-
Notifications
You must be signed in to change notification settings - Fork 4
/
Cox_Galois_Chapter8.tex
1628 lines (1144 loc) · 83.9 KB
/
Cox_Galois_Chapter8.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
%&LaTeX
\documentclass[11pt,a4paper]{article}
\usepackage[frenchb,english]{babel}
\usepackage[applemac]{inputenc}
\usepackage[OT1]{fontenc}
\usepackage[]{graphicx}
\usepackage{amsmath}
\usepackage{amsfonts}
\usepackage{amsthm}
\usepackage{amssymb}
\usepackage{tikz}
%\input{8bitdefs}
% marges
\topmargin 10pt
\headsep 10pt
\headheight 10pt
\marginparwidth 30pt
\oddsidemargin 40pt
\evensidemargin 40pt
\footskip 30pt
\textheight 670pt
\textwidth 420pt
\def\imp{\Rightarrow}
\def\gcro{\mbox{[\hspace{-.15em}[}}% intervalles d'entiers
\def\dcro{\mbox{]\hspace{-.15em}]}}
\newcommand{\be} {\begin{enumerate}}
\newcommand{\ee} {\end{enumerate}}
\newcommand{\deb}{\begin{eqnarray*}}
\newcommand{\fin}{\end{eqnarray*}}
\newcommand{\ssi} {si et seulement si }
\newcommand{\D}{\mathrm{d}}
\newcommand{\Q}{\mathbb{Q}}
\newcommand{\Z}{\mathbb{Z}}
\newcommand{\N}{\mathbb{N}}
\newcommand{\R}{\mathbb{R}}
\newcommand{\C}{\mathbb{C}}
\newcommand{\F}{\mathbb{F}}
\newcommand{\U}{\mathbb{U}}
\newcommand{\re}{\,\mathrm{Re}\,}
\newcommand{\ord}{\mathrm{ord}}
\newcommand{\Gal}{\mathrm{Gal}}
\newcommand{\legendre}[2]{\genfrac{(}{)}{}{}{#1}{#2}}
\title{Solutions to David A.Cox "Galois Theory''}
\refstepcounter{section} \refstepcounter{section} \refstepcounter{section} \refstepcounter{section}
\refstepcounter{section}\refstepcounter{section}\refstepcounter{section}
\begin{document}
\section{Chapter 8 : SOLVABILITY BY RADICALS}
\subsection{SOLVABLE GROUPS}
\paragraph{Ex. 8.1.1}
{\it Consider the groups $A_4$ and $S_4$.
\be
\item[(a)] Show that $\{e,(1\, 2)(3\, 4), (1\, 3)(2\, 4), (1\, 4)(2\, 3)\}$ is a normal subgroup of $S_4$.
\item[(b)] Show that $A_4$ and $S_4$ are solvable.
\ee
}
\begin{proof}
\be
\item[(a)] Write $a = (1\,2)(3\,4), b = (1\,3)(2\,4),c=(1\,4)(2\,3)$. Note that $e, a,b, c$ are even permutations, so $K = \{e,a,b,c\}\subset A_4$. They satisfy the relations
$$a^2 = b^2=c^2=e, a b = b a = c, a c = c a = b, b c = c b =a.$$
This gives the same Cayley table of the Klein's four-group $\mathbb{Z}/2 \mathbb{Z} \times \mathbb{Z}/2 \mathbb{Z}$, where we write
$$e' = (0,0) , a' = (1,0), b' = (0,1) , c' = (1,1).$$
The mapping ($e\mapsto e', a\mapsto a',b\mapsto b',c\mapsto c'$) is an isomorphism.
If $\sigma \in S_4$, $\sigma a \sigma^{-1} = \sigma [(1\,2)(3\,4))] \sigma^{-1}= (\sigma(1)\, \sigma(2)) (\sigma(3)\, \sigma(4)) \in K$, and the same is true for $b$ and $c$, so $K$ is normal in $S_4$ (a fortiori in $A_4$).
Conclusion: $K = \{e,(1\, 2)(3\, 4), (1\, 3)(2\, 4), (1\, 4)(2\, 3)\}$ is a normal subgroup of $S_4$ included in $A_4$, and $K$ is isomorphic to $\mathbb{Z}/2 \mathbb{Z} \times \mathbb{Z}/2 \mathbb{Z}$.
\item[(b)] So we obtain a chain $$S_4 \supset A_4 \supset K \supset \langle a \rangle\supset \{e\},$$
$\bullet$ $A_4$ has index 2 in $S_4$, so $A_4$ is a normal subgroup of $S_4$,
and $S_4/A_4 \simeq \mathbb{Z}/2 \mathbb{Z}$.
$\bullet$ By part (a), we know that $K$ is normal in $A_4$.
As $\vert A_4/K \vert = 3$, $A_4/K \simeq \mathbb{Z}/3 \mathbb{Z}$.
$\bullet$ $K$ being Abelian, $\langle a \rangle$ is normal in $K$, and $K/ \langle a \rangle \simeq \mathbb{Z}/2 \mathbb{Z}$.
$\bullet$ $ \langle a \rangle/\{e\} \simeq \mathbb{Z}/2 \mathbb{Z}$.
Conclusion: $S_4$ is solvable (and also $A_4$).
\ee
\end{proof}
\paragraph{Ex. 8.1.2}
{\it This exercise is concerned with the first part of the proof of Theorem 8.1.4.
\be
\item[(a)] Prove assertions (a)--(d) made in the proof of the theorem.
\item[(b)] Suppose that $\phi : M_1\to M_2$ is an onto group homomorphism. If $|M_1| = p$, where $p$ is prime, then prove that $|M_2| = 1$ or $p$.
\item[(c)] Explain how part (b) proves the assertion made in the text that $\tilde{G}_{i-1}/\tilde{G}_i$ either is trivial or has prime order.
\ee
}
\begin{proof}
Let $G$ solvable, and $H$ a normal subgroup of $G$. We must prove that $G/H$ is solvable.
There exist subgroups $G_i$ of $G$ such that
$$\{{e}\} = {G}_n \subset \cdots \subset {G}_i \subset {G}_{i-1} \subset \cdots \subset {G}_0 = G,$$
where ${G}_i \lhd {G}_{i-1}$, and $ {G}_{i-1} / {G}_i$ is of prime order.
\be
\item[(a)] Let $\pi : G \to G/H, g \mapsto gH$ the canonical projection, and $\tilde{G}_i= \pi(G_i)$.
$\bullet$ As $G_0 = G$, $\tilde{G}_0 = \pi(G) = G/H$ since $\pi$ is surjective.
$\bullet$ As $G_n = \{e\}, \tilde{G}_n = \pi(\{e\}) = \{ eH\} = \{\overline{e}\}$, where we write $\overline{e}$ the identity of $G/H$.
$\bullet$ We know that $G_i \lhd G_{i-1}$. Then we show that $\tilde{G}_i \lhd \tilde{G}_{i-1}$.
Let any $\overline{x} \in \tilde{G}_{i-1}$, and $\overline{y} \in \tilde{G}_{i}$, with $\overline{x} = xH, \overline{y} = yH, x \in G_{i-1},y\in G_i$.
$\overline{x}\, \overline{y} \, \overline{x}^{-1} = xH yH x^{-1} H = xyx^{-1} H = zH$, where $z=xyx^{-1} \in G_{i}$, since $G_i \lhd G_{i-1}$.
Therefore $\overline{x}\, \overline{y} \, \overline{x}^{-1} = \pi(z) \in \pi(G_i) = \tilde{G}_i$. So $\tilde{G}_i \lhd \tilde{G}_{i-1}$.
$\bullet$ Let $\varphi$ be the mapping
$$
\varphi :
\left\{
\begin{array}{ccc}
G_{i-1} & \to &\tilde{G}_{i-1}/\tilde{G}_i\\
g & \mapsto & \pi(g)\tilde{G}_i
\end{array}
\right.
$$
(if $g \in G_{i-1}, \pi(g) \in \tilde{G}_{i-1}$, thus $\pi(g) \tilde{G_i} \in \tilde{G}_{i-1}/\tilde{G}_i$).
If $g \in G_i, \pi(g) = gH \in \tilde{G}_i$, thus $\varphi(g)=\pi(g)\tilde{G}_i = \tilde{G_i}$, which is the identity of $\tilde{G}_{i-1}/\tilde G_i$, so $g \in \ker(\varphi)$. This proves
$$G_i \subset \ker(\varphi).$$
As $G_i \subset \ker(\varphi)$, if two elements $g,g'$ of $G_{i-1}$ are congruent modulo $G_i$, i.e. $gG_i = g' G_i$, then $g^{-1} g' \in G_i \subset \ker(\varphi)$, so $g,g'$ have the same image by $\varphi$. Consequently $\varphi(g)$ depends only of the class $gG_i$ of $g$ in $G_{i-1}/G_{i}$.
The mapping $\overline{\varphi} : G_{i-1}/G_i \to \tilde{G}_{i-1}/\tilde{G}_i$ defined by $g G_i \mapsto \pi(g)\tilde{G}_i$ is thus well defined, and is a group homomorphism.
Let $y = v \tilde{G}_i$ be any element of $\tilde{G}_{i-1}/\tilde{G}_i$, where $v \in \tilde{G}_{i-1}$, so $v = \pi(g)$ for some $g \in G_{i-1}$. Therefore $y = \pi(g) \tilde{G}_i = \varphi(g) = \overline{\varphi}(gG_{i})$, so $\overline{\varphi}$ is surjective.
\item[(b)] Let $\phi : M_1 \to M_2$ be a surjective group homomorphism, and suppose that $\vert M_1 \vert = p$ is prime.
Then $M_2 \simeq M_1/\ker(\phi)$. The order of the subgroup $\ker(\phi)$ of $M_1$ divides $\vert M_1 \vert = p$, so $\vert \ker(\phi) \vert = 1$ or $p$, thus $\vert M_2 \vert = \vert M_1 \vert / \vert \ker(\phi) \vert = p$ or $1$.
\item[(c)]
By hypothesis $|G_{i-1}/G_{i}| = p$ is prime. By part (b) applied to the surjective group homomorphism $\phi = \varphi:G_{i-1}/G_i \to \tilde{G}_{i-1}/\tilde{G}_i$, we know that $\vert \tilde{G}_{i-1}/ \tilde{G}_{i} \vert = 1$ or $p$.
Therefore $\tilde{G}_{i-1}/ \tilde{G}_{i}$ is trivial, or of prime order.
To conclude:
$$\{\overline{e}\} = \tilde{G}_n \subset \cdots \subset \tilde{G}_i \subset \tilde{G}_{i-1} \subset \cdots \subset \tilde{G}_0 = G/H,$$
with $\tilde{G}_i \lhd \tilde{G}_{i-1}, \tilde{G}_i / \tilde{G}_{i-1}$ trivial or of prime order. By discarding duplicates, we obtain a composition series which proves that $G/H$ is solvable.
\ee
\end{proof}
\paragraph{Ex. 8.1.3}
{\it Consider the map $\pi : G \to G/H$ used in the proof of Theorem 8.1.4. Given a subgroup $K \subset G/H$, define $\pi^{-1}(K)$ as in (8.1).
\be
\item[(a)] Show that $\pi^{-1}(K)$ is a subgroup of $G$ containing $H$.
\item[(b)] Show that $H$ is the kernel of $\pi$ and that $H = \pi^{-1}(\{eH\})$.
\item[(c)] Show that $G = \pi^{-1}(G/H)$.
\ee
}
\begin{proof}
Let $\pi : G \to G/H$ the canonical projection, and $K$ a subgroup of $G/H$.
\be
\item[(a)] $\pi^{-1}(K) \subset G$ is the pre-image of a subgroup by the group homomorphism $\pi$, so is a subgroup of $G$.
Moreover $\{\overline{e}\} = \{eH\} \subset K$, thus $H = \pi^{-1}(\{\overline{e}\}) \subset \pi^{-1}(K)$. So $\pi^{-1}(K)$ is a subgroup of $G$ which contains $H$.
\item[(b)] For all $x\in G$, $x \in \ker(\pi) \iff xH = H \iff x \in H$.
Thus $\ker(\pi) = H$.
$eH=H$ being the identity of $G/H$, by definition of the kernel, $H = \ker(\pi) = \pi^{-1}(\{eH\})$.
\item[(c)] As $\pi$ is a mapping of $G$ in $G/H$, $\pi^{-1}(G/H)$ is the whole group $G$.
\ee
\end{proof}
\paragraph{Ex. 8.1.4}
{\it In the situation of (8.2), prove that $G_i$ is normal in $G_{i-1}$ and that ${gG_i \mapsto \pi(g) \tilde{G}_i}$ gives the isomorphism (8.2).
}
\begin{proof}
As $\tilde{G}_i \lhd \tilde{G}_{i-1}$, and as $\pi$ is a group homomorphism, then $\pi^{-1} (\tilde{G}_i) \lhd \pi^{-1} (\tilde{G}_{i-1})$, so $G_{i} \lhd G_{i-1}$.
Indeed, if $x \in G_{i-1}, y \in G_{i}$, then $x' = \pi(x) \in \tilde{G}_i,y' =\pi(y) \in \tilde{G}_{i-1}$, where $\tilde{G}_i \lhd \tilde{G}_{i-1}$, so $x' y' x'^{-1} \in \tilde{G}_i$,
then $\pi(xyx^{-1}) = \pi(x) \pi(y) \pi(x)^{-1} \in \tilde{G}_{i}$, and $xyx^{-1} \in \pi^{-1}(\tilde{G}_i) = G_i$.
As $\pi$ est surjective, $\pi(G_i) = \tilde{G}_i$, and the situation is the same as in Exercise 2, where we have proved that $\overline{\varphi} : G_{i-1}/G_{i} \to \tilde G_{i-1}/ \tilde G_{i}$ given by $gG_i \mapsto \pi(g) \tilde{G}_i$ is well defined, and is a surjective group homomorphism. It remains to verify that $\overline{\varphi}$ is injective.
If $g G_i \in \ker(\overline{\varphi})$, then $\overline{\varphi}(gG_i) = \tilde G_i$, that is $\pi(g) \tilde G_i = \tilde G_i$, thus $\pi(g) \in \tilde G_i$, so $g \in G_i$, and $gG_i =G_i$ is the identity of $G/G_i$. Therefore $\overline{\varphi}$ is injective. $\overline{\varphi}$ is a group isomorphism:
$$ G_{i-1}/G_{i} \simeq \tilde G_{i-1}/ \tilde G_{i}$$
\end{proof}
\paragraph{Ex. 8.1.5}
{\it In this exercise, you will prove Theorem 8.1.7.
\be
\item[(a)] In any group, show that $\langle g \rangle$ is normal for all $g \in Z(G)$.
\item[(b)] Prove Theorem 8.1.7 using induction on $n$, where $|G| = p^n$ and $p$ is prime.
\ee
}
\begin{proof}
\be
\item[(a)]
Consider the right action of $G$ on itself defined by conjugation with $x^g = g^{-1} x g$, then $G$ in the disjoint union of the associate orbits:
$$G = \bigcup_{x \in S} O_x,$$
where $S$ is a complete set of representative for the conjugacy classes: for all $y \in G$, $\vert O_y \cap S \vert = 1$.
The stabilizer of $x$ is the set of $g \in G$ such that $g^{-1} x g = x$, so $xg = gx$: this is the normalizer $C_x$ of $x$.
Consequently $\vert O_x \vert = (G : C_x)$, and so
$$\vert G \vert = \sum_{x \in S} (G:C_x).$$
Note that $$(G:C_x) = 1\iff C_x = G\iff \forall g \in G,\, gx = xg \iff x \in Z = Z(G).$$ If we take apart these elements in the preceding sum, we obtain (noting that $Z \subset S$ since $O_x = \{x\}$ for all $x \in Z$)
$$ \vert G \vert = \sum_{x \in S} (G:C_x) = \sum_{x \in Z} (G:C_x) + \sum_{x \in S - Z} (G:C_x) = \vert Z \vert + \sum_{x \in S-Z} (G:C_x).$$
Writing $T = S - Z$, we obtain so the class formula
$$ \vert G \vert = \vert Z \vert + \sum_{x \in T} (G:C_x).$$
If $G$ is a $p$-groupe of order $p^n$, then for all $x \in S - Z$, $(G:C_x) >1$ is a power of $p$, so $(G:C_x) = p^k, k\geq 1$, thus $p$ divides $(G:C_x)$ for all $x \in T$.
As $p$ divides also $\vert G \vert$, the class formula implies that $p$ divides $\vert Z \vert \geq 1$, and so $\vert Z \vert \geq p$. Therefore the center of a $p\,$-group is not trivial.
\item[(b)] If $g \in Z(G)$, then for all $x \in G$, and for all $k \in \mathbb{Z}$, $ g^k \in Z$, so $x g^k = g^k x$, $x g^k x^{-1} = g^k \in Z$, therefore $\langle g \rangle$ is normal in $G$.
\item[(c)] If $n=1$, every group of order $p^n = p$ is cyclic, a fortiori solvable.
Using induction, suppose that all groups of order $p^k, k<n$ are solvable. Let $G$ be a group of order $p^n$.
Fix $g \in Z, g \neq e$. This is possible since $Z$ is not trivial. By part (b), $H = \langle g \rangle$ is a normal cyclic subgroup $G$, so $H$ is solvable, and $G/H$ as for cardinality a factor of $p^n$, so $|G/H| = p^k$, with $k<n$ since $\vert H \vert > 1$. The induction hypothesis implies that $G/H = G/\langle g \rangle$ is solvable.
As $H$ and $G/H$ are solvable, by Theorem 8.1.4, $G$ is also solvable, and the induction is done.
Every finite $p$-group is solvable.
\ee
\end{proof}
\paragraph{Ex. 8.1.6}
{\it In this exercise you will prove that groups of order 30 are solvable.
\be
\item[(a)] Use the method of Example 8.1.10 to prove that groups of order 10 or 15 are solvable.
\item[(b)] Show that a group of order 30 is solvable if and only if it has a proper normal subgroup different from $\{e\}$.
\item[(c)] Let $G$ a group of order 30. Use the third Sylow theorem to show that $G$ has one or ten $3$-Sylow subgroups and one or six $5$-Sylow subgroups.
\item[(d)] Show that the group $G$ can't simultaneously have ten $3$-Sylow subgroups and six $5$-Sylow subgroups. Conclude that $G$ must be solvable.
\ee
}
\begin{proof}
\be
\item[(a)] Let $G$ be a group of order 10, and $N$ the number of 5-Sylow subgroups of $G$. Then $N\equiv 1\pmod 5$ and $N\mid 10$. Therefore $N= 1$. A group of order 10 has a unique 5-Sylow $H$, and as all 5-Sylow subgroup are conjugate, $H$ is normal in $G$, since the conjugate of a 5-Silow is a 5-Sylow. Then $\vert H \vert= 5$ and $\vert G/H\vert = 2$ are prime, so $H$ and $G/H$ are cyclic of prime order. This implies that $G$ is solvable.
Same reasoning if $\vert G \vert = 15$: then $N \equiv 1 \pmod 5$, and $N \mid 15$, therefore $N=1$.
\item[(b)] Let $G$ be a group of order 30.
\be
\item[$\bullet$] If $G$ is Abelian, a fortiori solvable, it contains a 5-Sylow subgroup, necessarily normal in $G$.
\item[$\bullet$] If $G$ is a non Abelian solvable group, there exists a composition series $$\{e\}=G_n \subsetneq G_{n-1} \subsetneq \cdots \subsetneq G_0 = G.$$
Then $n\geq 2$, otherwise a short series $\{e\} = G_1 \subsetneq G_0=G$, with $G_0/G_1=G$ Abelian, is in contradiction with the hypothesis "$G$ non Abelian". Then the subgroup $G_1$ is normal in $G$, and $G_1\neq \{e\},G_1\ne G$.
Consequently the hypothesis $G$ solvable implies the existence of a proper non trivial subgroup of $G$.
\item[$\bullet$] Conversely, suppose that $G$ has a normal subgroup $H$, with $H \neq \{e\}, H \neq G$.
then $q = \vert H \vert$ divides 30, and $q\neq1, q\neq 30$, so $q \in \{ 2,3,5,6,10,15\}$.
If $q = 10$ or $q=15$, then $H$ is solvable by part (a), and the quotient group $G/H$ is then of order $3$ or $2$ both prime, so $G/H$ is cyclic. This proves that $G$ is solvable.
If $q=2$ or $q=3$, then $H$ is cyclic, and $G/H$ of order 15 or 10 is solvable by part (a), so $G$ is solvable.
A group of order 5 is cyclic, a fortiori solvable, and a group of order 6 est isomorphic to the cyclic group $C_6$, or to $S_3$, and in both cases is solvable.
If $q = 5$ or $q=6$, $H$ and $G/H$ are both solvable, so $G$ is solvable.
\ee
Conclusion: a group $G$ of order 30 is solvable if and only if it contains a proper non trivial normal subgroup.
\item[(c)] The number $N$ of 3-sylow subgroups of $G$ satisfies $N \equiv 1 \pmod 3$, and $N$ divides $30$, so $N = 1$ or $N = 10$.
The number $N'$ of 5-sylow subgroups of $G$ satisfies $N \equiv 1 \pmod 5$, and $N'$ divisdes $30$, so $N' = 1$ or $N'=6$.
\item[(d)] Suppose that $G$ contains ten 3-Sylow subgroups of $G$ and six 5-Sylow subgroups.
Two distinct cyclic subgroups of order $p$, with $p$ prime, have a trivial intersection, otherwise any non trivial element in the intersection would be a generator of these two subgroups, which would then be identical.
Consequently, two 3-Sylow subgroups have an intersection reduced to the identity of $G$, and this is the same for the 5-Sylow.
The union of 3-Sylow has then $1+10\times2 = 21 $ elements, and the union of the 5-Sylow $1+6\times 4= 25$ elements. As a 5-Sylow and a 3-Sylow have a trivial intersection, $G$ would contains at least $21 + 25 -1 =45$ elements, in contradiction with $|G| = 30$.
Therefore $N=1$ or $N'=1$. In the first case, a 3-Sylow is normal in $G$, and in the second case, this is a 5-Sylow. By part (b), $G$ is solvable.
\ee
\end{proof}
\paragraph{Ex. 8.1.7}
{\it Use Burnside's $p^nq^m$ Theorem (Theorem 8.1.8) to show that groups of order $<60$ are solvable, with the possible exception of groups of order 30 or 42. When combined with the previous exercise and Example 8.1.11, this implies that groups of order $<60$ are solvable.
}
\begin{proof}
The positive integers $n<60$ have at most two prime factors, except $30 = 2\times 3 \times 5$ and $42 = 2\times3 \times 7$. The Burnside's $p^nq^m$ Theorem shows then that groups of order $<60$ are solvable, with the possible exception of groups of order 30 or 42. Exercise 6 proves that the groups of order 30 are solvable, and Example 8.1.11 shows that the groups of order 42 are also solvable. So all groups of order less that 60 are solvable.
\end{proof}
\paragraph{Ex. 8.1.8}
{\it Let $G$ be a finite group, and suppose that we have subgroups
$$\{e\} = G_n \subset \cdots \subset G_0 = G,$$
such that $G_i$ is normal in $G_{i-1}$ for $i=1,\ldots,n$.
\be
\item[(a)] Prove that $G$ is solvable if $G_{i-1}/G_i$ is Abelian for $i=1,\ldots,n$.
\item[(b)] Prove that $G$ is solvable if $G_{i-1}/G_i$ is solvable for $i=1,\ldots,n$.
\ee
}
\begin{proof}
Suppose that
$$\{e\} = G_n \subset \cdots \subset G_0 = G,$$
with $G_i \lhd G_{i-1},\ i=1,\cdots,n$.
\begin{enumerate}
\item[(a)]
Suppose that $G_{i-1}/G_i$ is Abelian for $i=1,\cdots,n$.
Then $G_{i-1}/G_i$ is solvable (Proposition 8.1.5), so we can find a composition series $(\tilde{G}_{i,k})_{0\leq k \leq n_i}$ of $G_{i-1}/G_i$, with cyclic quotients of prime order. The proof of Theorem 8.1.4 (see Exercises 2,3,4) shows that the pre-images $G_{i,k} = \pi^{-1}(\tilde{G}_{i,k})$ of $\tilde{G}_{i,k}$ by the canonical projection $\pi : G_{i-1} \to G_{i-1}/G_i$ form a composition series
$$G_i = G_{i,n_i} \subset G_{i,n_i-1}\subset \cdots \subset G_{i,k} \subset G_{i,k-1} \subset\cdots \subset G_{i,0} = G_{i-1},$$
such that $G_{i,k} \lhd G_{i,k-1}, \ i=1,\cdots n_i$ and such that $(G_{i,k-1} :G_{i,k})$ is prime.
If we glue together all these composition series for $ i=1,\cdots,n$, we obtain a composition series of $G$ where all quotients are of prime order, so $G$ is solvable according to Definition 8.1.1.
\item[(b)]
The proof of part (a) shows that it is sufficient that the quotients $G_{i-1}/G_i$ are solvable to prove that $G$ is solvable.
\end{enumerate}
\end{proof}
\subsection{RADICAL AND SOLVABLE EXTENSIONS}
\paragraph{Ex. 8.2.1}
{\it As in Example 8.2.3, let $L$ be the splitting field of $x^3+x^2-2x-1$ over $\Q$. Also let $\zeta_7 = e^{2\pi i/7}$.
\be
\item[(a)] Show that the roots of $x^3+x^2-2x-1$ are $2\cos(2j\pi/7) = \zeta_7^j + \zeta_7^{-j}$ for $j=1,2,3$.
\item[(b)] Show that $\Q \subset L \subset \Q(\zeta_7)$, and explain why $\Q \subset \Q(\zeta_7)$ is radical.
\ee
}
\begin{proof}
\be
\item[(a)] Let $f = x^3+x^2-2x-1\in \mathbb{Q}[x]$.
The polynomial $\Phi_7 = \frac{x^7-1}{x-1} = x^6+x^5+x^4+x^3+x^2+x+1$ has the roots $e^{\frac{2ik\pi}{7}}, {1 \leq k \leq 6}$.
As $\Phi_7(0) \neq 0$, for all $z \in \mathbb{C}$,
$$ \Phi_7(z) = 0 \iff \left(z^3 + \frac{1}{z^3}\right) + \left(z^2+\frac{1}{z^2}\right) + \left(z + \frac{1}{z}\right) +1 = 0.$$
Writing $u = z + \frac{1}{z}$, we obtain $u^2 = z^2+\frac{1}{z^2} + 2$, that is $z^2+\frac{1}{z^2} = u^2-2$.
$u^3 = z^3 + \frac{1}{z^3} + 3(z + \frac{1}{z})$, so $z^3 + \frac{1}{z^3} = u^3 - 3u$.
Therefore
\begin{align*}
\Phi_7(z) = 0 &\iff \exists u \in \mathbb{C}, u =z + \frac{1}{z}\ \mathrm{and} \ (u^3 - 3u) + (u^2 - 2) +u+1 = 0\\
&\iff\exists u \in \mathbb{C}, u =z + \frac{1}{z}\ \mathrm{and} \ f(u)=u^3 + u^2 - 2u -1 = 0\\
\end{align*}
Applying this equivalence to $z = e^{\frac{2ik\pi}{7}}, 1 \leq k \leq 3$, we obtain $$f(2\cos(2k\pi/7))=f(\zeta_7^k + \zeta_7^{-k})=0,\quad k = 1,2,3.$$
These 3 roots of $f$ are distinct, since the function $\cos$ is strictly decreasing on $[0,\pi]$.
Therefore
\begin{align*}
f = x^3+x^2-2x-1 &= (x -2\cos(2\pi/7)) (x -2\cos(4\pi/7)) (x -2\cos(6\pi/7))\\
&= (x-\zeta_7-\zeta_7^{-1}) (x-\zeta_7^2-\zeta_7^{-2}) (x-\zeta_7^3-\zeta_7^{-3})
\end{align*}
\item[(b)] $L$ is the splitting field of $f$ over $\mathbb{Q}$, so by definition $$L = \mathbb{Q}(\zeta_7+\zeta_7^{-1},\zeta_7^2+\zeta_7^{-2},\zeta_7^3+\zeta_7^{-3}).$$
Therefore $L\supset \mathbb{Q}$, and as the three roots of $f$ lie in $\mathbb{Q}[\zeta_7]$,
$$\mathbb{Q} \subset L \subset \mathbb{Q}(\zeta_7).$$
As $\zeta_7^7 = 1 \in \mathbb{Q}, \mathbb{Q}(\zeta_7)$ is by defintion a radical extension of $\mathbb{Q}$, so $\Q \subset L$ is a solvable extension.
\ee
\end{proof}
\paragraph{Ex. 8.2.2}
{\it In the situation of Example 8.2.3, assume that $\Q \subset L$ is radical. Prove that $L = \Q(\gamma)$, where $\gamma^m \in \Q$ for some $m\geq 3$.
}
\begin{proof}
$f(x) = x^3+x^2-2x-1 \in \mathbb{Q}[x]$: $\sigma_1=-1,\sigma_2=-2,\sigma_3=1$.
Note that $f$ of degree 3 is irreducible over $\mathbb{Q}$, otherwise it would have a rational root $\alpha =p/q \in \mathbb{Q}, p \wedge q = 1,q>0$.
Then $p^3 + p^2 q - 2 pq^2-q^3 = 0$, so $p\mid q^3,p\wedge q = 1$, therefore $p\mid 1$, and similarly $q \mid 1$, thus $\alpha = \pm1$, but neither 1 nor $-1$ is a root of $f$, so $f$ is irreducible.
By Exercise 8.2.1, the splitting field of $f$ over $\Q$ is
$$L = \mathbb{Q}(\zeta_7+\zeta_7^{-1},\zeta_7^2+\zeta_7^{-2},\zeta_7^3+\zeta_7^{-3})\subset \R.$$
$\mathrm{discr}(f) = -4\sigma_2^3-27\sigma_3^2+\sigma_1^2\sigma_2^2-4\sigma_1^3\sigma_3+18\sigma_1\sigma_2\sigma_3 = 32-27+4+4+36 = 49$.
$\mathrm{discr}(f) =49$ is a square in $\mathbb{Q}$, so $\mathrm{Gal}(L/ \mathbb{Q})$ is the cyclic group $C_3 \simeq \mathbb{Z}/3\mathbb{Z}$ (Prop. 7.4.2), and so $[L : \mathbb{Q}] = 3$.
Assume that the extension $\mathbb{Q}\subset L$ is radical.
As $[L : \mathbb{Q}] = 3$, it doesn't exists any strict sub-extension of $\mathbb{Q}\subset L$, so the definition of a radical extension imply the existence of $\gamma \in L$ and of an integer $m > 1$ such that $L = \mathbb{Q}(\gamma)$ and $\gamma^m \in \mathbb{Q}$. As the extension is of degree 3, it it not a quadratic extension, so $m\geq 3$.
We conclude the reasoning (Example 8.2.3):
Let $p$ be the minimal polynomial of $\gamma$ over $\mathbb{Q}$. Then $\deg(p) = [L:\mathbb{Q}] = 3$. As $\gamma$ is a root of $x^m - \gamma^m \in \mathbb{Q}[x]$, $p$ divides $x^m - \gamma^m$.
As the extension $\mathbb{Q} \subset L$ is a Galois extension, all the roots of $p$ are in $L$ so are real since $L \subset \R$. Thus the three real distinct roots of $p$ are among the roots of $x^m- \gamma^m$, that is $$\gamma, \zeta_m \gamma,\zeta_m^2 \gamma,\cdots,\zeta_m^{m-1} \gamma,\qquad \gamma \in \mathbb{R}.$$
But this is impossible since at most two of these roots are real.
Conclusion: $\mathbb{Q}\subset L$ is not a radical extension (but $\mathbb{Q} \subset L \subset \mathbb{Q}(\zeta_7)$, so $\mathbb{Q} \subset L$ is solvable).
\end{proof}
\paragraph{Ex. 8.2.3}
{\it Here you will prove two properties of compositums.
\be
\item[(a)] Prove that the compositum $K_1K_2$ exits.
\item[(b)] Prove (8.3)
\ee
}
\begin{proof}
\be
\item[(a)] Let $A$ be the set of the subfields of $L$ containing $K_1$ and $K_2$. Then $A \neq \emptyset$, since $L \in A$.
The intersection of the subfields of $L$ containing $K_1$ and $K_2$ is a subfield of $L$ containing $ K_1$ and $K_2$, thus
$\bigcap\limits_{X \in A} X$ is an element of $A$, and this is the smallest element of $A$ for inclusion.
So there exists a smallest subfield of $L$ containing $K_1$ and $K_2$, that is $K_1K_2$.
\item[(b)] Suppose that $K_1 = F(\alpha_1,\alpha_2,\cdots,\alpha_n), K_2 = F(\beta_1,\beta_2,\cdots,\beta_m)$.
$K = K_1K_2$ contains $K_1$ and $K_2$, so contains $F$ , and also $\alpha_1,\alpha_2,\cdots,\alpha_n,\beta_1,\beta_2,\cdots,\beta_m$, therefore $K \supset F(\alpha_1,\alpha_2,\cdots,\alpha_n,\beta_1,\beta_2,\cdots,\beta_m)$.
Conversaly, as $K = F(\alpha_1,\alpha_2,\cdots,\alpha_n,\beta_1,\beta_2,\cdots,\beta_m)$ is a subfield of $L$ containing $K_1 = F(\alpha_1,\alpha_2,\cdots,\alpha_n)$ and $K_2 = F(\beta_1,\beta_2,\cdots,\beta_m)$, $K \in A$, so $K$ contains the smallest element of $A$, which is $K_1K_2$.
Conclusion: $K_1K_2 = F(\alpha_1,\alpha_2,\cdots,\alpha_n,\beta_1,\beta_2,\cdots,\beta_m)$.
\ee
\end{proof}
\paragraph{Ex. 8.2.4}
{\it This exercise is concerned with the proof of Proposition 8.2.6.
\be
\item[(a)] Show that $K = F(\alpha_1,\ldots,\alpha_r)$ is the Galois closure of $F \subset L$.
\item[(b)] Prove that the conjugates of $L$ in $M$ are the fields $F(\alpha_i)$ for $i=1,\ldots,r$.
\ee
}
\begin{proof}
\be
\item[(a)] $F\subset L \subset M$, and $F \subset M$ is a Galois extension.
The Theorem of the Primitive Element implies that $L = F(\alpha)$ for some $\alpha \in L$.
Since $F \subset M $ is a Galois extension, the minimal polynomial $h$ of $\alpha$ over $F$ is separable and splits completely over $M$, say $h(x) = (x-\alpha_1)\cdots(x-\alpha_r)$, where $\alpha_1 = \alpha$.
$K = F(\alpha_1,\alpha_2,\cdots,\alpha_r)$ is a Galois extension of $F$ containing $L$, since $K$ is the splitting field of a separable polynomial $h \in F[x]$.
We show that $F\subset K$ is the Galois closure of $F \subset L$.
$\bullet$ $F\subset K$ is a Galois extension (as previously proved) and $K \supset L$.
$\bullet$ Suppose that $L \subset K'$ is another extension such that $K'$ is Galois over $F$.
As $F\subset K'$ is normal, and $\alpha \in L \subset K'$, the polynomial $h \in F[x]$ with a root in $K'$ splits completely over $K'$:
$h(x) = (x-\beta_1)\cdots(x-\beta_r)$, where $\beta_1 = \alpha$, and $\beta_i \in K', 1 \leq i \leq r$.
Let $K'' = F(\beta_1,\cdots,\beta_r)$.
$K$ and $K''$ are both splitting fields of $h$ over $F$. By the Theorem of Unicity of the splitting field, there exist an isomorphism $\varphi : K\to K''$ which is the identity on $F$. As $K'' \subset K'$, $\varphi$ gives a field homomorphism of $K$ in $K'$ which is the identity on $F$.
By definition of a Galois closure, $F \subset K$ is a Galois closure of $F \subset L$.
Note: in Exercise 7.3.13, we have seen that there exists a {\it unique} sub-extension of $F \subset M$ which is a Galois closure of $F \subset L$, so $K$ is the unique Galois closure of $F\subset L$ included in $M$, the smallest subfield $K$ of $M$ such that $F\subset L \subset K \subset M$ which is Galois over $F$. If $\alpha \in K'$, $L\subset K'$, and $F\subset K'$ is a Galois extension, then $\alpha = \alpha_1 \in K'$ implies that $\alpha_1,\ldots,\alpha_r \in K'$, so $K =F(\alpha_1,\ldots,\alpha_r) \subset K'$.
\item[(b)] If $L'$ is a conjugate field of $L$ in $M$, there exists a $F$-automorphism $\sigma$ of the field $M$ such that $L' = \sigma(L)$, $\sigma \in \mathrm{Gal}(M/F)$.
As $\sigma$ sends $\alpha$ on a conjugate of $\alpha$, $\sigma(\alpha) = \alpha_i,\ 1 \leq i \leq r$, and
$$L' = \sigma(L) = \sigma(F(\alpha)) = F(\alpha_i).$$
Indeed, an element $u \in L$ is of the form $u = g(\alpha), g \in F[x]$, so $\sigma(u) = g(\alpha_i) \in F(\alpha_i)$, therefore $\sigma(L) \subset F(\alpha_i)$.
Conversely, an element $v \in F(\alpha_i)$ is of the form $v = g(\alpha_i)$. So $v$ is the image of $u = g(\alpha) \in L$ by $\sigma$, so $L' = \sigma(L) = F(\alpha_i)$, therefore the conjugates of $L$ are among the $F(\alpha_i), 1 \leq i \leq r$.
Moreover, if $L'' = F(\alpha_i)$, for any $i=1,\ldots,r$, we know that there exists $\sigma \in \mathrm{Gal}(M/F)$ such that $\sigma(\alpha) = \alpha_i$ (Prop.5.1.8). As previously proved, $L'' = F(\alpha_i) = \sigma(F(\alpha)) = \sigma(L)$.
The conjugate fields of $L$ in $M$ are the $r$ subfields $F(\alpha_i), 1 \leq i \leq r$.
\ee
\end{proof}
\paragraph{Ex. 8.2.5}
{\it This exercise will complete the proof of part (b) of Lemma 8.2.7.
\be
\item[(a)] Prove (8.5).
\item[(b)] Prove that the field $F'_n$ defined in (8.4) is the compositum $K_1K_2$.
\ee
}
\begin{proof}
\begin{enumerate}
\item[(a)]
By hypothesis, $F\subset K_2$, so $F_0=F \subset K_2 = F'_0$.
Using induction, if we suppose that $F_{i-1} \subset F'_{i-1}$ for some $i, 1\leq i < n$, then $F_{i-1}(\gamma_i) \subset F'_{i-1}(\gamma_i)$, therefore $F_i \subset F'_i$.
Conclusion: $\forall i, 0 \leq i \leq n, \ F_i \subset F'_i$.
Consequently, $\gamma_i^{m_i} \in F_{i-1} \subset F'_{i-1}, \ i=1,\cdots,n$, so $K_2 \subset F'_n$ is a radical extension.
\item[(b)]
We show that $K'_n = K_1K_2$.
$K_1 = F(\gamma_1,\cdots,\gamma_n)$, and $F \subset K_2$, therefore $K_1K_2 = K_2(\gamma_1,\cdots,\gamma_n) = F'_n$.
Indeed, $F'_n = K_2(\gamma_1,\cdots,\gamma_n)$ by construction, and
\be
\item[$\bullet$]
\begin{align*}
& K_2(\gamma_1,\cdots,\gamma_n) \supset K_2,\\
& K_2(\gamma_1,\cdots,\gamma_n) \supset F(\gamma_1,\cdots,\gamma_n) = K_1,
\end{align*}
therefore $F'_n \supset K_1, F'_n\supset K_2$.
\item[$\bullet$]If $K$ is any subfield of $L$ which contains $K_1,K_2$, $K \supset K_1 =F(\gamma_1,\cdots,\gamma_n)$, so
\begin{center}
$K \supset \{\gamma_1,\cdots,\gamma_n\}$ and $ K \supset K_2,$
\end{center}
therefore $K \supset K_2(\gamma_1,\cdots,\gamma_n) = F'_n$.
\ee
So $F'_n$ is the smallest subfield of $L$ which contains $K_1$ and $K_2$,
$$ K_1K_2 = F'_n.$$
We conclude that $K_1K_2$ is a radical extension of $K_2$.
\end{enumerate}
\end{proof}
\paragraph{Ex. 8.2.6}
{\it Suppose we have finite extensions $F \subset L \subset M$ and $\sigma \in \Gal(M/F)$, and assume that $F\subset L$ is radical. Prove that $F \subset\sigma L$ is also radical.
}
\begin{proof}
Let $\sigma \in \Gal(M/F)$.
There exists an ascending series $(F_i)_{0\leq i \leq n}$ of subfields of $L$ such that
$$F = F_0 \subset F_1=F_0(\gamma_1) \subset \cdots \subset F_i = F_{i-1}(\gamma_i)\subset \cdots \subset F_n = F_{n-1}(\gamma_n)=L,$$
where $\gamma_i^{m_i} \in F_{i-1}$.
Write $F'_i = \sigma F_i,\ i=0,\cdots,n$, and $\gamma'_i = \sigma(\gamma_i)$. Then $F'_0 = \sigma F_0 =\sigma F = F$ and $F'_n = \sigma L$.
As $F_i = F_{i-1}(\gamma_i), \ i=1,\cdots,n$, then $$F'_i = \sigma F_i = \sigma(F_{i-1}(\gamma_i)) = (\sigma F_{i-1})(\sigma (\gamma_i)) = F'_{i-1}(\gamma'_{i-1}).$$
Therefore
$$F = F'_0 \subset F'_1=F'_0(\gamma'_1) \subset \cdots \subset F'_i = F'_{i-1}(\gamma'_i)\subset \cdots \subset F'_n = F'_{n-1}(\gamma_n)=\sigma L.$$
Moreover, $(\gamma'_{i})^{m_i} = \sigma(\gamma_i)^{m_i} = \sigma(\gamma_i^{m_i}) \in \sigma(F_{i-1}) = F'_{i-1}$.
Consequently $F \subset \sigma L$ is a radical extension.
\end{proof}
\paragraph{Ex. 8.2.7}
{\it Suppose that we have extensions $F \subset K_1 \subset L$ and $F\subset K_2 \subset L$ such that $F\subset K_1$ and $F \subset K_2$ are Galois. Prove that $F \subset K_1K_2$ is Galois. This will show that the compositum of two Galois extensions is again Galois.
}
\begin{proof}
$F\subset K_1$ and $F \subset K_2$ are Galois extensions, so are separable extensions. By the Theorem of the Primitive Element, $K_1 = F(\alpha), K_2=F(\beta), \ \alpha \in K_1,\beta \in K_2$, with $\alpha, \beta$ separable over $F$, therefore $K_1K_2 = F(\alpha,\beta)$ is separable (Proposition 7.1.6).
By Proposition 7.1.7, the Galois closure exists: let $M$ be a Galois closure of $F \subset K_1K_2$, and $\sigma$ be any element of $\Gal(M/F)$.
Let $\gamma \in K_1 K_2 = F(\alpha,\beta)$. Then $\sigma(\gamma)\in F(\sigma(\alpha),\sigma(\beta))$, and $\sigma(\alpha) \in K_1, \sigma(\beta)\in K_2$ since $F\subset K_1$ and $F \subset K_2$ are normal extensions. Therefore $\sigma(\gamma) \in K_1K_2$.
Consequently $$\sigma (K_1 K_2) \subset K_1K_2.$$
Applying this result to $\sigma^{-1}$, we obtain $\sigma^{-1}(K_1K_2) \subset K_1K_2$, therefore $K_1K_2 \subset \sigma(K_1K_2)$, and so
$$\forall \sigma \in \Gal(M/F),\ \sigma (K_1K_2) = K_1 K_2.$$
By Theorem 7.2.5, we conclude that $K_1K_2$ is a Galois extension of $F$.
\end{proof}
\subsection{SOLVABLE EXTENSIONS AND SOLVABLE GROUPS}
\paragraph{Ex. 8.3.1}
{\it Let $m$ be a positive integer, and let $L$ be a field of characteristic $0$. Then let $L \subset M$ be the splitting field of $x^m-1\in L[x]$.
\be
\item[(a)] Prove that $x^m-1$ is separable.
\item[(b)] Prove that the roots of $x^m-1$ lying in $M$ form a group under multiplication.
\ee
}
\begin{proof}
\begin{enumerate}
\item[(a)]
Let $f=x^m-1,\ m\in \N^*$. Then $f'=m x^{m-1}$ is relatively prime with $f$. Indeed $m\neq 0$ in $L$ since the characteristic of $L$ is $0$, and $-f +m^{-1} x f'= -x^m+1 + x^m = 1$ is a B\'ezout's relation between $f$ and $f'$. Therefore (Prop. 5.3.2), $f$ is a separable polynomial, so $x^m-1$ has $m$ distinct roots in $M$, the splitting field of $f$ over $L$.
\item[(b)] We show that $\mathbb{U}_m = \{\alpha \in M\ \vert \ \alpha^m=1\}$, the set of the $m$ roots of $f$ in $M$, is a subgroup of $M^*$.
\be
\item[$\bullet$] $1^m = 1$, therefore $1\in \mathbb{U}_m \neq \emptyset$.
\item[$\bullet$] If $\alpha, \beta \in \mathbb{U}_m$, then $(\alpha \beta^{-1})^m = \alpha^m (\beta^m)^{-1} = 1$, therefore $\alpha \beta^{-1} \in \mathbb{U}_m$.
\ee
The roots of $x^m-1$ in $M$ form a group under multiplication, subgroup of the multiplicative group of a field, so it is cyclic.
\end{enumerate}
\end{proof}
\paragraph{Ex. 8.3.2}
{\it Assume that $F\subset L$ is a Galois extension and that $F$ has characteristic 0. Also, consider the extension $L \subset L(\zeta)$ obtained by adjoining a primitive $m$th root of unity. Prove that $F\subset L(\zeta)$ is Galois.
}
\begin{proof}
Let $\zeta$ a primitive root of $f = x^m-1$, in other words a generator of $\mathbb{U}_m$.
As the characteristic of $F$ is 0, by Exercise 1, $f$ is a separable polynomial, and $f=(x-1)(x-\zeta)\ldots(x-\zeta^{m-1})$
$F(\zeta) = F(1,\zeta,\cdots,\zeta^{m-1})$ is the splitting field over $F$ of the separable polynomial $f$, so $F\subset F(\zeta)$ is a Galois extension. By hypothesis, $F\subset L$ is also a Galois extension. By the Theorem of the Primitive Element, there exists $\alpha\in L$ such that $L=F(\alpha)$. Then the compositum of $L=F(\alpha)$ and $F(\zeta)$ is $F(\alpha,\zeta) = L(\zeta)$. By Exercise 8.2.7, this is a Galois extension of $F$.
\begin{center}
$F \subset L(\zeta)$ is a Galois extension.
\end{center}
\end{proof}
\paragraph{Ex. 8.3.3}
{\it Prove (8.9), where $\zeta$ is a primitive $p$th root of unity and $1\leq i \leq p-1$.
}
\begin{proof}
As $\zeta^p=1,\zeta^i\neq 1 (1\leq i \leq p-1)$,
$$1+\zeta^{-i}+\zeta^{-2i}+\cdots+\zeta^{-(p-1)i} = \frac{1-\zeta^{-ip}}{1-\zeta^{-i}} = 0.$$
\end{proof}
\paragraph{Ex. 8.3.4}
{\it Consider the extension $F_{i-1} \subset F_i$ of (8.11). In the discussion following (8.11), we showed that this extension is Galois. We now describe its Galois group.
\be
\item[(a)] Let $\sigma \in \Gal(F_i/F_{i-1})$. Show that there is a unique integer $0 \leq l \leq m_i-1$ such that $\sigma(\gamma_i) = \zeta_i^l \gamma_i$.
\item[(b)] Show that $\sigma \mapsto [l]$ defines a one-to-one homomorphism $\Gal(F_i/F_{i-1}) \to \Z/m_i\Z$, where $[l]$ is the congruence class of $l$ modulo $m_i$.
\item[(c)] Conclude that $\Gal(F_i/F_{i-1})$ is cyclic.
\ee
}
\begin{proof}
Recall the context: the extension $F_{i-1} \subset F_i$ is a Galois extension,
where $F_i = F_{i-1}(\gamma_i) = F(1,\gamma_i,\zeta^i \gamma_i,\cdots,\zeta^{m_i-1} \gamma_i)$ is the splitting field of $x^{m_i} - a_i$ over $F_i$,
$a_i = \gamma_i^{m_i} \in F_{i-1}$ and $\zeta_i$ is a $m_i$th primitive root of unity.
\begin{enumerate}
\item[(a)]
Let $\sigma \in \Gal(F_i/F_{i-1})$. As $\gamma_i$ is a root of $x^{m_i}-a_i \in F_{i-1}(x)$, $\sigma(\gamma_i)$ is another root, so
$$\sigma(\gamma_i) = \zeta_i^l \gamma_i,\qquad 0 \leq l \leq m_i-1.$$
Such an $l$ is unique, since $ \zeta_i^l \gamma_i = \zeta_i^{l'}\gamma_i, \ 1\leq l,l' \leq m_i$ implies $\zeta_i^l = \zeta_i^{l'}$, thus $m_i \mid l'-l$, so $\l \equiv l' \pmod{ m_i}$. As $\vert l' -l \vert \leq m_i-1$, then $l'-l=0$.
\item[(b)]
Let
$$
\varphi :
\left\{
\begin{array}{ccc}
\Gal(F_i/F_{i-1})&\to &\Z/m_i\Z \\
\sigma & \mapsto & [l] : \sigma(\gamma_i) = \zeta_i^l \gamma_i
\end{array}
\right.
$$
This mapping is well defined, since $l$ is known modulo $m_i$.
$\bullet$ We verify that $\varphi$ is a group homomorphism.
Let $\sigma,\tau \in \Gal(F_i/F_{i-1})$. Then $\sigma(\gamma_i) = \zeta_i^l \gamma_i, \tau(\gamma_i) = \zeta_i^k \gamma_i$.
Thus $(\sigma \tau)(\gamma_i) = \sigma(\zeta_i^k \gamma_i) = \sigma(\zeta_i)^k \sigma(\gamma_i) = \zeta_i^k \sigma(\gamma_i)$, since $\zeta_i \in F$, therefore $\zeta_i \in F_{i-1}$.
$(\sigma \tau)(\gamma_i) = \zeta_i^k \zeta_i^l \gamma_i = \zeta_i^{l+k} \gamma_i$, so $\varphi(\sigma \tau) = [l+k] = [l]+[k] = \varphi(\sigma) + \varphi(\tau)$.
$\bullet$ $\varphi$ is an injective homomorphism:
if $\sigma \in \ker(\varphi)$, $[l] = [0]$, so $\sigma(\gamma_i) = \zeta_i^l \gamma_i = \gamma_i$. As $\sigma$ fixes the elements of $F_{i-1}$ and also $\gamma_{i}$, this is the identity on $F_i=F_{i-1}(\gamma_i)$. $\ker(\varphi) = \{e\}$, and $\varphi$ is injective.
\item[(c)]
Therefore $\Gal(F_i/F_{i-1})$ is isomorphic to a subgroup $H$ of $\Z/m_i\Z$. As every subgroup of a cyclic group is cyclic, $H$ is cyclic, so
\begin{center}
$\Gal(F_i/F_{i-1})$ is a cyclic group.
\end{center}
\end{enumerate}
\end{proof}
\paragraph{Ex. 8.3.5}
{\it Suppose that we have extensions $F \subset F_{i-1}\subset F_i \subset L$ such that $L$ is Galois over $F$ and $F_i$ is Galois over $F_{i-1}$. Prove that $|\Gal(F_i/F_{i-1})|$ divides $|\Gal(L/F)|$.
}
\begin{proof}
$F \subset F_{i-1} \subset F_i \subset L$.
As $F\subset L$ is a Galois extension, $F_i \subset L$ and $F_{i-1} \subset L$ are also Galois, we have
$$[L:F_i] =\vert \Gal(L/F_i) \vert,\qquad [L:F_{i-1}] =\vert \Gal(L/F_{i-1}\vert.$$
$\Gal(F_i/F_{i-1})\simeq \Gal(L/F_{i-1})/\Gal(L/F_i)$, thus $\vert \Gal(F_i/F_{i-1}) \vert$ divides $\vert \Gal(L/F_{i-1}\vert = [L/F_{i-1}]$.
As $ \vert \Gal(L/F) \vert= [L:F] = [L:F_{i-1}][F_{i-1} : F]$,
\begin{center}
$\vert \Gal(F_i/F_{i-1}) \vert$ divides $\vert \Gal(L/F)\vert $.
\end{center}
\end{proof}
\paragraph{Ex. 8.3.6}
{\it Let $L$ be a field containing a primitive $m$th root of unity $\zeta$ and let $n$ be a positive divisor of $m$. Prove that $\zeta^{m/n}$ is a primitive $n$th root of unity.
}
\begin{proof}
For all $k\in \Z$,
$$(\zeta^{m/n})^k = 1 \iff \zeta^{mk/n} = 1 \iff m \mid mk/n \iff n\mid k.$$ The order of $\zeta^{m/n}$ is so $n$. In other words, $ \zeta^{m/n}$ is a primitive $n$th root of unity.
\end{proof}
\paragraph{Ex. 8.3.7}
{\it Let $F \subset L$ be Galois and solvable (with $F$ of characteristic 0). This exercise will consider a variation of Corollary 8.3.4. Let $p_1,\ldots,p_r$ be the distinct primes dividing $[L:F]$.
\be
\item[(a)] Show that $F$ contains a primitive $(p_1\cdots p_r)$th root of unity if and only if $F$ contains a primitive $p_i$th root of unity for $i=1,\ldots,r$.
\item[(b)] Prove that $F\subset L$ is radical when $F$ contains a primitive $(p_1\cdots p_r)$th root of unity.
\item[(c)] Prove that $F \subset L(\zeta)$ is radical, where $\zeta$ is a primitive $(p_1\cdots p_r)$th root of unity.
\ee
}
\begin{proof}
$F\subset L $ a solvable Galois extension, with $F$ of characteristic 0. $p_1,\cdots,p_r$ are the distinct prime numbers which divide $[L:F]$.
\begin{enumerate}
\item[(a)]
{\bf Lemma 1.} {\it Suppose that two elements $a,b$ of an Abelian group $G$ are of respective order $p,q$, where $p,q$ are relatively prime. Then the order of $ab$ is $pq$. }
{\it Proof of Lemma 1:}
$a^p= b^q=e$, therefore $(ab)^{pq} = (a^p)^q (b^q)^p = e$.
For all $k\in \Z$, since $p \wedge q = 1$,
$$(ab)^k = e \Rightarrow (ab)^{qk} = e\Rightarrow a^{qk} b^{qk}=e \Rightarrow a^{qk} = e \Rightarrow p \mid qk \Rightarrow p \mid k.$$
Similarly
$$(ab)^k = e \Rightarrow (ab)^{pk} = e\Rightarrow a^{pk} b^{pk}=e \Rightarrow b^{pk} = e \Rightarrow q \mid pk \Rightarrow q \mid k.$$
Consequently, using again $p\wedge q = 1$, $$(ab)^k = e \Rightarrow (p \mid k \ \mathrm{and} \ q \mid k) \Rightarrow pq\mid k.$$
To conclude,
$$\forall k \in \Z,\ (ab)^k=e \iff pq \mid k,$$
The order of $ab$ is thus $pq$. \qed
\bigskip
{\bf Lemma 2.} {\it If $G$ is an Abelian group and $c_1, \cdots,c_r \in G$ are of respective order $p_1, \cdots, p_r$, where $p_1, \cdots,p_r$ are pairwise relatively prime, then the order of $c = c_1\cdots c_r$ is $p_1\cdots p_r$.}
{\it Proof of Lemma 2:} Using the induction hypothesis $\vert c_1\cdots c_k \vert = p_1\cdots p_k, \ k<r$, and applying Lemma 1 to $a=c_1\cdots c_k$ of order $p_1\cdots p_k$, and $b = c_{k+1}$ of order $p_{k+1}$, where $p_1\cdots p_k \wedge p_{k+1}=1$, then $\vert c_1\cdots c_{k+1} \vert = p_1\cdots p_{k+1}$. \qed
$\bullet$ Suppose that $F$ contains a root of unity $c$ of order $n = p_1\cdots p_r$. Write $c_i = c^{n/p_i},\ i=1,\cdots r$. Then $c_i\in F$ and Exercise 6 proves that $c_i$ is of order $p_i$.
$\bullet$ Conversely, suppose that $F$ contains some elements $c_i$ of order $p_i$, for all $i, 1\leq i \leq r$. The $p_i$ are distinct prime numbers, so are pairwise relatively prime.
Let $c = c_1\cdots c_r$. Lemma 2 applied in the Abelian group $G = F^*$ shows that the order of $c_1\cdots c_r$ is $p_1\cdots p_r$.
Conclusion: if $p_1,\cdots,p_r$ are distinct prime numbers, $F$ contains a primitive $(p_1\cdots p_r)$th of unity if and only if it contains primitive $p_i$th roots of unity for all $i = 1,\cdots,r$.
\item[(b)]
Suppose that $F$ contains a primitive $p_1\cdots p_r$th root of unity $\zeta_{p_1\cdots p_r}$. By part (a), $F$ contains also $p_i$th primitive roots of unity $\zeta_{p_i}$ for $ i=1,\cdots,r$.
So the condition (8.12) is satisfied: $F$ contains a $p$th primitive root of unity for all $p$ dividing $\Gal(L/F) = [L:F]$. The part $(b)\Rightarrow(a)$ (special case) in the proof of Theorem 8.3.3 shows that $F\subset L$ is a radical extension.
\item[(c)]
Let $\zeta$ a $p_1\cdots p_r$th primitive root of unity.
As proven in the text, there exists an injective group homomorphism (8.13)
$$\Gal(L(\zeta)/F(\zeta)) \to\Gal(L/F)$$
thus $\vert \Gal(L/F) \vert$ is a multiple of $\vert \Gal(L(\zeta)/F(\zeta)) \vert$.
So $F(\zeta)$ contains a $p$th primitive root of unity for all $p$ dividing $\vert \Gal(L(\zeta)/F(\zeta)) \vert$, and then the part (b) proves that $F(\zeta)\subset L(\zeta)$ is a radical extension. As $F \subset F(\zeta)$ is radical, by Lemma 8.2.7(a), $F \subset L(\zeta)$ is a radical extension.
\end{enumerate}
\end{proof}
\paragraph{Ex. 8.3.8}
{\it This exercise concerns the details of our derivation of Cardan's formulas.
\be
\item[(a)] Use the computational methods of Section 2.3 the formulas for $\alpha_1^3$ and $\beta_1$ stated in the text.
\item[(b)] Prove (8.15).
\ee
}
\begin{proof}
\begin{enumerate}
\item[(a)]
We know that $\alpha_1 = x_1+\omega^2 x_2+\omega x_3$, so
\begin{align*}
\alpha_1^3 =&\ ( x_1+\omega^2 x_2+\omega x_3)^3\\
=&\quad x_1^3+x_2^3+x_3^3\\
&+3\omega\, (x_1^2x_3+x_2^2x_1+x_3^2x_2)\\
&+3\omega^2(x_1^2x_2+x_2^2x_3+x_3^2x_1)\\
&+6x_1x_2x_3.
\end{align*}
Thus $\alpha_1^3$ is of the form
$$\alpha_1^3 = p + 3\omega r + 3 \omega^2 s + 6q,$$
where
\begin{align*}
p &=x_1^3+x_2^3+x_3^3 = \sigma_1^3-3\sigma_1\sigma_2+3\sigma_3,\\
q &= x_1x_2x_3 = \sigma_3,\\
r&=x_1^2x_3+x_2^2x_1+x_3^2x_2,\\
s&=x_1^2x_2+x_2^2x_3+x_3^2x_1.
\end{align*}
Since $r,s$ are fixed by the permutations of $A_3$, they must be of the form $A+B\sqrt{\Delta}, \ A,B\in K = \Q(\sigma_1,\sigma_2,\sigma_3)$, where we choose
\begin{align*}
\sqrt{\Delta} &= (x_2-x_1)(x_3-x_2)(x_3-x_1)\\
&=(x_1^2x_3+x_2^2x_1+x_3^2x_2) -(x_1^2x_3+x_2^2x_1+x_3^2x_2)\\
&=r-s.
\end{align*}
The pair $(r,s)$ is thus the solution of the system
\begin{align*}
r-s&=\sqrt{\Delta},\\
r+s &= \sigma_1\sigma_2 - 3 \sigma_3.
\end{align*}
Therefore
\begin{align*}
r &= \frac{1}{2}(\sigma_1 \sigma_2 - 3 \sigma_3 +\sqrt{\Delta}),\\
s &= \frac{1}{2}(\sigma_1 \sigma_2 - 3 \sigma_3 -\sqrt{\Delta}).
\end{align*}
Then
\begin{align*}
\alpha_1^3 =&\ ( x_1+\omega^2 x_2+\omega x_3)^3\\
=&+\sigma_1^3-3\sigma_1\sigma_2+3\sigma_3\\
&+\frac{3}{2}\omega\,( \sigma_1 \sigma_2 - 3 \sigma_3 +\sqrt{\Delta})\\
&+\frac{3}{2}\omega^2(\sigma_1 \sigma_2 - 3 \sigma_3 -\sqrt{\Delta})\\
&+6\sigma_3\\
=& \sigma_1^3-3\sigma_1\sigma_2+9\sigma_3 -\frac{3}{2}(\sigma_1 \sigma_2 - 3\sigma_3)+\frac{3}{2}(\omega - \omega^2)\sqrt{\Delta}\\
=&\sigma_1^3 -\frac{9}{2} \sigma_1 \sigma_2 +\frac{27}{2} \sigma_3 +\frac{3\sqrt{3}}{2} i \sqrt{\Delta}
\end{align*}
Therefore
\begin{align*}
\alpha_1^3 &= -\frac{27}{2} q + \frac{3\sqrt{3}}{2} i \sqrt{\Delta}\\
&=\frac{27}{2} \left ( - q +\sqrt{\frac{-\Delta}{27} }\right )
\end{align*}
where
$$q = -\frac{2}{27} \sigma_1^3 +\frac{1}{3} \sigma_1 \sigma_2 - \sigma_3.$$
So
\begin{align*}
\alpha_1 &= x_1+\omega^2 x_2 + \omega x_3\\
&= 3\ \sqrt[3]{\frac{1}{2}\left ( -q +\sqrt{\frac{-\Delta}{27} }\right)},
\end{align*}
and
\begin{align*}
\beta_1 &= (2 3)\cdot \alpha_1\\
&= x_1+\omega^2 x_3 + \omega x_2\\
&= 3\ \sqrt[3]{\frac{1}{2}\left ( -q - \sqrt{\frac{-\Delta}{27} }\right)}.
\end{align*}
Indeed the same calculation gives $\beta_1$, by the exchange of $x_2$ with $x_3$, which sends $\sqrt{\Delta}$ on $- \sqrt{\Delta}$.
\item[(b)]
The system of equations
\begin{align*}
\sigma_1 &= x_1 + x_2 +x_3\\
\alpha_1 &= x_1+\omega^2 x_2+\omega x_1\\
\beta_1 &= x_1+\omega x_2+\omega^2 x_3,
\end{align*}
has for solution
\begin{align*}
x_1 &= \frac{1}{3} (\sigma_1 + \alpha_1+\beta_1)\\
x_2 &= \frac{1}{3} (\sigma_1 + \omega \alpha_1+ \omega^2\beta_1)\\
x_3 &= \frac{1}{3} (\sigma_1 + \omega^2 \alpha_1+ \omega \beta_1)\\
\end{align*}
And these are the Cardan's formula for the roots of $\tilde{f} = x^3 - \sigma_1 x^2+\sigma_2 x -\sigma_3$.
\end{enumerate}
\end{proof}
\subsection{SIMPLE GROUPS}
\paragraph{Ex. 8.4.1}
{\it Let $G$ be a non trivial finite Abelian group. Prove that $G$ is simple if and only if $G\simeq \Z/p\Z$ for some prime $p$.
}
\begin{proof}
Let $G$ be a non trivial finite Abelian group.
\be
\item[$\bullet$] If $G \simeq \Z/p\Z$, $G$ is cyclic of order $p$. Every subgroup $H$ of $G$ has a cardinality dividing $p$, so its order is 1 or $p$, therefore $H = \{e\}$ or $H = G$. The only subgroups of $G$, normal or not, are $\{e\}$ or $G$. So $G$ is simple.
\item[$\bullet$] Suppose that $G$ is a non trivial finite Abelian simple group. As $G$ is Abelian, every subgroup of $G$ is normal in $G$, thus $G$ has no other subgroup that $\{e\}$ or $G$. As $G$ is not trivial, there exists $x \in G, x\neq e$. Then $\langle x \rangle$ is a subgroup of $G$ with cardinality greater than 1, therefore $G = \langle x \rangle$. $G$ being cyclic, it is isomorphic to $\Z/n\Z,\ n\in \N, n>1$. If $n$ was not prime, $n$ would be divisible by some integer $d, 1<d<n$. Then $\langle [d]_n \rangle$ is a subgroup of $\Z/n\Z$ of order $n/d, 1<n/d<n$, so $\Z/n\Z$ would have a non trivial subgroup, and also $G$. This is a contradiction, so $n=p$ is prime:
\begin{center}
$G\simeq \Z/p\Z$, $p$ prime.
\end{center}
\ee
\end{proof}
\paragraph{Ex. 8.4.2}
{\it Prove that $A_n$ is generated by 3-cycles when $n\geq 3$.
}
\begin{proof}
We notice that for all $i,j,k$ such that $i \ne j, j \ne k$, $(i\, j)(j \,k) = (i j k)$.
As every permutation in $A_n$ is the product of an even number of permutations, it is sufficient to prove that the product of $(i \, j)(k \, l), i\neq j, k \neq l$, is a product of 3-cycles.
$\bullet$ If $\{i,j\} ,\{k,l\}$ are disjointed, then $i,j,k,l$ are distincts, so $$(i\, j)(k\, l) = (i\, j)(j\, k)(j\, k)(k\, l) = (i \, j \, k) (\,j\, k\, \l).$$
$\bullet$ If $\{i,j\} ,\{k,l\}$ have one common element, say $i = k, i \neq l$, then $$(i\, j)(k\, l) = (i\, j)(i \,l) = (j\,i)(i\, l)= (j\, i \,l)$$ is a 3-cycle.
$\bullet$ If $\{i,j\} = \{k, l\}$, then $(i\,j)(k\,l) = (i \,j)^2 = ()= e$ is the empty product.
Conclusion: $A_n$ is generated by 3-cycles.
\end{proof}
\paragraph{Ex. 8.4.3}
{\it This exercise is concerned with the proof of Theorem 8.4.3.
\be
\item[(a)] Prove (8.17).
\item[(b)] Verify the identities (8.18), (8.19) and (8.20).
\item[(c)] Verify the conjugation identity (8.21).
\ee
}
{\bf Theorem.} {\it The alternating group $A_n$ is simple for all $n\geq 5$.}
\begin{proof}
Let $H \neq \{e\}$ a normal subgroup of $A_n$. It is sufficient to prove that $H = A_n$. We show first that $H$ contains a 3-cycle. As $H \neq \{e\}$, $H$ contains an even permutation $\sigma \neq e$. For any 3-cycle $(j_1 j_2 j_3)$, $(j_1 j_2 j_3) \in A_n$, and $H \lhd A_n$, so
$$\sigma^{-1} (j_1 \, j_2\, j_3)^{-1} \sigma\, (j_1\, j_2\, j_3) \in H.$$
\begin{enumerate}
\item[(a)]
Suppose that $j \not \in \{j_1,j_2,j_3\}$ and $ \sigma(j)\not \in \{j_1,j_2,j_3\}$.
We show then that $\sigma^{-1} (j_1 \, j_2\, j_3)^{-1} \sigma\, (j_1\, j_2\, j_3)$ fixes $j$.
As $j \not \in \{j_1,j_2,j_3\}$, then $(j_1\, j_2\, j_3)\cdot j = j$, and $[\sigma (j_1\, j_2\, j_3)]\cdot j = \sigma(j)$.
As $\sigma(j) \not \in \{j_1,j_2,j_3\}$, and $(j_1 \, j_2\, j_3)^{-1} = (j_3 j_2 j_1)$, $$[(j_1 \, j_2\, j_3)^{-1} \sigma\, (j_1\, j_2\, j_3)]\cdot j = (j_1 \, j_2\, j_3)^{-1}\cdot \sigma(j) = \sigma(j).$$
Therefore
$$[ \sigma^{-1} (j_1 \, j_2\, j_3)^{-1} \sigma\, (j_1\, j_2\, j_3)]\cdot j = \sigma^{-1}(\sigma(j)) = j.$$
This proves that this commutator is a permutation in $H$ that moves at most 6 elements of $\{1,\ldots,n\}$, the elements $j_1,j_2,j_3,\sigma^{-1}(j_1),\sigma^{-1}(j_2),\sigma^{-1}(j_3)$.
\item[(b)]
\begin{enumerate}
\item[$\bullet$] {\bf Case 1. } First suppose that one of the cycles in the cycle decomposition of $\sigma$ has a length $\geq 4$, say
$$\sigma = (i_1\,i_2\,i_3\,i_4 \cdots)(\cdots)\cdots.$$
In this case we claim that
$$\lambda = \sigma^{-1} (i_2\,i_3\,i_4)^{-1} \sigma\, (i_2\,i_3\,i_4) = (i_1\,i_3\,i_4).$$
We already know that $\lambda$ fixes every element $k$ which is not in
$$A = \{i_2,i_3,i_4,\sigma^{-1}({i_2}),\sigma^{-1}({i_3}),\sigma^{-1}({i_4})\} = \{i_1,i_2,i_3,i_4\}.$$
\be
\item[$\bullet$] If $k \not \in A $, then $\lambda(k) = k = (i_1\,i_3\,i_4)\cdot k $.