STUDENT’S SOLUTION MANUAT ° Abstract Algebra - IN. Herstein University of Chicago ing Company Collier Macmillan Publish@rs LondoPreface {As | stressed in ho book Abstract Algabra itis vory important that you try to salve many of the problems. fs nice to solve a given problem, but itis more important fo make an attempt to solve it “There is no better way for you fo test how much conirl you have over the material. tis too much to expect that you will ty, or solve, al, or most, ofthe large number of problems appearing in tho ‘book. The chiet thing is that you take a stab at a good eross-saction of thom, In tis addendum to the book you wil ind solutions io many of the problems. Roughyy one-all of the so-called Easier Problems are solved hora; a slighty highor tracton ot tho Middlo-Lovel Problems also have their solution gwen. Tho coluion is given fora rathor largo fraction of the Harder Problems of Very Hard Probloms, Thoro is no systomatic way-for instanco, every second ‘one—in which I chose thase exereiges for which I giva solutions. The choice was made on the basis of how instructive of usolul a given solution would be for the student who tempted a given problom and di not succood in coving it Even if you do soe a particular ono, it might bo of interst to you to compare your approach and mine to that exercise. There fs a great tamptaton to peek at the solution ater an inal attempt ata problem tals, Don't ‘glo into this tamptavon. Only after you have trough about a problem and approached trom a ‘variety of anqles, and stif didn’t g2t Kk, should you look up the solution inthis booklet. To give in and [peok prematuraly is Sol-dofoating for takes away trom both the challenge and learning ‘experiance that You could have i thinking through the mater on your own. ‘What | have wrtten above applies particulary to those problems designatad as Harder Problems ‘of Very Hard Problems, These problems, in general, were designed to challenge the would-be ‘soe! tothe utmost So don tbe discouraged I you lal to solve a great many of them. The fun you rive should be in the attempt, and not necessarly inthe success, of grappling with something cicutt ‘The problems are occasionally not rated as to dficlty. This is deliberate on my part The ability to.quess or 10 xige the cificulty of a situation can be an essential part ofthe learning process. Solutions fo some problems depend on those of other ones. There is often some cross-reference Indicated inthe statement ofthe problem. Some problems come a litle before the required material needed to handle them has been ciscussed. This serves two purposes. First it gives you a chance to develop some ofthe neecied notions on your own. Sacond, it lends fo show you that this material hhas some bite fit and that it allows you ta de things which you were unable to do without it ‘Atany rate, Ihope you have lots of fun with the problems, that they challenge you encugh, and that they doa't eustrate you too often InkContents CHAPTER 1: Things Familiar and Less Familiar CHAPTER 2: Grouse .... CHAPTER 2: The Symmevic Group CHAPTER 4: Ring Theory 25 6s, 701 Things Familiar and Less Familiar SECTION 2. Easier Problems. 3. AnBaC is the set of all women residents of the United States who are Canadian citizens, A - 8 is the set of all non-Canadian residents of the United States, A-C is the set of all male residents of the United States, and C - A is the set of all the women in the world not residing in the ‘United States, 4 a-9 6 IfACBand ifue Au thenue Aoruec; fue A thenueB since ACB, hence u is certainly inB uC. On the other hang, if ue C then u € BUC, Thus, in either case, ue Bu, whence AUC C BuC. B. if uisin(A-B)u(B~A) then u ts certainly inAu6; ifuisinA-8 then u is got in B so is not in An B hence w ts in (Au B) - (AaB). Similarly, if u is in - Athen u 1s in (AUB) - (AaB) Therefore (A - 8) u (B ~ A) C(A UB) - (ANB). On the other hand, if ue (Au B) - (A 9B) then ue Au 8 andu is not in both A and B ; thus If u is in A then wis gotin Bhence u is in A - B Similarly, if u is in B then u is not in 8 - A Thus in all cases u is in (A - B) u (B - A), whence (Au 8) - (Aa 8) C(A-B)u (B-A). We thus get the required equality of these two sets, 10. Since A u(B aC) Is contained in both A uB and AuC we get that (AUB) a (AUC) 2 Au (BAC). For the other way, ifue(AuB)a(AuC), then u is in both AuB and AUC and, if isnot in A, then ue Bandu eC so 12./ Student's Solutions Manual that ue BAC, and so.u€ Av (B nC). If u should happen to be in A then certainly u Is in Au (BC). So we get that If u ts in(A UB) n (AUC) then u is in Au(B nC), hence (Au 8) aCAUC)CAU(BAC) This proves the desired equality of the sets 11, The subsets of S = (1,2,3,4) are 16 in number. They are: S, (1),(2),(3), (4), (1,201,391, 41,(2,3),(2,4),(3,4),(1,2,5),1,2,4),(1,3,4),(2,3,4), and the empty set Middle-Level Problems. 12. (@). If ue (An BY then u cannot be in both A and B. If wis not in A then UIs in A’ and if u is not in B then u is in B’, hence u ¢ A’ UB" Thus (An By CA’ UB’ Since AD AnB, AC (AnBY; similarly BC (AN BY, hence A’ vB’ (An BY’ This proves that (An BY = A UB (b) For any subset C of S, (C= C from the very def inition of C Thus, by Part (a), (Cu DY = (CY 0 (0 =CaD Let A=C’ andB =D"; we get (AuB)’=(CY nO) =A" 13.(@), AB =(A-8)0(B-A)=(8-A)U(A~B) =8 +A tb), ArO=(A-0)U(0-A)=AUO=A (AAS ANAZA (). Ar As(A~A)U(A~A) #0 (e). To show that A+ (BC) = (A+ 8) + C we first prove the Theorem. A+ (BC) = (AuBUC) ~((An BUCA n Chu(B n Cd] - (An BAC) Proof, LetO = ((AnB)u(AnC) u(8 aC) - (AM BAC) and EXAUBUC)-D If UeE and if uisnot in A then ue BUC. We claim that if uis inBaC then uc A+B + C.For, I UeBAC, then, since u ts not in A, so not in ‘ANB AC, u must de In D, contradicting that it is in €. Thus wis not in B aC(CHAPTER 1: Things Familiar and Less Famitiar / 3 and being in B uC we have that ue B + C, hence in A + (B + C). On the other hang, if'we A then u then, if u isnot inB+C thenu€ A - (B+ C) Ax(BeC) Hence ue A+ (B+ C) and so CA + (B+ C) To finish the proof we show that A+ (B+ C)CE If xe A+ (8 + C) andif x is not in A then x 15 1nB + C; therefore x 1s inB uC but not inB aC, thus x is not in D, $0 must be in If x 1S INA then x iS not InB + C; if x is not inBuC then x 1S not in D hence must be in E.1f, on the other hand, x is in BUC then since it is inB + Cit isnot inBAC so isnot in ABn C thus not in D , hence must be in E. This finishes the proof. To finish Part (e) we note that (A +B) * € *C + (A+B), andusing the theorem above, C + (A +B) = Cu AUB - {{(CnA)u(CAB)U(AnB)| - (A nBaC)) = Av(B+C) (IPAS BEAYC then a+ (A+ B)= A+ (A+ C), = by Part (e) we get (A+ A) +B=(A+ Al © C. From Parts (d) and (b) we ge" that B = C (A (B+C)= AaB -C)UNC -B)) =(An(6 -C) NU (AAC -B)) by the result of Problem 9. The analagous proof to that ef Problem 9 snows that AngB = C)= (AaB) -(AnC), Thus, from the above, ACB C)=((AnB) ~(An cu (ANC) ~(An Bl =AnB + Aac =AB + AC 15. Let D= BUC; then mA UB UC) =m(AUD)= mA) + m(D) - m(AAaD) = m(A) + m(B) + m(C) - mB nC) ~ mA 0B) (An C)) = mA) » mB) + mC) ~ M(B AC) - (mA 9B) + m(A AC) - mMANB NC) =m(Ay * miB) + m(C) - MA NB) -mM(AAC)-miBAC)+mANBAC) 16 Toobtainm(A, uv . wA,) we form the sum of the m(A,) for f= 1,0 and subtract from this the sum of the m’s of all the intersections of every m - 1 distinct Aj's plus the sum of.the m's of all tne :atersections of4/ Student's Solutions Manual every m - 2 distinct Ay’s and so on. (The formal proof can be carried out after we have studied induction in Section 6.) 17. Let Abe the set of Americans that have gone to high-school and let B be the set of all Americans that read a daily newspaper. Then AuB ' consists of at,most all, Americans. If there are n Americans then 12 m(A vB) = m¢A) + m(B) - m(A mB) = Bn + 7m - m(AnB) from the data given. Hence m(A a) > 8 *.7n-n= Sn, thus at least SOX of all iy ‘Americans are In An B, that Is, nave gone to high-school and read a datly newspaper. 19 Y& Arguing as in the solution to Problem 17 we easily obtain that the set of veterans who lost an eye, an ear, and an arm is at least 25% of the group Of disabled veterans, Thus those who, In addition, lost a leg is at least 85 +25 - 100 = 10 percent 21 GBY@ Lets =fa),..., ag) and let Abe a subset of 5, We assign to A a binary number as follows: if a € A the 1" digit of this number is 1 and if 1 ay is not in A the i” digit is 0. Thus the subset {@),a5 24) has the binary Number 10110 assigned to It, Shas 11111 assigned to it, the empty set has 00000 assigned to It, and soon. On the other hand, given a 5S-digit binary number It 1s the number assigned to some subset of S. For instance, 01101 would come from the subset (a,,a5,a5). Since the number of S-digit binary numbers is 32 there are 52 subsets in S. (Of course we could solve s the problem by enumerating all 32 subsets of S) (D). Since there Is only one subset of S having 4 elements which does Not contain a given aj and there are 5 a,'s we have exactly five 4-element subsets of SCHAPTER 1: Things Familiar and Less Familiar / 5 (c). The number of subsets of S having 2 elements Is the number of ways we can pick 2 ‘objects out of a set of 5 objects without regard to order. This number is 10 = 5.4/2 22. (a). Arguing as above in the solution to Problem 21, by assigning an n-binary number to each subset of S we see that Shas 2” subsets. (b), The number of subsets of S having exactly m elements is equal to the number of ways of picking m objects, without regard to order, out of a set on n objects. This number Is ni/(mi(n-m)). A formal proof of this can be more readily given once we have studied Induction SECTION 3. Easier Problems. 1. (a). fig not a mapping since not every woman has a husband (b). C13 not a mapping of S into T for f(1) = 0 and 0 ts not in T. (c). f is. a mapping from the positive integers to the nonnegative ones. (a). 1 1s not a mapping of the nonnegative integers into themselves for £(Q) = -1 which ts not in T. (e), 1s amapping on the set of all integers. (1), f lgnot a mapping from the reals into themselves since Vi Is not areal number. (q). f 13 amapping from the positive real numbers into themselves because every positive real number has a positive square root 3. If isa 1-1 mapping of S onto T then (7! defines a mapping from T to § since, given t in T, there is one and only one element s in’ such that t= f(s), ands 13 defined to be f7!(t}. That £7! 13 onto and 1-1 follows from the fact that { 18 a 1-1 mapping from S onto T, that it is onto 15 a consequence of the fact that f is defined on all of S so that every element6 / Student's Solutions Manual $ In$ has image t= f(s) in T, nence s= 17 !(t) It 1s 1-1 because, If clap =O Nty) then ty = (710g) = ty = ty S. Given u in U tnenu = f(t) for some t in T since f maps T onto U; because f maps $ onto T, the t above 1s of the form t = f(s) for some s in S. Thus u=f(t) = f(g(s)) = (1 gis), whence { g Is a mapping of S onto U, 6 If tisin T then t = (S) for some s in S since f ts onto, Thus g(t) = {U(S)) = (g Ms) =n XS) = RECS) = NUED; therefore q = h 8. (a), { defines a function from S to T. (0). Since even + even 1s even, oad + odd Is even, and even + odd is odd we easily verify that f(s; + $9) = 1) (So) (c) By the definition of f, (3) = -1 and ((2) = 1, yet 1 = 116) = f(3)f(2) = -1, hence f(s 82) 1s, tn general, not equal to f(s f(s). fis* N=(s6 12 9 (a). fs) +s andgls) = $+ 1 thus (f gis) = (gts (b). (g Ms) = git(s)) = g(s2) =s2 +1 (€)._ from (a) and (o} we see that f g* fF 10. (. (fay fe ghs)= (pcs * = ales +a) b= acs+ ad + b= Facadep(s), Renee fay fe.d * fac,adeo (b). From the result of (a) we know that {3,5 fc,q* fac,adep and Tod Fab * fea,ched » thus these two products of mappings are equal if and only if ad + b= cb +d. This does not hold, for instance,if a=b=d= 1 and c-2 (). Ife= = 1 then the result of (b) shows that for fy fy 1 tobe equal tof, ; fap) we must have a+b = b+d, that ts,a=d. So the onlyCHAPTER 1: Things Familiar and Less Familiar / 7 ‘ones satisfying the condition are all the f, (A). Suppose that fy °F p= 4,9. the Identity mapping. By the formula of Part (a), xa= | and xb + y= 0. Thus x= a7! andy = -a”'b are solutions . Furthermore, for this x andy, fa. fy * faxay + p and since ax= 1 and ay + b= a(-a”!b) + b= Owe See that {5 »: 1,0, Thus ip {5 the required inverse for fp Middle-Level Problems. 12. (a). the f So defined Is not a mapping, since, for instance, 1(1/2) = 2!52 wnmte (01/2) = 1(2/4) = 2234 = 2!32, so we cannot assign a unique value to (1/2), nence f is not a mapping. (0). given a positive rational number r we can write ras m/n where mand n are positive integers having.na common factor Then f defined on 5 to T by f(r) = 23" does define a legitimate function (mapping). 13. To show that f defines a function we must show that to every s Ins, f assigns a unique element of T. 1f 23" = 235° where m,n.a\b are non - negative integers, then nb and 2-4 = 30° ang since the left side of this is even white the a andn =b. For suppose m > a, say: then right one Is odd, this cannot nappen. So every element in Shas a unigue representation in the form 23" and the rule (23%) = m/n is thus a map- ping from S to T. 15, Let =f where a *0 and b are integers. Applying the formula fap’ fap * '22,aben We get that foFef =Ta.y°la2 anety * fa3,a2beapebi IN12 / Student's Solutions Manual (b) “ Xp x3 ‘sy T XQ X3 “4 x Ky Rats Aq 2K X3 Xa (fg 7x) xQ%x3% yy (B_afymy xD xg XG leckeist tna (0). A similar computation to the ones above shows that (fg)> = ig, ang n= is (A. From (c} and (a) we see that (£gXx}) = xz while (gtXx,) = x), hence fg = of 11. See the solution for Sp for any n given for Problems 12 and 13 Middle-Level Problems. 12. For the solution see that of Problem 31 in Secticn 2 which shows that, givens in'S and nS, there 1s a positive integer k, depending on s, such that fK(s) «5. If m is the product of the k's for all the elements of S then M(t) = t for all t in. Thus f™ =| V4 Let S = (x), xp, J and T= (y), ¥9,-.¥ql where m ¢n, Define F 182 V+ V2 ti FFOM Spy LO Sp DY: IF F € Spy and f(x) = xj (| depending on 1) tet FUN) be in Sp defined by FONY,) = yj for 1 C1 m. Clearly F ts I-1 ang from Its very definition F(1g) = FEC). 15, If Shas 3 or more elements define f. 54 5 by (54) = Sp, ((S9) = 8, and f(s) = § for all the other s in S. Define gS 3S by 4(51) = 83, g(s3)= 5) and g(s)=s for all the other s in S. Thus (19X54) = f(q(34))= 1135) = 53, a FS aCHAPTER 1: Things Famifar and Less Familiar / 13 while (gfS1) = g(1(s1)) = (So) = 8). Thus fg * gf. 16. (a). Suppose that f € 1 moves only 5), 52... Sp, leaving the other elements of S fixed, and g €M moves only ty, ta, ty, leaving the other elements of S fixed. Then, since (fg)(s) = f(g(s)), we see that fg leaves all elements of S-fixed with the possible exception of 9}, .., Smity, ty, ty Thus f(s) # s for at most a finite number of s, hence, by the definition of MH fgisint (0). If f eM and f(s) = for all s other than $4, $9... Spq then f(s) = for all s other than $),59,.. , Sip,hence fem, 18 (2), Since UT) Is the set of all {In AS) which take T Into Itself, and i(t) = t for every t in T, clearly 1 takes T into itself Thus i UT), (b). If f.g €UCT) then g takes T into itself, so if t eT then g(t) eT, hence (fg)t) = f(git)) 1s in T since f takes T into itse'’ Thus fq € U(T). Suppose that the elements of T are ty... tyy and {(ty) = ty where ty is in. Let the mapping F be defined by F(TKS)) = $j, 1 ch¢m.. FC) Hes In Sp) and, from its definition, maps U(T) onto Sm, for Ife € Spy and a(5)) = 5 for 1 Owe praceed by induction on n. If n= 1 then certainly (a*p)! = @*b. Suppose that for some m we know that (a¥b)m = ab! then Campy"! = Campyn(ampy™ = (ab)(a™D™) = ancde(a™en™)) =arccona™ yo) -ax((aanyed™) = axta™a(pe™) = Cana )*(beD™) = a™*1ay*!. naving made use of the fact that G ts abelian and * is associative. This completes the induction and proves the result for all positive integers n By definition a° = e for all a inG so that (a*b)° =e = e*e = a%b0 Finally, if n< 0 then n = -m where m> O anda = (a7!) =ar!mpe! so (amp)? = (arp)! = (at bey- FM = (a7 M*(o7 HY (by the result we proved form > 0) = a™p” since G 1s abelian, (ab) In future calculations we shall not be as formal as above and will use the associative law freely, and avoid these long chains of equalities,CHAPTER 2: Groups / 27 9. Suppose a2 = e for every a in the group G. If a,b are in G then (akb)2 = e, thus a*b*a*b = e; multiply both sides of this relation by a to obtain a2sp*a*p = a, and since a2 = e, b*a*D = a Multiply both sides of this relation on the left by D to obtain b2a¥ = ba, and since D2 = e, we end up with a*p = b¥a, Thus G fs abelian 11, Since the * in the example is just the composition of mappings, for ease of notation we drop it and write the product a*b simply as ab. We are considering the elements f'n! where (2 = n3 =e and th = nv!r. 2f; so tht = h'r for all integers t, and htr= tht. thus ceindycinty = clini = Ant) and (tini cen’ = rind; these two results: Thus fh? =n“! fh = can be succinctly written as (t'hJy(¢Sn¢) = fp? where a= i+ j and b= t+ (-1)5}, Thus 6 is closed under the product of mappings. Since 2+ (2h3, eis ing. Also, (fh)! «le! = eh) so 's in G. (Don't forget, the exponent of f is calculated mod 2 and that of h is “nod 5.) Finally,since we are talking about the product of mappings, the product 1s associative. Thus G is a group. That G is of order 6 is easy since, in f'n), 1 nas 2 pessibilities and j has 3, and these give rise to 6 distinct elements (Check it) That G is non-abelian is clear since fh = nf 13. Suppose that G has 4 elements; let e, a, and b be 5 distinct elements of G, Thus both ab and b*a are in G, if they are not equal then b*a ~ e, a, or b. If akb = e then we quickly get b¥2 =e and so ab = b*a. If ab «a then b =e and if atb= b then a =e (see the next problem for this), both of which are contradictions. So we get that a*b = b*a and G consists of e, a,b, and a*b. To check that G 1s abelian one should also check that a(a*b) -(a*b)a and b(a*b) = (a*b)b, we leave these to the reader. 14 See the proof of Lemma 22.228 / Student's Solutions Manual 16. Since a2» e for every ain G, by the definition of a~! we have that a*a”!. Thus, if a and bare inG then atb » (a*p)"! =p" Ia! « bxa, hence G Is abelian. 16. Suppose that 6 13a finite group of even order; 1f a= a”! for every a inG other than e, since a= (a~!)”! we get an even number of elements which are not e, together with e this would give G an odd number of elements, contrary to our assumption Middle-Level Problems. 21. If G is of order S, then for every element a inG there Is a least positive integer k, depending on a, such that a* = e if k = 5 then G must consist merely of e, 2, a2, a%, and a, so is abelian. If k= 4andb inG is not a power of a then a a! any i soe, a, a2, a3, and aX exhaust 6; now b*a is in G and is not a power of a, for if b¥a = a! we immediately get that ball, a contradiction Thus bMa is forced to equal 2*b, and so we see that G is abelian. If k = 3, then e, 2, a? are already 3 distinct elements of G. Let b in G not be a power of a, then, as above, b = a*b, so the elements of Gare e, a, 2%, b, and aXb. What can b*a possibly be? As above we quickly arrive at a*b = b¥a. So we are left with the only possiblity, namely that every a in G satisfies a2 = e, By the result of Problem 9, G must be abelian, In actual fact, as we shall see in Section 4, the first case, k =5, is the only possibility if G has order 5. 24 G is generated by 2 elements f and h satistying {2 = n= e and th = nly, and all the elements of 6 are of the unique form tnd where 1 = 0 or 1 and } canbe any integer 0 0. Therefore a to some positive power falls inH. Thus there 1s a smallest positive power k such that a is in H Suppose that y ts in H; then y = a" for some m. By the Euclidean Algorithm, m = gk * r where O £7 0. Therefore r= 0, hence m = qk. This shows that y = (at-9; thus H is a cyclic group with generator ak 14 Suppose that G has no proper subgroups. Let a = © be in G and consider H= (al {any integer). As is immediate, if x and y are in H then xy and x"! are In Thus H ts a subgroup of G, and H = (e) since 2 = e is inH. Thus, by hypothesis, H = G. Thus G is cyclic with a as generator Middle-Level Problems. 16. By the result of Problem 14 any element of G other than e generates G. If ais inG and a2 = e then the group generated by a has 2 elements, that is, Ghas 2 elements, If a2 e then every element of G is a power of a2: in particular, a= (22) = 22 nence 22M"! = e, for some integer m. Since 2m- 1 =O one of 2m-1>Oor! -2m>O At any rate we get that a? =e for some smallest positive integer p. Thus 6 consists of the Pdistinct elements e, a, a2, .., a?! We claim that p is a prime. If p = uv where both u> Vand v> 1 then if b= 24 e the subgroup T generated by b consists of rCHAPTER 2: Groups / 33 the v elements e, b, D2, .. oY"! since bY = aX = a? =e But T =6, sov=p since v is the order of T, and 1s that of G But then u= I, contrary to u> I. Thus p is a prime and [ol ~ p 18. Since S is finite A(S) is a finite group. If f and g are in T(x) then £00 CX and g(X) ¢ x; hence (Fg)(X) = (Q(X) © FX) CX and so fg is in TO). Since A(S) is 2 Finite group and TCX) Is closed under the product of A(S), T(X) is a subgroup of ACS) 19, Suppose that x= a)b, andy = agbo are In AB, where a),.a) are in A and b), by are inB, Since G Is abelian, xy = ajbyagbg = ajaqb)bz € AB, and 1 x7 T+ (ayp))"! =p) "!ay7! =a, 'b, “| AB. Thus AB is a subgroup of 6. 22 and 23. We saw in Problem 19 that AB is a subgroup of G. How can AB -[ab a. € A, 0 €B) fail to have mn distinct elements? Only if there is some collapsing of these elements, that is, if and onty if for some distinct pairs (2),04), (agba) we have a,b; = ajbo. But this implies that ap7!ay = bgb,7! and since the left side ts in A and the right side is in 6, the elements on both sides are in AB, Thus ay =ay¢ and b, = ¢”'by where cis inAn6.But, ifceAnBand if aeA,be6 then a, = ac Is inAand by =7'b is inBand.ayn, = (anc! ‘ab, Thus there are exactly 1A n Bl pairs giving rise to the same ab, Thus [ABI ~ mn/IA n él = LAIIBI/IA 9 Bl. This solves Problem 23. However, to solve the present problem, we must show 1 This is most easily done after we learn Lagrange’s Theorem. Note that the argument used for [ABI did not depend on G being abelian . This will be used ‘many times in the problems in the rest of this chapter. 24 That N is 2 subgroup follows because the intersection of any number that if JAl andl are relatively prime then AaB = (e] so that |Aa 894 / Student's Solutions Manual of subgroups of G is a subgroup of G (This is a slight generalization of the result in Problem 1; the praof we gave there works in this more general situation). If x,y 6 then y~! (x !Hxdy = (xy) !HCxy) and as x runs over 6 with y fixed then xy runs over all the elements of 6. Thus y~Yin (x !rixdhy = (xy) 'HOxy), as x runs over G, =n x" !Hx, as x runs over G by the remark above Thus y~!Ny = N. Harder Problems. 25. Let Se the set of all the integers and let x be the set of positive Integers. Let f be the 1-1 mapping of $ onto itself def ined by fin) =n + 1 for every integer n, Then certainly f(X) ¢ X, nence f e T(x) but (7! Is not tn TOO since f-1(1) = 0, which is not inX. Thus T(X) cannot be a subgroup of AGS) 26. See the proof of Lagrange’s Theorem (Theorem 2 42) in the next ‘section. 28. We first check that MN Is a suDgroup of G.If mn, myn; are inPN, where m,m) are inMand n,n, are in N then ¢mn}(m ny) = mamn” nny and by the hypothesis on M, nmn”! is int thus mnmn™! is inM, and pn; is INN. Therefore (mn\(mynj) is in MN, hence MN is closed under the product. inG Also (may! = 9“!m=! = 9" hm“ !ndn! so is in MN since n= !m~!n is in Mandi”! is InN Thus MN is a subgoup of 6. We xeG then x° | (etx = Oo !Mx)07 Nx) © MN by our hypothesis on M and N 30. Consider the element a = mnm='n7!; bracketing it one way, a= (mm 'n-! so is inN since mam”! and n”! are in N.On the other hand, bracketing it another way, a= minm™!n~!) so is in M since m and am7'n7! aCHAPTER 2: Groups / 36 are inM. Thus a€MaN-~(e). This tells us that e= a= mam !n7!, which implies that mn = nm, SECTION 4. Easier Problems. 2. The relation ~ defined on R by a~b if both a> b and b > a satisties the symmetry and transitivity properties but fails to satisfy a ~ a 3. The argument starts with “if a~b . *, however there may be no element b which satisfies this, as is exemplified in Problem 2. However, if we insist that for every a there is some b such that a ~b then the argument is valid and ~ is then an equivalence relation. 4 Suppose that S Is the union of the mutuaily-dis aint, non-empty subsets 5. Thus, given s in S there is one and only one S,. such that s ¢ Sq. 0 if we define a ~b if a andb lie in the same S,. then we easily see that ~ is an equivalence relation and the equivalence class of s is precisely that Sq, in which s Nes. 8. Suppose every left coset of H in G is aright coset of H in G. Thus, if a is in G, Ha must be a right coset of H in G, thus Ha = bH for some b in G. But ais in Ha so a must be in bH, because a is in aH and, by Problem 5, the right cosets are equivalence classes, we have that DH = aH Thus Ha = aH, hence H= aa! 11. Let M be the set of all left cosets of H in G and N the set of all right cosets of H in G. Define the mapping F from M to N by F(Ha) = a”! This ts clearly a mapping of M onto N because, given the right coset xH, then36 / Student's Solutions Manual xH = F(X), Is F 1-17 Yes, because if F(Ha) = F(Hb) then a”! = bl, and SoH ab”'H, this puts ab“! in H whence Ha = Hb. So, even for infinite Groups there is a 1-1 correspondence of M onto N; for finite groups this translates into: Ml and N have the same number of elements. Thus there are the same number of left cosets of H in G as there are right cosets of H inG 12. The answer is no, For instance if 6 = Sx and H is the subgroup in Problem 6, then Hg = (g, fg) and Hg = (fg, 12g = g) = Hg, while oH = (9, gf) and Fg = (fg, fg =g]. Because fg * gf we see that fgH = gH yet Hig = Hg 13. The elements of Ug are ((1), (51, (71, (111, (131, (170. The orders of these are: o({1}) = 1, of{S)) =6, of7)) = 3, o((1 1) = 6, off 13) = 3, offt7)) = 2 We verify one of them, namely that o({7]) = 3; the otner verifications are similar. (7]!» (7), (712 = (49] = (13, (7 = [71031 = (911 = (1) Thus the order of [7] is 3 since that is the first positive power of [7] which {s the identity element of Ujg. The group is cyclic since o({S}) = 6, so the powers of [5] sweep out all of Usg, 1S. If x? we 1 mod p then pl (x2 = 1)= (x 1x + 1). Since p is a prime this tells us that either p I(x - 1) or pI (x + 1) The first of these yields that x @ | mod p and the second yields x a -1 mod p. 16. For every a inG there is an inverse a”! in G, 1 aa”! then in the Product 2) 35..dp, a cancels against a”! since G is abelian. Thus the only terms remaining uncancelled in ay...dp are those elements of G which are their own inverses Since each such element has square equal to e, we get, again using that G is abelian, that ()a).a,)2 =eCHAPTER 2: Groups / 37 20. Recall the basic multiplication rule: T.5T.¢,q = Tacjaden from which - -1 -1 we saw that Tog cola THUS Teg TapTog*To,d 'Tacadeb * T, nc laTacjadeb * Tajc"Madeb) - eNa 7 Ta where x = 7 1dCa- 19D) ; by choosing appropriate appropriate c and d we can realize any x in the above form provided that not both a= | andb #0. Thus, if Tay *Ty,9 the identity map, then the conjugacy class of Ty * (Ta x1 all x! 21. The dinearal group of order 8 is the group generated by f and h where 12=n4=eanam = n7!t (nf), A computation shows that there are 4 conjugacy classes, namely cl(e) = (e), cl(h) = (hn, cin) = tn), cP) = (f,n2f1, and ci(tn) = (thn 24 If p is a prime of the form 4n + 3 then Up is a group of order p - | = 4n + 250 its order is not divisible by 4. However, If 22 = -1 mod then [al has order 4 in Up, which would force 4 to divide [U,l, 2 contradiction. So there 1s no such a. Middle-Level Problems. 27. Suppose that aH = DH forces Ha = Hb; but if h is in H then aH = ahH, thus Ha = Hah, and so H = Hana”!, that is, aha”! €H for all heH andall ae. Thus aa”! CH for all a inG, by Problem 29 of Section 3 we get that atta”! =H for all a in 6. 28 Let G be acyclic group of order n and a a generator of G. When is b= al also a generator of G, that is, when is b of order n? If (in) =d= 1 then p= (aly/4~ e since di. On the other hand, if (i,n)= 1 then if bk = alk «© then n| ik (see Problem 31 below), because (i,0) = 1 we must38 / Student's Solutions Manual have that nk, hence kn, and 0k =n. Thus a! has order n if and only if i and n are relatively prime (and 0 ¢ 1 <1); thus the number of generators of G equals the number of positive integers less than n and relatively prime ton, that is, @(n), 29. Suppose that aba”! =p! Then a2ba~2 = ataba!)a7! = ablan! « (aba™!)! = (b= y™ where m = 12 Continue in this Way, or use induction, to prove that a°ba"" = bK where k = i. 32 Suppose that o(a) = qf(a) + r where O then @(a) = e forall a in G, hence in this case @ is neither 1-1 nor onto. If Gis acyclic group of order 3 andn = 2 then, as 1s easily checked, » is both |-I and onto, Toor Wy) ; therefore (~! ts an isomorphism of Gz onto Gy, thus Gy = 6). 3. (2), Givenx in then x = (xala!, so x = Lala), hence Lg 1s onto Moreover, if L(x) = La(y) then xa~! = ya"! so that x=y. Thus La is 1-1 Therefore Ly € ACG) (b). IF x is In G then (LalpXx) = Lally (0 = Latxd™!) = (xd ar x007la7!) = x(aby"! = Lay (ud; thus Ly = Laly (c). Wab) = ab * Lal = Wla)ulb), by Part(b). If wa) = w(b) then, Llp hence a”!=Lg(e) = Ly(e) = b“!, thus a = b. So w ts a monomorphism of G into A(G). 5. Suppose that V €G satisfies VT, = T3V for all a inG Let c= Vie); then (VT 3e) = VT (e)) = V(x) for all x in G. Also (T,VXe) = T,(v(e)) =CHAPTER 2:.Groups / 43 Tc) = xc. Since VTy = TV for all x in G we get that V(x) = xe = Ly(x) where d="! Thus V=Ly 6. Let 9(G) be the image of G in G; if x, y are in @(G) then x = (a) and y= @(d) for some a, b in G, hence xy = @a)e(b) = @(ab) so is in @(G), as Is olay! 8 Define f from G to G' by f(a) = 23; then fa + b) = 23° = 2220 = Hayri) so that f is a homomorphism of G into G’. It is 1-1 because 2? = 2b implies (a7!) in @(G). Thus @(G) is a subgroup of G. that a = b. Finally, if ¢ = log. (a) then f(c) = 2© = a; therefore f is onto G’. 10. Computing, fg! = tge= gtr ="! = 93, ctghigttgly”! = tglgg ty! = tf! = 95, and glgg™! = g are all nH So aga”! is in H for all a inG; thus (aga“!) = agla"! is also In H for every i. Hence H is 2 normal subgroup of G. 13. We already know that @ is a homomorphism of G into itself by (e) of Problem 1. The kernel of ¢ is the set of all the elements a in G such that a™ =e. Thus this kernel consists of e alone only if a =e forces a=e. This happens if and only if m and n are relatively prime 16, This problem occurred as Problem 28 in Section 3; see the solution there. 21. Lets, t, and v be 3 distinct elements of S. There exists an ¢ in A(S) ‘such that f(s) = and {(t) =v; also there exists ag in A(S) such that Q(s) = t. Thus (g™! rgXs) = g”!cteg(s))) g°'(t) =, Thus, although f is inH(S), gg is not in H(S); thus H(S) cannot be normal in G. yHrcen = g” lew) = s because 24 (a) Gis clearly closed under the product. Also fey,e9), where e, is the unit element of Gand e, that of Go ts the unit element of G since (gy.92Kej.€5) = (944.9262) = (Qy.92), and, simitarly (e),e919,99) *44 / Student's Solutions Manual (9},9). A similar verification shows that (q,~',g97!) acts as the inverse 0f (9,92). The associative law easily checks out as a consequence of the fact that the associative law holdsin G, and Go, Thus G fs a group (b). The mapping @) defined by (a)) = (24,9) is 1-1. Also #4(a)29) = (319,09) = (@),e2Kaz.e9) = @ (a; 94 (ap), nence @} is a homomorphism, thus 1s a monomorphism, (©). Trivially the similar argument works for Gy. (4), Given (ay,ay) in G then (2,89) = (2),¢ Key a9), and (24,€9) 18 in #1(G 1) and (e429) Is in 026G9). If (@.29) 18 In @ (6) n @o(Go) then aj = @; and ay = ep, so this intersection consists of the identity element org, Middle-Level Problems. 26. (a) if a,b are in6 then, for all g in G, (ogopXg) = 6,(dg07!) ratogo" "a7! = (abig(ab)”! = o.y(9) Therefore ogy = gop, whence (ad) = ay = Fady = Wa)W0), so is ahomomorphism of G into A(G) (). If 2 € 206) then 7g) = 292"! = g for all g InG; thus ¢7 ts the Identity mapping on G, hence z is in Ker w. Therefore Z(G) ¢ Ker w. For the other direction note that if ae Ker y then og = (a) = identity mapping on 6, Nence g = gig) = aga”!, from which we get that ga = ag for all gin G Ths puts 2 in 2(G), Therefore Ker w Z(G). Thus we get that 2(G) = Ker w. 27, fg 1s1nG the, since # Is onto, g = 8¢a) for some a in G. Thus g~!a(N)g = a¢a)"!a(N)8(a) = a¢a"!yetnvaCa) = a(a~!Na) ¢ (ND, Since N Is normal in G. aCHAPTER 2: Groups / 45 Thus @(N) is normal inG (the result is also a consequence of the result in Problem 15). 29. (a). The mapping ag defined by o4(x) * a 'xa is an automorphism of G, thus if Mts a characteristic suogroup of G then a” 'ra = e4(M) CH for alla inG, Thus M ts normal in G (b). Since M and N are normal in G we already know that MN is a (normal) subgroup of 6. If @ is an automorphism of 6 then, since @(M) CM. and @(N) CN, we get that @(MN) = @(M@(N) C MN. Thus MN is 2 characteristic subgroup of G. (c). Let G be the group of order 4 having the elements e, a, b, ab where a2 = b2=e, and where ab =ba. Since the group G is abelian, every subgroup of G is normal in G, thus A = (e,al is anormal subgroup of G. The mapping @ defined on G by #(e) = e, ola) = b, @(b) = a, and (ab) = ab can be seen to be an automorphism of G But @(A) = {2,b) is not contained in A Thus A is not a characteristic subgroup of G. 3 Since H is of order p and 1s normal in G, if @ 1s an automorphism of G and if @(H) =H then H@(H) is a subgroup of G and is of order p2. (See Problem 16 to see that HH) is a subgroup of G; to see why it nas order p? see the argument given in Problem 22 of Section 3). Thus p@ = HoH must divide [Gl = pm, by Lagrange’s Theorem, Thus pI pm, and so p|m, contrary to assumption. Thus H is a characteristic subgroup of 6. 33. If Nis normal in then, for a in G, 6 defined by a_{x) = a !xa is an automorphism of G and a(N) C N since N 1s normal in G. Thus ¢g induces (gives rise to) an automorphism of N, hence takes 11 into itself because M is a characteristic subgroup of N. Which is to say 64(t1) = a” !Ma CM for all a in. Thus M is normal in G.46 / Student's Solutions Manual 34. Let @ be an automorphism of G and consider o, the automorphism of 6 defined by 03(%) = a” !xa for all x in 6. Then (60.387!) = a¢e.4(8° H(x)) = 9(a7!8°!(x)a) = @¢a)” !x0¢a) = og¢ay(), hence 08-1 = g(a) 50 15 in 166) Thus 1(G) 1s normal nig). Harder Problems, ‘37. Let G be a non-abelian group of order 6. If every element were of order 2 then, by Problem 9 of Section I, G would be abelian. Also, if there were an element of order 6 in G then G would be cyclic. Since G Is of even order It nas an element a =e such that 2? =e. If b =e inG is of not of order 2, by what we said above and Lagrange's Theorem, b has order 3. By {the result of Problem 30 the subgroup & = (e, b, 2) is normal in G. Now, since 6 is non-abeltan, ab = ba, yet aba is in B since B is normal in G Thus abar! = 6°! Since a? = b3 = @ and ab=b°!a the mapping of 6 onto 385 which sends a to f and b tog, where {? = 93 = e and fg = g7!r gives an ‘somorphism of 6 onto Sy 38. (a). (TyTe Ha) = Ty(T (Had) = Ty(Hac™!) = Hac™!'p=! = Hatbey! = Tyc(Ha) for every a in G, Thus The * The (O). Suppose u is in Ki); thus Ty = Wu) ~ ig Thus, for every a ind, Hau = Ty(Ha) = Ha, and so Haya” i Therefore aua’' ‘5 in H far every a in G. Conversely, if aua”! is in H for every a in G the argument reverses to Show that Ty = Wu) = ig. Thus KU) = (ue G | ava”! €H for every a in G) This tells us that if u is tn K(w) then u is in every a7! Ha, from this we get that K(w) is the intersection of all a~'Ha as a runs over 6, aCHAPTER 2: Groups / 47 (c). K(w) is a normal subgroup of 6, being the kernel of a homomar~ phism of G, and lies in H since ava’! is in H for every a in G, so in particular, for a= e Thus K() CH. Suppose that NCH is a normal ‘subgroup of G; then aNa”! CN CH, hence NC K(W) 40. Suppose that H is a subgroup of G, IG! = n, and that n does not divide Ig(H)! If S ts as in Problem 36, A(S) has elements has igiH)! elements, $0, by Lagrange’s Theorem, nas no subgroup of order n. Thus the mapping w of Problem 38 cannot be an isomorphism. Therefore K(w) = (e) is a normal subgroup of G contained in H. 4l. If IG= 21 and is 2 subgroup of order 7, then ig(H) = 3, and since 7 does not divide 3! = 6, H contains anormal subgroup N = (e) of G. But, since ‘the only subgroup of H which is different from (e) is H itself, we conclude ‘that N= H. Hence H is normal in G. 43,, 44, and 45, Let G be a group of order p2 where p is a prime. If G ts cyclic then we are done, for then G Is abelian. So if a = @ Is inG then ofa) = p, and the subgroup A = (a) is of order Thus ig(A) = p2/p = p, and p” coe not divide pl, Gnas anormal subgroup T = (e) contained in A. Hence T= Aand A is normal in G. So if b is in then bay”! = a! since A is normal and generated by a From this, since 0? = e, we get that a = bPab"P = a” where m = iP. (See Problem 29 of Section 4 for the kind of argument needed for this last step) Since a"! = @ anda Is of order p, p must divide m-1 =i? - |, however by Fermat's theorem, iP = i mod p. The outcome of all this 15 that 1 | mod p. Hence a! = a and so bab”! = a, that is ab = ba for all inG This argument held for any a * in G. Thus all elements of G are in Z(G), So G is abelian. Note, for Problem 44, that if G is cyclic and generated by a then a generates a subgroup of order p; If G Is not cyclic48 / Student's Solutions Manual then every a= e In G Is of order p, so generates a subgroup of order p. At any rate, G must have @ subgroup of order p, and it is normal since G is abelian. If G is of order 9 then p = 3, andG is abelian. 46. If G is cyclic with a as generator then.a? has order S and a5 has order 3, and we would be done. So suppose that 6 is not cyclic. Every non- identity element has order a divisor of 15, so has order 3 or 5. ‘Suppose that there aren't elements of both orders 3 and 5. So every element has order 5 or every element has order 3. If a and b are of order'S and b is not a! for any i, then the elements albK, where J, k take on alt values between 0 and 4 give us 25 distinct elements-~ far too many for 6 which only has 15 elements. ‘Suppose then that every element in G other than e has order 3. If a= @ is in G and ba = ab we claim that b =a! for some i. if not, since the ‘subgroups 6 = (b) and A = (a) satisfy AB = BA, ABis 2 subgroup of G of ‘order 9, and since 9 does not divide 15 this is not possible. Suppose that c isnot in A; thus the 3 elements c, aca”!, a@ca™2 = a-!ca are distinct, soc Gives rise toa triple of distinct elements in this way If d is not e nor any ofc, aca”! a 'ca then d, ada”!, a” !aa give us 3 new elements. For, if ada”! say, is one of these earlier elements then ada”! = alca™!, leading to the contradiction that d= a!~!ca°(!"!). Continue this way to get k distinct triples. These together with e exhaust G so the number of elements in G is 3k + 1 = 15, which implies that 3| 14 which is false. So not every element of G can have order 3 Very Hard Problems. 49. We first show that if ig(A) and i¢(B) are finite for the subgroups A and B of G then ig(A n B) is also finite. Let Auy, Ay, .. ,AUp, be all theCHAPTER 2: Groups / 49 distinct left cosets of A inG, and Bv), Bv9, .., Bvp those of B in G. Since AnB isa subgroup of B, B is the (possibly infinite) union of left cosets (An dw, where the w, are in 8, We claim that there are at most m distinct left cosets of AB inB For suppose Wy, .., Wms give us m+ such distinct cosets. Since G is the union of the Au, each wy = aguj where the aj are in A Since the number of uj is m, we must have that for two different k, q the same | appears for Wy and Wa; that 1s w= ayuy and Wa 7 ag, But these imply that w,Wa”! = ayaq7! so is in A; but wg! 15 tn since each of wy and Wa is. Thus WW a! Is in An, contrary to the fact that they give distinct left cosets of An B inB. Thus 6 is the union of at most m left cosets (A 1 8)Wp, hence By, i= the union of (An B)v,v;, and so G 1s the union of the (Aa B)w,v), which are at most mn in number Thus ig(A 0B) is finite a By induction we easily then get that 1G), Go, .., G, are of finite index in Gthen A; M Ag @. Ag 1S of Finite Index in Gif H is of finite index in G we claim that there are only a finite number of distinct a”'Ha in G. By the result of Problem 19, NH) = (a € Gl a” 'Ha = H) Is a subgroup of G and contains H, $0 15 of of finite index in G, in fact igiN(H)) < ig(H). Also the number of distinct a” 'Ha equals ig(NH)) (Provel). Hence there are only a finite number of distinct a”'Ha in 6, Each of these is of finite index in G (Prove!) , 50 their intersection N is of finite in G. By Problem 18, Nis formal in 6 50 Let aand bbe such that a2 -b*=e anda = e=b, and a=b, and50 / Student's Solutions Manual ab = ba The group, N, they generate (e, 2, b, ab] is abelian, hence all its ‘subgroups are normal. Let G be generated by 2, b, and g where g2 =e, g2 = bg, gb = ag, and gab = abg, Then N is normal in G but M =(e,a) which is normal in N is not normal in G, for gag”! = b is not in tH. SI. Let f be the mapping defined by f(x) = lx)” !; if f(x) © fly) then 26x! = ely"! sox ly = bx)” ely) = (x !y). By our hypothesis on @ we must have x !y = and sox = y. Thus f is 1-1, hence maps G onto itself. Therefore, given a inG, then a= @(xx"! for some x in G, thus @(a) = 22(x)@(x7!) = xa(x)"! = a”! since @2 ts the identity automorphim of 6. Thus b°!a"! = (ad)! = (ab) = g(aig(d) = 2” 'b“!, whence G is abelian 52. Let A= (a€ G1 ofa) = a”), and suppose that b € A. Thus both A and Ab have more than 3/4 of the elements of G, hence Am Ab has more than half ihe elements of G If x is In Am Ab then x = ab, where a is also in A, and (x) = x7!) = pla"! But @60) = @fabe(b) = a7 !b7!; consequently ab = ba follows. So whenever ab is in Awe must have that ab - ba, The number of ‘such a is more than half the elements in G, so the subgroup C(b) = (x €8 | xb = bx) nas order greater than |GI/2 yet divides {61 by Lagrange’s Theorem. Hence C(b) = G So b € Z(G); thus A ¢ 206). Therefore 2(G) has order larger than 3IGI/4, so must be all of G. Therefore G is abelian. Because G Is abelian, A becomes a subgroup of G, and since its order 1s larger than 3IG//4, A= G. Thus @(x) = x7! for all x inG. SECTION 6. 2. If aisareal number identity Na with lal; since the cosets of N inG multiply via NaNb = Nab which jibes with the fact that fab| = falfbl 4 — If gisinG then, since M ts normal in G/N, if X = Ng then X”! Mx CM; this gives us that Ng 'Mg @M, and so g”'Mg CM Thus M is normal in 6. aCHAPTER 2: Groups / 51 6. Every point in the plane has a mate in the unit square wnere 04x < 1 and 0 = bk where k = 19, and so DK"! + @. This implies that 71 (5-1), and since, by Fermat's Theorem, 71 (i® - 1) we get that 71 (1 - 1) But this says that 0! = b, and so ab = ba But then c= ab is of order 57 = 35, hence G Is cyclic. 4, Let Gbe generated by a and b wnere a> =p? = e and aba”! =p? the 21 distinct elements blal, where 0s j «3 and 0< 1 <7, form.a group for, as can be verified from the relations between a and b, (blabla) = bas wherer=1+ 2) ands =m +n. 5. Suppose that [GI = pm where p does not divide m, and suppose that P is anormal subgroup of order p". If @ is an automorphism of G then Q= 0(P) 13 a subgroup of order pl and PO has IPIIOV/IP 0 Ol = p2PVIP n al elements. Thus, if © = Q, then [Pal = 9, where s 21+ | But p® does not divide pm since p does not divide m. With this contradiction we get that 9(P) = P , hence P 18 a characteristic subgroup of 6. 6 Since IAB! = [AIBA a BI <1Gi, 1A 0 B12 IAIBI/IGI 2 VIG! YIGW/IGI > 1. Thus AnB=(e) 1A.n8|= 1, hence AB has IAll6l = mn distinct elements 8 By Cauchy's Theorem G has an element @ of order 11. Thus for the suvgroup A = (2) of order 11, ig(A) = 9 and 11 does not divide 9}, hence, by Problem 40 of Section 5, A is anormal subgroup of G aCHAPTER 2: Groups / 55 10. By Problem 9, Ghas a narmal subgroup of N of order 7; thus G) = G/N is a group of order 6 AS such, G has a normal subgroup T of order 3. By the Second Homomorphism Theorem (Theorem 2.7.2) the subgroup T= (aeG|NaeT)} 1s anormal subgroup of G and T/N = T). Since Ty = IT/MI = ITI/IN:, we get that ITI =IT IN| = 37 = 21 Harder Problems. 12. Since G is a group of order 21 it has an element a of order 7 and an element b of order 3. The subgroup A = (a) of order 7 is normal in G since 7 does not divide ig(All= 3!= 6, Therefore bab”! = al. Since G ts non-abeltan, i= 1. But since DS = e we get that a = a€ where k = 19, thus 13 - 1 ts divisible by 7 This gives us that |= 2or 4 If t= 2 tren p2an72 = a4, so in all circumstances 6 has an element c such that cac™ af If G, ts anotner non-abelian group of order 21, the same argument shows that G ; has elements u and v such that uf= v3 =e and vuv"! = u4. Define the mapping £ of G to Gy by f(a) = u and f(c) =v and fale} = lv). This mapping ts an isomorphism of G onto G Very Hard Problems. 13. By Cauchy's Theorem G has an element a of order 11, and since A = (a) is a subgroup of order 11, ig(A) = 9; because 11 does not divide 9! we have that Aisa normal subgroup of G We claim that A ¢ Z(G); for if gis InG then gag”! = al since it is in A, nence g! 'ag™!! = a™ where m=! ! By Fermat's Theorem, i!! mi mod 11; thus a!” = al. The net result of all thts56 / Student's Solutions Manual is that g! Tag"! = al = gagé!, trom which we get that g!a = ag!9. Since 10 does not divide 99 = |él, we easily get from this that ga = ag. Thus 2 € 266), hence A= (a) ¢ 266) Also G/A is of order 9, hence is abelian. Thus if u and v are inG then ww" lv"l is in A (see Problem 12 In Section 6). Hence uv = zvu where z is inA, thus in 206), and 2!! = @. Thus u2v = u(uy) = u2vu = zuvu = zvu2, since Z is in Z(G). Continuing this way we get that uly = z'vu!_ in particular, if 11, since z that ull = u2=y =e, wegetul!y«vull Thus if u is of order 3 we get 1, 1 80u!y= vu"! for ally inG. in short, umust be in 2G), Thus 2(G) has order at least 33 since it contains an element of order |, namely 2, and an element of order 3, namely u. Thus the order of G/Z(G) is 1 or 3; at any rate, G/2(G) is cyclic. By Problem 11 of Section 6, G must be abelian. 14 Consider the group generated by the two elements a and b where we impose the conditions that a? = b= e and bab”! = al. what value should ‘we assign to i in order to get consistency with the relations a = pU= e {and to insure that the group so obtained is a non-abelizn group of order pa? AS we have done many times, this implies that bab" = a where m = if. Thus ifr =a, since bY = e we get a= a™, and so a! This would require that q1 (m ~ 1) = (i9~ 1) and (since we want G non-abelian) q does not divide 1-1. Can we find such anr ? Yes, since Up 18 a cyclic group and 1 -1) there is an element [i] = (11 in Up such that (119 = (1], that ts, an integer i, where | d, we get that (G),G2) cannot be cyclic58 / Student's Solutions Manual 4 Since Py and P are of relatively prime orders, the subgroup PP Is of order py ™1pq"2. Continuing by induction we get that P;Po..Py is of order py ™1pg™2..9,k = IG), Thus G = PyPo..Py. Moreover every element g in G has a unique representation in the form g = 2,29. where each a; is in Py, because, if ajap..a, = bjb bj are two such representations of g then byay7! = bp..beay ta = (bgag7 toga"), 80 byay~! As in Po..Py. But, since bya;~' is in P, its order is a power of py; the subgroup Po..Pk 18 of order p2™2..p,™k and since py does not divide p2™2..0,™k we get that ba, ~'=e, hence a, = b). Similarly we get that aj = bj for all the i's. Thus g has a unique representation in the form g = a)..a,. By the definition of internal direct product, G is the internal direct product of Py yoo Pgs thus by Theorem 2.9.4, GaP) x Pax. xP, 5. The order of NyNo..Ny 13 at most INjIINol.INg! = IGI, if for any two different products njn9..ny =myMp..my where each of m, andn, are in Ny, for every i, then we cannot achieve this maximum for the number of elements in Ny..Ny. Thus every element of G = NjNo..N has a unique representation in the form nyna..n. By the definition of internal direct product and Theorem 2.9.4 we get G is the direct product of Ny, No, = « Ny 6. To show that G Is the direct product of Ny. No, ... Ne. given (a) and (b) we merely must show that each g in G has a unique representation in the form g = nj. where each n, is in, From the hypothesis (b) we have that Ny aN) = (e) IF 1 = J since Ny N,N Nye p-Nk and, since the Ny areCHAPTER 2: Groups / 59 normal In G, we get that nin) = nya, and nym, = myn, In consequence, it 4G. My..ry = My..my where each m, Is also in Ny then we obtain that nye Eee pH Oj Mey Eng”! = mynd” so nim”! is in Ny n(Ny Ni pNjsy Ny) * (e) Thus nj ~ mj for each i, whence g has a unique representation in the form g = n}..ny. Therefore G is the direct product of Ny, No, Ny SECTION 11. Earier Problems. 1. The conjugacy classes in Ss are (e), (f,fg, gf), and (g,92), where f and Q generate Ss, and =e andtg=g"'t (See Problem 19 In Section 4). Also Cle) = S3, Cif) = (e,f) and Cig) = (e.9.94) $0 ICC) = 2 and IC(g)l = 3, and ISGVICCEM = 1, ISsIZIC(E = 2 and ISsVICCQN = 2, and 1+ 2+3=6is the check on the class equation 2. The dihecral goup of order 8 ts generated by a and b where a2=b4=e and ab = b"!a The conjugate classes are (e}, (D2), (b,b9), (a,ab2), (2b,ab3] and C(e) = 6, C(b2) = G, Ctd) =(e,0,02,0), C(a) = (e,a,ab2,02). Therefore IG/\Cley = 1, Iev/ictn?y = 1, IGI/IC(Dy = 2, |GIICCa)| = 2, and IGI/ICCaby = 2; thus the class equation checks out as 1 + 1 + ae 242-6. 6 If P is normal inG and Q = P 1s 2 p-Sylow subgroup of G of order p™ then PO = OP, so PQ ts a subgroup of G and [POI = (PIII ad) = p2"/1P wal» 1%" 'must divide IGi Since I6i = pm whe (mp) =1, this 1s not possible Thus P = Qand P is the only p-Sylow subgroup of 6.60 / Student's Solutions Manual 6 We proceed by induction on {61 to prove that if the prime p divides I6| then G has an element of order p. IFIGI = p the result ts trivially true since every a =e inG is of order P. Suppose that the theorem is true for all groups H such that IMI 4 Since it is a subgroup of T and Tl = 27, Iker wl isa ‘subgroup of T, its order must divide 27 We thus have that [Ker wl = 9 or 27. 18. Let abe in N(N(P)); thus a” 'N(PIa © NCP), and since P © NP), a7!Pa Is contained in N(P). But then a” !pa Is a p-Sylow subroup of NCP). By the result of Problem 6 we know that P 1s the only p-Sylow subgroup in N(P). Thus a7 'Pa =P, whence a € NP); therefore N(N(P)) € N(P). Since N(P) C NN(P)) (Since H.C NCH) for any sugrouD H of G), NONCP)) = N(P). 19. We go by Induction on n. For any n=1 a group of order p has an element of order p, thus a subgroup of order p. Thus the result is correct for n= 1. Suppose that any group G of order p" nas a subgroup of order p™ ror all 0m we get + 2/D = 2, and so b?- 2b + 2=0, that is, (b- 1)? =-1 This 1s impossible for any real, hence for any rational, number Thus the given polynomial 1s irreducible over a (Note the irreducibiity of the polynomials in Parts (b) and (1) 1s an88 / Student's Solutions Manual Immediate consequence of the Eisenstein Criterion in Section 6). 1 If pCa) 19 not irreducible it 13 the product of two polynomials of degree less than 3. So one of these is of degree I, hence Is of the form ax+b, where a= 0 and b are in F. Thus p(x) = (ax * bXcx2 + ox + e); hence if = -b/a thenr is in F and p(r) = 0. Middle-Level Problems. 13. Let be the ideal (x? + 1LIf u=x + J is the image of x in BixI/J then every element in A = RIxI/J is of the form a + bu, This is true since for every polynomial fx) in Rbxd, £(x) = gixXx2 + 1) + (a+ bx), by the division algorithm, hence (Ox) + J (a + bx) + xN(x2 + 16 = (a+ Ox) eJ= a+ bu, Since gxXx2 + 1) 15 in the ideal J = (x2 + 1). Also, u2 = (x + WZ =x2+y= “1+ J, which ts -1 in A= BUd/J. The mapping @ from A to € def ined by (2 > bu) = a+ bi is clearly an isomorphism of A onto € 14. (a). There 1s no a in Z) such that a2 + 1 = 0, as can be verified by a Girect check of the 11 elements in F = Z, ). Thus by the remark preceaing Problem 10 (or by a quadratic version of Problem 11), x2 + 1 is irreducible 192) fel Therefore Flxl/(x? + 1) 4s a field. Exactly as in Problem 13, every element in Fixj/(x? + 1) is of the form a+ du, where u= x + (x2 + 1) and 3, D €F. Moreover every element Is of the unique form a bu, for if a+ bu = c+ du then (a~c) + (b- du Hence (a - c) + (b= d)x is in(x2 + 1); since 2 every non-zero element in (x? + 1) is a multiple of x2 + 1 it has degree at least 2 Thus (a ~ c) + (b- dx = 0, and so a= c and b= d Since we have 112 Independent choices for a and b and two different choices give rise to different elements in FlxV/(x? + 1), Flex? + 1yhas 112 121 elements,CHAPTER 4: Ring Theory / 99 (b), A direct check of all the elements in F = Z,, reveals that for no a inF is.a3 a4 40 Thus, as before, plx)= x3 + x + 4 ts irreducible in Fix] , hence Fix]/(p(0)) is a field. Similar to the argument in Part (a), every element in Flxl/(p(x)) has the unique form a + bu + cu2 where u = x + (p00) and a,b, ¢ are in F, Thus FixI/(p(x)) has 117 elements, 17 Define the mapping @ from F[x] into FIxI/(p,(x)) @ . @ FLxI/(p,(x)) by CFC) = (Fx) + (py O02, £00 + (DQLK)), 5 (00 + (B40) for every £00 in F{x]. This mapping is a homomorphism, since this is true in every Component. Furthermore, Ker @ = ((0x) € FIX} (0x) € (P00) for t= 1,.,K), that 1s, Ker @ = (p)(x)) 9 (pa(x)) a... (by (xd) We claim that @ maps Fle] onto FIXi/(p yx) @ . @ FLXI/(p OO). This 15a special case of what in algebra 1s called the Chinese Remainder Theorem We prove this more general result and apply it to the problem at hand Let Rbe a commutative ring with | andM), ., Mp distinct maximal ideals of R, then R/(My Ary A. AMA) = A/M, BRM, @ . @ RM, We proove it for n= 2, leaving the few details needed for induction to the reader goes as follows. Since?) =f are maximal ideats of R, Hy +My =R; thus, given x andy in R, x= m4 + mg andy = ny * ny where m; andny are int; and my and ng are in My. Thus (x +My, ¥ * Mg) = (mj + my *My.ny emg + Mp) = (mz +My, * Mg) = @{c), where the element r= mo +n). Thus @ is onto Returning to our problem, if p(X), .., Pk(x) are irreducible90 / Student's Solutions Manual polynomials in Flxl, then the ideals (D(x), .., (Pg(x)) are distinct maximal ideals of Ffxl, thus a direct application of the Chinese Remainder Theorem Proves the desired result. 18 Suppose that the irreducible polynomials of F{x] are bounded in degree; then, since F is finite, there would only be a finite number of irreducible polynomials in Flxl If these are pj (x), .., n(x), conisider the Polynomial a(x) = 1 + py(x)..0,(x) ; a(x) must be divisible by some irreducible polynomial, that is, by some p,(x). But this is clearly false. With this contradiction we get that there must be irreducible polynomials of arbitrarily high degree in Fix] 22. Since R does not have a unit element we know, by Problem 21, that R iS not a Euclidean ring We claim that d(2) < da) for any non-zero even integer. For, 1f d(a) « 2) then, if the Euclidean algorithm held in R, we would have 2= ga +r, where r = 0 or dir) < d(a), and a (s even. By the choice of a, we must have r = 0 and 2 = qa where q and a are even, which is impossible. If we consider 2 and 6 in R and if 6 = 2m +b where m is even, Since (b) cannot be less than d(2), we must have b = 0 and 6 = 2m where m is even, This is nonsense, so the Euclidean algorithm does not hold for 2 and 6, Harder Problems. 23° That x5 -2 and x5 + 2 are irreducible foliows from Problem 11, since a direct check shows that there are no elements a and b in re) such that a3 = 2 or D3 = -2 A= FIMO ~ 2) and r= x» (x3 2) In A then every element in A is of the form a+ br + cr2 where a,b, ¢ are in Ty and‘CHAPTER 4: Ring Theory / 91 3 +2 Similarly, if 8 = Flxl/(x3 + 2) and s = x + (x3 + 2) in Bevery element IN BIS of the form u + vs + ws? where U, v, W are In Zy ands) = -2 Map A hs + ¢s@ Checking the action of @ ‘onto 8 by defining ola + br + cr2} ‘on sums and products by a direct calculation shows that ¢ is an isomorph- tsm of A onto B 24 By Problem 11, x2 +x + | Is irreducible over Q, so FIdIx2 +x + 1) is a field in which r= x *(x2 +x + I) satisfies r? «r+ 1 = Q and such that every element is of the form a+ br where a, D are inQ@ Map the set A =(a+ bo }.a, b€Q) onto Fixx? + x + 1) by sending a + be onto a + br. This is easily seen to be an isomorphism; since A is isomorphic to the field Fixx? + x 61), Als itself a field Suppose that a + ba * 0, we claim that (a + bake + da) = | has a solution with ¢ and d rational For, (2 + baeKe * dex) * (ae - bd) + (ad + be -bdlec, nence we want that ac - bd = | andbe + (a - bid = Ohavea solution in @ for ¢ ana d This !s possible provided a(a - p) + b2 =o, but if a2 + p2- ab = 0 then 0 ¢a2 +b? = ab and a2 + b2 - ap -(a- 0)? + 2090 unigss a= = 0. SoA isa field since every non-zero element in A has its sverse nA 25. See the part on the Eisenstein Criterion in the next section. 26. ft =a ine then (t')? = 0 so, by hypothesis, 1 50; continuing thts way we get t = 0. Suppose now that agx"+ayx"l + +a, *otsa zero-divisor in R; then (agx" +. + 2_bgX™ + by) +. Day) = Owith by #0 Thus agb, = 0, agb; + aby = 0, aybg + aby + Andy = 0, .... Thus 1p * a;092 <0, from which we get ayb)2 = 0 and s0 (ajq)2= 0, giving + aybyb9? * ayb,5 = 0 from which we get that92 / Student’s Solutions Manual 29b95 = 0 and 50 (agbg)° = 0 and so apb9 = 0. Continuing with the other Coefficients we obtain a;b,' = 0, hence (ajbg)! = 0 and so a,bg = 0 for each 1 = 0,1, ..,m Thus dy * 0 does the trick 27. (a). We shall show that Ix] ts an ideal of Rfx] by showing in Part (b) that Ix] is the kernel of a homomorphism, (b). Define @ : Rix] > (R/INLxl by lage + ayx™ + and =(@y ex olay eT 6 6 (a, ets immediate from the addition and multiplication of polynomials that @ is a homomorphism of Rix} onto (R/iiIx} What is Ker @ 7 It is precisely the set of polynomials such that ay += i, ay +1=1,.., ay ¢4=4, that is the set of polynomials such that ag, ay, .. .dp are all in|. In other words, Ker @ = Hed Thus ROxd/1ix) = (R/N0 Very Hard Problems. 28 Suppose that f(x) = agp aX Ts + a5, god = Dgx™ +. + by FO are such that f(x)g(x) = 0 and g(x) ts of lowest degree with this Property. Then Bmp * 0 (Provel). We claim that (Xm = 0. Certainly aqdm #0 since this Is the coefficient of the lowest term In f(x)g(x). 1f 20m =9 for all | then (xm * 0 and we would be done. Suppose then that ap. jdm + 0 for ibm=1* = (ap. 4DgX™! + + ap.j)x and £10 g(x)a,.4) = , which teads us te the fact Vet and ap. tDpy #0. Thus for i < t, Qxlag_y = ap jdox” + ’n-tm ’n-1 ~ An-1Po' that (ap \Dgx™ | +. + dp_sDm) = 0, contradicting that g(x) is the i aCHAPTER 4: Ring Theory /:93 olynomtal of lowest degree with this property. Therefore g(x}ay =O for AML €t Thus 0 = f(xdatx) = (gx * + ay px’ + + agigixd = (a + an -xUidg xl + + ‘m)s but this implies the contradiction that n-th = 0 29. If ns a positive integer and u + vi is in thenu -ayn* ry and v= agh era where Ir] ¢n/2 andIrol < 9/2, that 1s, a(r)) ? = n, a positive integer So, by the above, (€ + diXa~ 1) = qn + r= aia + biM2- bi) +r where air! < din) = dla + b2) = ‘Si 2- DI), and so dir) = tel? = ts!2la = vif? = ashe? + 2) 1. But then @(agx” + ajx0-! « + an agli” +CHAPTER 4: Ring Theory / 95 ay 000"! +. + ay 13 of degree mn > n. Thus x, which Is of degree 1, cannot be the Image of any polynomial in F{x). Thus we get that @(x) = bx + ¢ for. some b * 0 andc in F But then, since @(a) = a for every a in F, @(f(x)) = fe) = (0x + 0), 13. The only invertible elements in Q[x] are the elements of 0. If # is an automorphism of [x] then @(a) is invertible if 2 is Thus @(0) CQ Sow induces an automorphism of Q. But any automorphism of O must be the identity map For @(1) = 1 since @ is an automorphism Therefore if n> 0 is an integer wind= ll + i+ + N= elle @(l+ telat t le eten Also @(-1) @(1) since @ 15 1-1, but (@(-1))? = @((-1)2) = @(1) = 1, Thus w(-1) =-1 From this we get that im) =m for all integers. If ¢ = m/n is. rational (m and n integers) then #(r) = w(m/n) = @(m)/e(n) = m/n =r. So any automorphism of Ofx] automatically leaves every 2iement of @ fixed. SECTION 7. Problems. 1. (fa, bl + fc, d}) + [e, fl = [ad + be, bd} + fe, 1] = {ad + be)F + bde, bat] = fadf + bef + bade, pdf), while (a, b) + (ic, dl + [e, f)) = (a,b) + (cf + de, df] = aaf + pict + de), bat] = fadf + bef * bde, baf] We thus see that the assoc~ lative law of addition in F checks out 4 If Kis. field which contains D then K must contain all fractions a/b where 2 andb * 0 are nD. Since the set of all these fractions is precisely F , we get that K DF