it suffices to compute H2(H, K) first. It is easy to find that αk = k ⇒ k = (x, x) for x ∈ F3 and 1 + α + α2 = 0. Hence NK = 0 and H2(H, K) =Z/3Z.
We first check the trivial case, that is, semi-direct product of H by K. Since
· 0 1 1 −1
¸ · 0 1
−1 −1
¸ · 0 1 1 −1
¸−1
=
· 1 1 0 1
¸
, we change α to
· 1 1 0 1
¸
for simplic- ity. Then G = ha, b, c|a3 = b3 = c3 = 1, ab = ba, ac = ca, bc = cabi.
Now we check that x3 = 1 for all x ∈ G:
First note that a is in the center of G and from bc = cab, we have cb = a2bc, bcb2 = ac and c2bc = ba. Thus
(bc)3 = b(cb)cbc = ba2b(ccbc) = ba2bba = a3b3 = 1.
The verification of (bc2)3 = (b2c)3 = (b2c2)3 = 1 are similar, and left to the reader.
Moreover, since (ax)n= anxn, so (aibjck)3 = a3i(bjck)3 = 1. Hence the result. Thus G
is the desired counter-example. ¤
Remarks: (1) Since H2(H, K) ' Z/3Z, we can find a nontrivial factor set f : H × H → K defined by f (h, h) = ab, f (h2, h) = f (h, h2) = 1, f (h2, h2) = a2b2, and f (1, x) = f (x, 1) = 1 for x ∈ H. Let G be the set of all pairs (k, x) ∈ K × H with the composition
(k, x)(k0, y) = (k(xk0)f (x, y), xy).
Where xk0 is defined as follows: (i) If x = 1 then xk0 = k0; (ii) x = h, k0 = aibj then xk0 = ai+jbj; (iii) if x = h2, k0 = aibj, then xk0 = ai+2jbj. Then
(1, h)3 = (1 · (h1) · ab, h2)(1, h) = (ab, h2)(1, h) = (ab, 1).
2
Thus (1, h) has order 9, which does not satisfy our condition.
(2) The above discussions also hold for any odd prime p. That is, G = ha, b, c|ap = bp = cp = 1, ab = ba, ca = ac, bc = cabi is nonabelian group such that xp = 1 for all x ∈ G.
(3) For the group extensions and cohomology of groups, we refer to J. J. Rotman:
The theory of groups; An introduction; or S. MacLane: Homology.
(4) This exercise may be regard as a special case of Burnside’s problem: Let G be finitely generated and n is the l.c.m of the orders of elements, is G a finite group? This exercise shows that, if n = 2, G is abelian. In fact, if n = 3, G is a finite nilpotent group of class ≤ 3. If n = 4 or 6, G is a finite group. We refer the reader to B. Huppert:
Endlich Gruppen, or M. Hall: The theory of groups, Chap. 18, for more defails.
4. For a given binary composition define a simple product of the sequence of elements a1, a2, . . . , an inductively as either a1u where u is a simple product of a2, . . . , an or as vanwhere v is a simple product of a1, . . . , an−1. Show that any product of ≥ 2relements can be written as a simple product to r elements (which are themselves products).
Proof. We prove it by induction on r. There is nothing to prove for r = 1. Any product of n elements a1, . . . , an, n = 2r, r > 1, has the form of (a1, . . . , ai)(ai+1, . . . , an), where (a1, . . . , ai) is a product of a1, . . . , ai. Then one of the sequence {a1, . . . , ai} and {ai+1, . . . , an} has length ≥ 2r−1, say {a1, . . . , ai}. By the inductive hypothesis, (a1, . . . , ai) is a simple product of r − 1 elements and (a1, . . . , ai)(ai+1, . . . , an) is a
simple product of r elements. ¤
Remark. The condition of ≥ 2r may be refined to ≥ 2r−1+ 1.
The reader may try to give a product of 8 elements which is not a simple product of 4 elements.
3