forked from jubnoske08/linear_algebra
-
Notifications
You must be signed in to change notification settings - Fork 0
/
chapter_6.tex
1090 lines (1035 loc) · 52.3 KB
/
chapter_6.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
\documentclass{extarticle}
\sloppy
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% PACKAGES %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\usepackage[10pt]{extsizes}
\usepackage{amsfonts}
\usepackage{amsthm}
\usepackage{amssymb}
\usepackage[shortlabels]{enumitem}
\usepackage{microtype}
\usepackage{amsmath}
\usepackage{mathtools}
\usepackage{commath}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% PROBLEM ENVIRONMENT %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\usepackage{tcolorbox}
\tcbuselibrary{theorems, breakable, skins}
\newtcbtheorem{prob}% environment name
{Problem}% Title text
{enhanced, % tcolorbox styles
attach boxed title to top left={xshift = 4mm, yshift=-2mm},
colback=blue!5, colframe=black, colbacktitle=blue!3, coltitle=black,
boxed title style={size=small,colframe=gray},
fonttitle=\bfseries,
separator sign none
}%
{}
\newenvironment{problem}[1]{\begin{prob*}{#1}{}}{\end{prob*}}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% THEOREMS/LEMMAS/ETC. %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newtheorem{thm}{Theorem}
\newtheorem*{thm-non}{Theorem}
\newtheorem{lemma}[thm]{Lemma}
\newtheorem{corollary}[thm]{Corollary}
\newtheorem*{definition}{Definition}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% MY COMMANDS %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newcommand{\Z}{\mathbb{Z}}
\newcommand{\R}{\mathbb{R}}
\newcommand{\C}{\mathbb{C}}
\newcommand{\F}{\mathbb{F}}
\newcommand{\bigO}{\mathcal{O}}
\newcommand{\Real}{\mathcal{Re}}
\newcommand{\poly}{\mathcal{P}}
\newcommand{\mat}{\mathcal{M}}
\DeclareMathOperator{\Span}{span}
\newcommand{\Hom}{\mathcal{L}}
\DeclareMathOperator{\Null}{null}
\DeclareMathOperator{\Range}{range}
\newcommand{\defeq}{\vcentcolon=}
\newcommand\widebar[1]{\mathop{\overline{#1}}}
\newcommand{\restr}[1]{|_{#1}}
\DeclarePairedDelimiterX{\inp}[2]{\langle}{\rangle}{#1, #2}
\DeclarePairedDelimiter\Mod{\lvert}{\rvert}
\DeclarePairedDelimiter\Norm{\lVert}{\rVert}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SECTION NUMBERING %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\renewcommand\thesection{\Alph{section}:}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% DOCUMENT START %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\title{\vspace{-2em}Chapter 6: Inner Product Spaces}
\author{\emph{Linear Algebra Done Right}, by Sheldon Axler}
\date{}
\begin{document}
\maketitle
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SECTION A
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Inner Products and Norms}
% Problem 1
\begin{problem}{1}
Show that the function that takes $\left((x_1,x_2), (y_1,y_2)\right)\in \R^2\times \R^2$ to $\Mod{x_1y_1} + \Mod{x_2y_2}$ is not an inner product on $\R^2$.
\end{problem}
\begin{proof}
Suppose it were. First notice
\begin{align*}
\inp{(1, 1) + (-1, -1)}{(1, 1)} &= \inp{(0, 0)}{(1, 1)}\\
&= \Mod{0\cdot 1} + \Mod{0\cdot 1}\\
&= 0.
\end{align*}
Next, since inner products are additive in the first slot, we also have
\begin{align*}
\inp{(1, 1) + (-1, -1)}{(1, 1)} &= \inp{(1, 1)}{(1, 1)} + \inp{(-1, -1)}{(1, 1)}\\
&= \Mod{1\cdot 1} + \Mod{1 \cdot 1} + \Mod{(-1)\cdot 1} + \Mod{(-1) \cdot 1}\\
&= 4.
\end{align*}
But this implies $0 = 4$, a contradiction. Hence we must conclude that the function does not in fact define an inner product.
\end{proof}
% Problem 3
\begin{problem}{3}
Suppose $\F=\R$ and $V\neq \{0\}$. Replace the positivity condition (which states that $\inp{v}{v}\geq0$ for all $v\in V$) in the definition of an inner product (6.3) with the condition that $\inp{v}{v} > 0$ for some $v\in V$. Show that this change in the definition does not change the set of functions from $V\times V$ to $\R$ that are inner products on $V$.
\end{problem}
\begin{proof}
Let $V$ be a nontrivial vector space over $\R$, let $A$ denote the set of functions $V\times V\to\R$ that are inner products on $V$ in the standard definition, and let $B$ denote the set of functions $V\times V\to \R$ under the modified definition. We will show $A = B$.\\
\indent Suppose $\inp{\cdot}{\cdot}_1\in A$. Since $V\neq\{0\}$, there exists $v\in V-\{0\}$. Then $\inp{v}{v}_1>0$, and so $\inp{\cdot}{\cdot}_1\in B$. Thus $A\subseteq B$.\\
\indent Conversely, suppose $\inp{\cdot}{\cdot}_2 \in B$. Then there exists some $v'\in V$ such that $\inp{v'}{v'}_2 > 0$. Suppose by way of contradiction there exists $u\in V$ is such that $\inp{u}{u}_2 < 0$. Define $w = \alpha u + (1- \alpha) v'$ for $\alpha\in\R$. It follows
\begin{align*}
\inp{w}{w}_2 &= \inp{\alpha u + (1- \alpha) v'}{\alpha u + (1- \alpha) v'}_2\\
&= \inp{\alpha u}{\alpha u}_2 + 2\inp{\alpha u}{(1 - \alpha)v'}_2 + \inp{(1 - \alpha)v'}{(1 - \alpha)v'}_2\\
&= \alpha^2\inp{u}{u}_2 + 2\alpha(1-\alpha)\inp{u}{v'}_2 + (1-\alpha)^2\inp{v'}{v'}_2.
\end{align*}
Notice the final expression is a polynomial in the indeterminate $\alpha$, call it $p$. Since $p(0) = \inp{v'}{v'}_2 > 0$ and $p(1) = \inp{u}{u}_2 < 0$, by Bolzano's theorem there exists $\alpha_0\in(0, 1)$ such that $p(\alpha_0) = 0$. That is, if $ w = \alpha_0u + (1 - \alpha_0)v'$, then $\inp{w}{w}_2 = 0$. In particular, notice $\alpha_0\neq 0$, for otherwise $w = v'$, a contradiction since $\inp{v'}{v'}_2 > 0$. Now, since $\inp{w}{w}_2 = 0$ iff $w = 0$ (by the definiteness condition of an inner product), it follows
\begin{equation*}
u = \frac{\alpha_0 - 1}{\alpha_0} v.
\end{equation*}
Letting $t = \frac{\alpha_0 - 1}{\alpha_0}$, we now have
\begin{align*}
\inp{u}{u}_2 &= \inp{tv'}{tv'}_2\\
&= t^2\inp{v'}{v'}_2\\
&> 0,
\end{align*}
where the inequality follows since $t\in(-1, 0)$ and $\inp{v'}{v'}_2 > 0$. This contradicts our assumption that $\inp{u}{u}_2 < 0$, and so we have $\inp{\cdot}{\cdot}_2\in A$. Therefore, $B\subseteq A$. Since we've already shown $A\subseteq B$, this implies $A = B$, as desired.
\end{proof}
% Problem 5
\begin{problem}{5}
Let $V$ be finite-dimensional. Suppose $T\in\Hom(V)$ is such that $\Norm{Tv}\leq \Norm{v}$ for every $v\in V$. Prove that $T-\sqrt{2}I$ is invertible.
\end{problem}
\begin{proof}
Let $v\in\Null(T - \sqrt{2}I)$, and suppose by way of contradiction that $v\neq 0$. Then
\begin{align*}
Tv - \sqrt{2}v = 0 &\implies Tv = \sqrt{2}v\\
&\implies \Norm{\sqrt{2}v}\leq \Norm{v}\\
&\implies \sqrt{2}\cdot \Norm{v}\leq \Norm{v}\\
&\implies \sqrt{2} \leq 1,
\end{align*}
a contradiction. Hence $v = 0$ and $\Null(T - \sqrt{2}I) =\{0\}$, so that $T-\sqrt{2}I$ is injective. Since $V$ is finite-dimensional, this implies $T-\sqrt{2}I$ is invertible, as desired.
\end{proof}
% Problem 7
\begin{problem}{7}
Suppose $u,v\in V$. Prove that $\Norm{au + bv} = \Norm{bu + av}$ for all $a,b\in\R$ if and only if $\Norm{u} = \Norm{v}$.
\end{problem}
\begin{proof}
$(\Rightarrow)$ Suppose $\Norm{au + bv} = \Norm{bu + av}$ for all $a,b\in\R$. Then this equation holds when $a = 1$ and $b = 0$. But then we must have $\Norm{u} = \Norm{v}$, as desired.\\
\indent $(\Leftarrow)$ Conversely, suppose $\Norm{u} = \Norm{v}$. Let $a,b\in\R$ be arbitrary, and notice
\begin{align}
\Norm{au + bv} &= \inp{au + bv}{au + bv} \nonumber \\
&= \inp{au}{au} + \inp{au}{bv} + \inp{bv}{au} + \inp{bv}{bv} \nonumber \\
&= a^2\Norm{u}^2 + ab\left(\inp{u}{v} + \inp{v}{u}\right) + b^2\Norm{v}^2. \label{eq1}
\end{align}
Also, we have
\begin{align}
\Norm{bu + av} &= \inp{bu + av}{bu + av} \nonumber \\
&= \inp{bu}{bu} + \inp{bu}{av} + \inp{av}{bu} + \inp{av}{av} \nonumber \\
&= b^2\Norm{u}^2 + ab\left(\inp{u}{v} + \inp{v}{u}\right) + a^2\Norm{v}^2. \label{eq2}
\end{align}
Since $\Norm{u} = \Norm{v}$, \eqref{eq1} equals \eqref{eq2}, and hence $\Norm{au + bv} = \Norm{bu + av}$. Since $a,b$ were arbitrary, the result follows.
\end{proof}
% Problem 9
\begin{problem}{9}
Suppose $u,v\in V$ and $\Norm{u}\leq 1$ and $\Norm{v}\leq 1$. Prove that
\begin{equation*}
\sqrt{1 - \Norm{u}^2}\sqrt{1 - \Norm{v}^2}\leq 1 - \Mod{\inp{u}{v}}.
\end{equation*}
\end{problem}
\begin{proof}
By the Cauchy-Schwarz Inequality, we have $\Mod{\inp{u}{v}}\leq \Norm{u}\Norm{v}$. Since $\Norm{u}\leq 1$ and $\Norm{v}\leq 1$, this implies
\begin{align*}
0 \leq 1 -\Norm{u}\Norm{v} \leq 1 -\Mod{\inp{u}{v}},
\end{align*}
and hence it's enough to show
\begin{equation*}
\sqrt{1 - \Norm{u}^2}\sqrt{1 - \Norm{v}^2} \leq 1 -\Norm{u}\Norm{v}.
\end{equation*}
Squaring both sides, it suffices to prove
\begin{equation}
\left(1 - \Norm{u}^2\right)\left(1 - \Norm{v}^2\right) \leq \left(1 -\Norm{u}\Norm{v}\right)^2. \label{eq3}
\end{equation}
Notice
\begin{align*}
\left(1 -\Norm{u}\Norm{v}\right)^2 - \left(1 - \Norm{u}^2\right)\left(1 - \Norm{v}^2\right) &= \Norm{u}^2 - 2\Norm{u}\Norm{v} + \Norm{v}^2\\
&= \left(\Norm{u} - \Norm{v}\right)^2\\
&\geq 0,
\end{align*}
and hence inequality \eqref{eq3} holds, completing the proof.
\end{proof}
% Problem 11
\begin{problem}{11}
Prove that
\begin{equation*}
16 \leq (a + b + c + d)\left(\frac{1}{a} + \frac{1}{b} + \frac{1}{c} + \frac{1}{d}\right)
\end{equation*}
for all positive numbers $a,b,c,d$.
\end{problem}
\begin{proof}
Define
\begin{align*}
x = \left(\sqrt{a}, \sqrt{b}, \sqrt{c}, \sqrt{d} \right)\quad\text{and}\quad y = \left(\sqrt{\frac{1}{a}}, \sqrt{\frac{1}{b}}, \sqrt{\frac{1}{c}}, \sqrt{\frac{1}{d}}\right).
\end{align*}
Then the Cauchy-Schwarz Inequality implies
\begin{align*}
(a + b + c + d)\left(\frac{1}{a} + \frac{1}{b} + \frac{1}{c} + \frac{1}{d}\right) &\geq \left(\sqrt{a}\sqrt{\frac{1}{a}} + \sqrt{b}\sqrt{\frac{1}{b}} + \sqrt{c}\sqrt{\frac{1}{c}} + \sqrt{d}\sqrt{\frac{1}{d}}\right)^2\\
&= (1 + 1 + 1 + 1)^2\\
&= 16,
\end{align*}
as desired.
\end{proof}
% Problem 13
\begin{problem}{13}
Suppose $u,v$ are nonzero vectors in $\R^2$. Prove that
\begin{equation*}
\inp{u}{v} = \Norm{u}\Norm{v}\cos\theta,
\end{equation*}
where $\theta$ is the angle between $u$ and $v$ (thinking of $u$ and $v$ as arrows with initial point at the origin).
\end{problem}
\begin{proof}
Let $A$ denote the line segment from the origin to $u$, let $B$ denote the line segment from the origin to $v$, and let $C$ denote the line segment from $v$ to $u$. Then $A$ has length $\Norm{u}$, $B$ has length $\Norm{v}$ and $C$ has length $\Norm{u - v}$. Letting $\theta$ denote the angle between $A$ and $B$, by the Law of Cosines we have
\begin{align*}
C^2 = A^2 + B^2 - 2BC\cos\theta,
\end{align*}
or equivalently
\begin{align*}
\Norm{u - v}^2 = \Norm{u}^2 + \Norm{v}^2 - 2\Norm{u}\Norm{v}\cos\theta.
\end{align*}
It follows
\begin{align*}
2\Norm{u}\Norm{v}\cos\theta &= \Norm{u}^2 + \Norm{v}^2 - \Norm{u-v}^2\\
&= \inp{u}{u} + \inp{v}{v} - \inp{u-v}{u-v}\\
&= \inp{u}{u} + \inp{v}{v} - \left(\inp{u}{u} - 2\inp{u}{v} + \inp{v}{v}\right)\\
&= 2\inp{u}{v}.
\end{align*}
Dividing both sides by $2$ gives the desired result.
\end{proof}
% Problem 15
\begin{problem}{15}
Prove that
\begin{equation*}
\left(\sum_{j = 1}^na_j b_j\right)^2 \leq \left(\sum_{j = 1}^nj{a_j}^2\right)\left(\sum_{j=1}^n\frac{{b_j}^2}{j}\right)
\end{equation*}
for all real numbers $a_1,\dots,a_n$ and $b_1,\dots,b_n$.
\end{problem}
\begin{proof}
Let
\begin{equation*}
u = \left(a_1, \sqrt{2}a_2, \dots, \sqrt{n}a_n\right)\quad\text{and}\quad v =\left(b_1, \frac{1}{\sqrt{2}}b_2,\dots, \frac{1}{\sqrt{n}}b_n\right).
\end{equation*}
Since $\inp{u}{v} = \sum_{k=1}^na_kb_k$, the Cauchy-Schwarz Inequality yields
\begin{align*}
\left(a_1b_1 +\dots + a_nb_n\right)^2 &\leq \Norm{u}^2\Norm{v}^2\\
&=\left({a_1}^2 + 2{a_2}^2 + \dots + n{a_n}^2\right)\left({b_1}^2 + \frac{{b_2}^2}{2} + \dots + \frac{{b_n}^2}{n}\right),
\end{align*}
as desired.
\end{proof}
% Problem 17
\begin{problem}{17}
Prove or disprove: there is an inner product on $\R^2$ such that the associated norm is given by
\begin{equation*}
\Norm{(x,y)} = \max\{\Mod{x},\Mod{y}\}
\end{equation*}
for all $(x, y)\in\R^2$.
\end{problem}
\begin{proof}
Suppose such an inner product existed. Then by the Parallelogram Equality, it follows
\begin{align*}
\Norm{(1, 0) + (0, 1)}^2 + \Norm{(1, 0) - (0, 1)}^2 = 2\left(\Norm{(1,0)}^2 + \Norm{(0,1)}^2\right).
\end{align*}
After simplification, this implies $2 = 4$, a contradiction. Hence no such inner product exists.
\end{proof}
% Problem 19
\begin{problem}{19}
Suppose $V$ is a real inner product space. Prove that
\begin{equation*}
\inp{u}{v} = \frac{\Norm{u+v}^2 - \Norm{u - v}^2}{4}
\end{equation*}
for all $u,v\in V$.
\end{problem}
\begin{proof}
Suppose $V$ is a real inner product space and let $u,v\in V$. It follows
\begin{align*}
\frac{\Norm{u+v}^2 - \Norm{u - v}^2}{4} &= \frac{\left(\Norm{u}^2 + 2\inp{u}{v} +\Norm{v}^2\right) - \left(\Norm{u}^2 - 2\inp{u}{v} +\Norm{v}^2\right)}{4}\\
&= \frac{4\inp{u}{v}}{4}\\
&= \inp{u}{v},
\end{align*}
as desired.
\end{proof}
% Problem 20
\begin{problem}{20}
Suppose $V$ is a complex inner product space. Prove that
\begin{equation*}
\inp{u}{v} = \frac{\Norm{u+v}^2 - \Norm{u - v}^2 + \Norm{u + iv}^2i - \Norm{u -iv}^2i}{4}
\end{equation*}
for all $u,v\in V$.
\end{problem}
\begin{proof}
Notice we have
\begin{align*}
\Norm{u + v}^2 &= \inp{u + v}{u + v}\\
&=\Norm{u}^2 +\inp{u}{v} + \inp{v}{u} + \Norm{v}^2
\end{align*}
and
\begin{align*}
-\Norm{u - v}^2 &= -\inp{u - v}{u - v}\\
&=-\Norm{u}^2 + \inp{u}{v} + \inp{v}{u} - \Norm{v}^2.
\end{align*}
Also, we have
\begin{align*}
\Norm{u + iv}^2i &= i\left(\inp{u + iv}{u+iv}\right)\\
&= i\left(\Norm{u}^2 + \inp{u}{iv} + \inp{iv}{u} + \inp{iv}{iv}\right)\\
&= i\left(\Norm{u}^2 - i\inp{u}{v} + i\inp{v}{u} + \Norm{v}^2\right)\\
&= i\Norm{u}^2 + \inp{u}{v} - \inp{v}{u} + i\Norm{v}^2
\end{align*}
and
\begin{align*}
-\Norm{u - iv}^2i &= -i\left(\inp{u - iv}{u-iv}\right)\\
&= -i\left(\Norm{u}^2 - \inp{u}{iv} - \inp{iv}{u} + \inp{iv}{iv}\right)\\
&= -i\left(\Norm{u}^2 + i\inp{u}{v} - i\inp{v}{u} + \Norm{v}^2\right)\\
&= -i\Norm{u}^2 + \inp{u}{v} - \inp{v}{u} - i\Norm{v}^2.
\end{align*}
Thus it follows
\begin{align*}
\Norm{u+v}^2 - \Norm{u - v}^2 + \Norm{u + iv}^2i - \Norm{u -iv}^2i = 4\inp{u}{v}.
\end{align*}
Dividing both sides by $4$ yields the desired result.
\end{proof}
%% Problem 21
%\begin{problem}{21}
%A norm on a vector space $U$ is a function $\Norm{\cdot}: U\to [0,\infty)$ such that $\Norm{u} = 0$ if and only if $u=0$, $\Norm{\alpha u} = \Mod{\alpha}\Norm{u}$ for all $\alpha\in\F$ and all $u\in U$, and $\Norm{u + v}\leq\Norm{u} + \Norm{v}$ for all $u,v\in U$. Prove that a norm satisfying the parallelogram equality comes from an inner product (in other words, show that if $\Norm{\cdot}$ is a norm on $U$ satisfying the parallelogram equality, then there is an inner product $\inp{\cdot}{\cdot}$ on $U$ such that $\Norm{u} = \inp{u}{u}^{1/2}$ for all $u\in U$).
%\end{problem}
% Problem 23
\begin{problem}{23}
Suppose $V_1, \dots, V_m$ are inner product spaces. Show that the equation
\begin{equation*}
\inp{(u_1,\dots, u_m)}{(v_1,\dots, v_m)} = \inp{u_1}{v_1} + \dots + \inp{u_m}{v_m}
\end{equation*}
defines an inner product on $V_1\times \dots \times V_m$.
\end{problem}
\begin{proof}
We prove that this definition satisfies each property of an inner product in turn.\\
\textbf{Positivity: } Let $(v_1,\dots, v_m)\in V_1\times\dots V_m$. Since $\inp{v_k}{v_k}$ is an inner product on $V_k$ for $k = 1,\dots, m$, we have $\inp{v_k}{v_k}\geq 0$. Thus
\begin{equation*}
\inp{(v_1,\dots, v_m)}{(v_1,\dots, v_m)} = \inp{v_1}{v_1} + \dots + \inp{v_m}{v_m} \geq 0.
\end{equation*}
\textbf{Definiteness: } First suppose $\inp{(v_1,\dots, v_m)}{(v_1,\dots, v_m)} = 0$ for $(v_1,\dots, v_m)\in V_1\times\dots\times V_m$. Then
\begin{equation*}
\inp{v_1}{v_1} + \dots + \inp{v_m}{v_m} = 0.
\end{equation*}
By positivity of each inner product on $V_k$ (for $k = 1,\dots, m$), we must have $\inp{v_k}{v_k}\geq 0$. Thus the equation above holds only if $\inp{v_k}{v_k} = 0$ for each $k$, which is true iff $v_k = 0$ (by definiteness of the inner product on $V_k$). Hence $(v_1, \dots, v_m) = (0, \dots, 0)$. Conversely, suppose $(v_1,\dots, v_m) = (0, \dots, 0)$. Then
\begin{align*}
\inp{(v_1,\dots, v_m)}{(v_1,\dots, v_m)} &= \inp{v_1}{v_1} + \dots + \inp{v_m}{v_m}\\
&= \inp{0}{0} + \dots + \inp{0}{0}\\
&= 0 + \dots + 0\\
&= 0,
\end{align*}
where the third equality follows from definiteness of the inner product on each $V_k$, respectively.\\
\textbf{Additivity in first slot: } Let
\begin{equation*}
(u_1,\dots, u_m), (v_1,\dots, v_m), (w_1,\dots, w_m)\in V_1\times\dots\times V_m.
\end{equation*}
It follows
\begin{align*}
\langle (u_1, \dots, u_m) + &(v_1,\dots, v_m)), (w_1,\dots, w_m)\rangle \\
&= \inp{(u_1 + v_1, \dots, u_m + v_m)}{(w_1,\dots, w_m)}\\
&= \inp{u_1 + v_1}{w_1} + \dots + \inp{u_m + v_m}{w_m}\\
&= \inp{u_1}{w_1} + \inp{v_1}{w_1} + \dots + \inp{u_m}{w_m} + \inp{v_m}{w_m}\\
&= \inp{(u_1,\dots,u_m)}{(w_1,\dots, w_m)} + \inp{(v_1,\dots, v_m)}{(w_1,\dots,w_m)},
\end{align*}
where the third equality follows from additivity in the first slot of each inner product on $V_k$, respectively.\\
\textbf{Homogeneity in the first slot: } Let $\lambda\in\F$ and
\begin{equation*}
(u_1,\dots, u_m),(v_1,\dots, v_m)\in V_1\times \dots \times V_m.
\end{equation*}
It follows
\begin{align*}
\inp{\lambda(u_1,\dots, u_m)}{(v_1,\dots, v_m)} &= \inp{(\lambda u_1,\dots, \lambda u_m)}{(v_1,\dots, v_m)} \\
&= \inp{\lambda u_1}{v_1} + \dots + \inp{\lambda u_m}{v_m}\\
&= \lambda\inp{u_1}{v_1} + \dots + \lambda\inp{u_m}{v_m}\\
&= \lambda(\inp{u_1}{v_1} + \dots + \inp{u_m}{v_m})\\
&=\lambda\inp{(u_1,\dots, u_m)}{(v_1,\dots, v_m)},
\end{align*}
where the third equality follows from homogeneity in the first slot of each inner product on $V_k$, respectively.\\
\textbf{Conjugate symmetry: } Again let
\begin{equation*}
(u_1,\dots, u_m),(v_1,\dots, v_m)\in V_1\times \dots \times V_m.
\end{equation*}
It follows
\begin{align*}
\inp{(u_1,\dots, u_m)}{(v_1,\dots, v_m)} &= \inp{u_1}{v_1} + \dots + \inp{u_m}{v_m}\\
&= \widebar{\inp{v_1}{u_1}} + \dots + \widebar{\inp{v_m}{u_m}}\\
&= \widebar{\inp{u_1}{v_1} + \dots + \inp{u_m}{v_m}}\\
&= \widebar{\inp{(v_1,\dots, v_m)}{(u_1,\dots, u_m)}},
\end{align*}
where the second equality follows from conjugate symmetry of each inner product on $V_k$, respectively.
\end{proof}
% Problem 24
\begin{problem}{24}
Suppose $S\in\Hom(V)$ is an injective operator on $V$. Define $\inp{\cdot}{\cdot}_1$ by
\begin{equation*}
\inp{u}{v}_1 = \inp{Su}{Sv}
\end{equation*}
for $u,v\in V$. Show that $\inp{\cdot}{\cdot}_1$ is an inner product on $V$.
\end{problem}
\begin{proof}
We prove that this definition satisfies each property of an inner product in turn.\\
\textbf{Positivity: } Let $v\in V$. Then $\inp{v}{v}_1 = \inp{Sv}{Sv} \geq 0$.\\
\textbf{Definiteness: } Suppose $\inp{v}{v} = 0$ for some $v\in V$. This is true iff $\inp{Sv}{Sv} = 0$ (by definition) which is true iff $Sv = 0$ (by definiteness of $\inp{\cdot}{\cdot}$), which is true iff $v = 0$ (since $S$ is injective).\\
\textbf{Additivity in first slot: } Let $u, v, w\in V$. Then
\begin{align*}
\inp{u + v}{w}_1 &= \inp{S(u + v)}{Sw}\\
&= \inp{Su + Sv}{Sw}\\
&= \inp{Su}{Sw} + \inp{Sv}{Sw}\\
&= \inp{u}{w}_1 + \inp{v}{w}_1.
\end{align*}
\textbf{Homogeneity in first slot: } Let $\lambda\in\F$ and $u, v\in V$. Then
\begin{align*}
\inp{\lambda u}{v}_1 &= \inp{S(\lambda u)}{Sv}\\
&= \inp{\lambda Su}{Sv}\\
&= \lambda \inp{Su}{Sv}\\
&= \lambda\inp{u}{v}_1.
\end{align*}
\textbf{Conjugate symmetry} Let $u,v\in V$. Then
\begin{align*}
\inp{u}{v}_1 &= \inp{Su}{Sv}\\
&= \widebar{\inp{Sv}{Su}}\\
&= \widebar{\inp{v}{u}_1}.
\end{align*}
\end{proof}
% Problem 25
\begin{problem}{25}
Suppose $S\in\Hom(V)$ is not injective. Define $\inp{\cdot}{\cdot}_1$ as in the exercise above. Explain why $\inp{\cdot}{\cdot}_1$ is not an inner product on $V$.
\end{problem}
\begin{proof}
If $S$ is not injective, then $\inp{\cdot}{\cdot}_1$ fails the definiteness requirement in the definition of an inner product. In particular, there exists $v\neq 0$ such that $Sv =0$. Hence $\inp{v}{v}_1 = \inp{Sv}{Sv} = 0$ for a nonzero $v$.
\end{proof}
% Problem 27
\begin{problem}{27}
Suppose $u,v,w\in V$. Prove that
\begin{equation*}
\norm{w - \frac{1}{2}(u + v)}^2 = \frac{\norm{w - u}^2 + \norm{w - v}^2}{2} - \frac{\norm{u - v}^2}{4}.
\end{equation*}
\end{problem}
\begin{proof}
We have
\begin{align*}
\norm{w - \frac{1}{2}(u + v)}^2 &= \norm{\left(\frac{w - u}{2}\right) + \left(\frac{w - v}{2} \right)}^2\\
&= 2\norm{\frac{w - u}{2}}^2 + 2 \norm{\frac{w - v}{2}}^2 - \norm{\left(\frac{w-u}{2}\right) - \left(\frac{w-v}{2}\right)}^2\\
&= \frac{\norm{w-u}^2 + \norm{w - v}^2}{2} - \norm{\frac{-u + v}{2}}^2\\
&= \frac{\norm{w-u}^2 + \norm{w - v}^2}{2} - \frac{\norm{u - v}^2}{4},
\end{align*}
where the second equality follows by the Parallelogram Equality.
\end{proof}
The next problem requires some extra work to prove. We first include a definition and prove a theorem.
\begin{definition}
Suppose $\norm{\cdot}_1$ and $\norm{\cdot}_2$ are norms on vector space $V$. We say $\norm{\cdot}_1$ and $\norm{\cdot}_2$ are \emph{equivalent} if there exist $0 < C_1\leq C_2$ such that
\begin{equation*}
C_1 \norm{v}_1 \leq \norm{v}_2 \leq C_2 \norm{v}_1
\end{equation*}
for all $v\in V$.
\end{definition}
\begin{thm-non}
Any two norms on a finite-dimensional vector space are equivalent.
\end{thm-non}
\begin{proof}
Let $V$ be finite-dimensional with basis $e_1,\dots, e_n$. It suffices to prove that every norm on $V$ is equivalent to the $\ell_1$-style norm $\norm{\cdot}_1$ defined by
\begin{equation*}
\norm{v}_1 = \abs{\alpha_1} + \dots + \abs{\alpha_n}
\end{equation*}
for all $v = \alpha_1e_1 + \dots + \alpha_ne_n\in V$.\\
\indent Let $\norm{\cdot}$ be a norm on $V$. We wish to show $C_1\norm{v}_1 \leq \norm{v}\leq C_2\norm{v}_1$ for all $v\in V$ and some choice of $C_1,C_2$. Since this is trivially true for $v = 0$, we need only consider $v\neq 0$, in which case we have
\begin{align*}
C_1 \leq \norm{u} \leq C_2, \tag{*}\label{key}
\end{align*}
where $u = v/\norm{v}_1$. Thus it suffices to consider only vectors $v\in V$ such that $\norm{v}_1 = 1$.\\
\indent We will now show that $\norm{\cdot}$ is continuous under $\norm{\cdot}_1$ and apply the Extreme Value Theorem to deduce the desired result. So let $\epsilon > 0$ and define $M = \max\{\norm{e_1}, \dots, \norm{e_n}\}$ and
\begin{equation*}
\delta = \frac{\epsilon}{M}.
\end{equation*}
It follows that if $u,v\in V$ are such that $\norm{u - v}_1 < \delta$, then
\begin{align*}
\abs{\norm{u} - \norm{v}} &\leq \norm{u - v}\\
&\leq M\norm{u - v}_1\\
&\leq M\delta \\
&= \epsilon,
\end{align*}
and $\norm{\cdot}$ is indeed continuous under the topology induced by $\norm{\cdot}_1$. Let $\mathcal{S} = \{u\in V\mid \norm{u}_1=1\}$ (the unit sphere with respect to $\norm{\cdot}_1$). Since $\mathcal{S}$ is compact and $\norm{\cdot}$ is continuous on it, by the Extreme Value Theorem we may define
\begin{align*}
C_1 = \min_{u \in \mathcal{S}}\norm{u}~~~\text{and}~~~C_2 = \max_{u \in \mathcal{S}}\norm{u}.
\end{align*}
But now $C_1$ and $C_2$ satisfy \eqref{key}, completing the proof.
\end{proof}
% Problem 29
\begin{problem}{29}
For $u,v \in V$, define $d(u,v) = \norm{u - v}$.
\begin{enumerate}[(a)]
\item Show that $d$ is a metric on $V$.
\item Show that if $V$ is finite-dimensional, then $d$ is a complete metric on $V$ (meaning that every Cauchy sequence converges).
\item Show that every finite-dimensional subspace of $V$ is a closed subset of $V$ (with respect to the metric $d$).
\end{enumerate}
\end{problem}
\begin{proof}
\begin{enumerate}[(a)]
\item We show that $d$ satisfies each property of the definition of a metric in turn.\\
\textbf{Identity of indiscernibles: } Let $u, v\in V$. It follows
\begin{align*}
d(u, v) = 0 &\iff \sqrt{\inp{u-v}{u-v}} = 0\\
&\iff \inp{u - v, u- v} = 0\\
&\iff u - v = 0\\
&\iff u = v.
\end{align*}
\textbf{Symmetry: } Let $u, v\in V$. We have
\begin{align*}
d(u, v) &= \norm{u - v}\\
&= \norm{(-1)(u - v)}\\
&= \norm{v - u}\\
&= d(v, u).
\end{align*}
\textbf{Triangle inequality: } Let $u, v, w\in V$. Notice
\begin{align*}
d(u,v) + d(v, w) &= \norm{u - v} + \norm{v - w}\\
&\leq \norm{(u - v) + (v - w)}\\
&= \norm{u, w}\\
&= d(u,w).
\end{align*}
\item Suppose $V$ is a $p$-dimensional vector space with basis $e_1,\dots, e_p$. Assume $\{v_k\}_{k = 1}^\infty$ is Cauchy. Then for $\epsilon > 0$, there exists $N\in\Z^+$ such that $\norm{v_m - v_n} < \epsilon$ whenever $m,n > N$. Given any $v_i$ in our Cauchy sequence, we adopt the notation that $\alpha_{i, 1}, \dots, \alpha_{i, p}\in \F$ are always defined such that
\begin{equation*}
v_i = \alpha_{i, 1}e_1 + \dots + \alpha_{i, p}e_p.
\end{equation*}
By our previous theorem, $\norm{\cdot}$ is equivalent to $\norm{\cdot}_1$ (where $\norm{\cdot}_1$ is defined in that theorem's proof). Thus there exists some $c > 0$ such that, whenever $m,n > N$, we have
\begin{align*}
c\norm{v_m - v_n}_1 \leq \norm{v_m - v_n} < \epsilon,
\end{align*}
and hence
\begin{equation*}
c\left(\sum_{i = 1}^p\abs{\alpha_{m, i} - \alpha_{n, i}}\right) < \epsilon.
\end{equation*}
This implies that $\{\alpha_{k, i}\}_{k = 1}^\infty$ is Cauchy in $\R$ for each $i = 1,\dots, p$. Since $\R$ is complete, these sequences converge. So let $\alpha_i = \lim_{k \to \infty}\alpha_{k, i}$ for each $i$, and define $v = \alpha_1 e_1 + \dots + \alpha_p e_p$. It follows
\begin{align*}
\norm{v_j - v} &= \norm{(\alpha_{j, 1} - \beta_1)e_1 + \dots + (\alpha_{j, p}-\beta_p)e_p}\\
&\leq \abs{\alpha_{j, 1}-\alpha_1}\norm{e_1} + \dots + \abs{\alpha_{j, p}-\alpha_p}\norm{e_p}.
\end{align*}
Since $\alpha_{j, i}\to \alpha_i$ for $i = 1,\dots, p$, the RHS can be made arbitrarily small by choosing sufficiently large $M\in\Z^+$ and considering $j > M$. Thus $\{v_k\}_{k = 1}^\infty$ converges to $v$, and $V$ is indeed complete with respect to $\norm{\cdot}$.
\item Suppose $U$ is a finite-dimensional subspace of $V$, and suppose $\{u_k\}_{k=1}^\infty\subseteq U$ is Cauchy. By (b), $\lim_{k \to \infty}u_k \in U$, hence $U$ contains all its limit points. Thus $U$ is closed. \qedhere
\end{enumerate}
\end{proof}
% Problem 31
\begin{problem}{31}
Use inner products to prove Apollonius's Identity: In a triangle with sides of length $a$, $b$, and $c$, let $d$ be the length of the line segment from the midpoint of the side of length $c$ to the opposite vertex. Then
\begin{equation*}
a^2 + b^2 = \frac{1}{2}c^2 + 2d^2.
\end{equation*}
\end{problem}
\begin{proof}
Consider a triangle formed by vectors $v, w\in\R^2$ and the origin such that $\norm{w} = a$, $\norm{v} = c$, and $\norm{w - v} = b$. The identity follows by applying Problem 27 with $u = 0$.
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SECTION B
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Orthonormal Bases}
% Problem 1
\begin{problem}{1}
\begin{enumerate}[(a)]
\item Suppose $\theta\in \R$. Show that $(\cos\theta, \sin\theta)$, $(-\sin\theta,\cos\theta)$ and $(\cos\theta, \sin\theta)$, $(\sin\theta,-\cos\theta)$ are orthonormal bases of $\R^2$.
\item Show that each orthonormal basis of $\R^2$ is of the form given by one of the two possibilities of part (a).
\end{enumerate}
\end{problem}
\begin{proof}
\begin{enumerate}[(a)]
\item Notice
\begin{equation*}
\inp{(\cos\theta, \sin\theta)}{(-\sin\theta,\cos\theta)} = -\sin\theta\cos\theta + \sin\theta\cos\theta = 0
\end{equation*}
and
\begin{equation*}
\inp{(\cos\theta, \sin\theta)}{(\sin\theta,-\cos\theta)} = \sin\theta\cos\theta - \sin\theta\cos\theta = 0,
\end{equation*}
hence both lists are orthonormal. Clearly the three distinct vectors contained in the two lists all have norm $1$ (following from the identity $\cos^2\theta + \sin^2\theta = 1$). Since both lists have length $2$, by Theorem 6.28 both lists are orthonormal bases.
\item Suppose $e_1,e_2$ is an orthonormal basis of $\R^2$. Since $\norm{e_1} = \norm{e_2} = 1$, there exist $\theta, \varphi \in [0, 2\pi)$ such that
\begin{equation*}
e_1 = (\cos\theta, \sin\theta) \quad\text{and}\quad e_2 = (\cos\varphi, \sin\varphi).
\end{equation*}
Next, since $\inp{e_1}{e_2} = 0$, we have
\begin{equation*}
\cos\theta\cos\varphi + \sin\theta\sin\varphi = 0.
\end{equation*}
Since $\cos\theta\cos\varphi = \frac{1}{2}(\cos(\theta + \varphi) + \cos(\theta - \varphi))$ and $\sin\theta\sin\varphi = \cos(\theta - \varphi) - \cos(\theta + \varphi)$, the above implies
\begin{equation*}
\cos(\theta - \varphi) = 0
\end{equation*}
and thus $\varphi = \theta + \frac{3\pi}{2} - n\pi$, for $n \in \Z$. Since $\theta,\varphi\in[0, 2\pi)$, this implies $\varphi = \theta \pm \frac{\pi}{2}$. If $\varphi = \theta + \frac{\pi}{2}$, then
\begin{align*}
e_2 &= \left(\cos\left(\theta + \frac{\pi}{2}\right), \sin\left(\theta + \frac{\pi}{2}\right)\right)\\
&= (-\sin\theta, \cos\theta),
\end{align*}
and if $\varphi = \theta - \frac{\pi}{2}$, then
\begin{align*}
e_2 &= \left(\cos\left(\theta - \frac{\pi}{2}\right), \sin\left(\theta - \frac{\pi}{2}\right)\right)\\
&= (\sin\theta, -\cos\theta).
\end{align*}
Thus all orthonormal bases of $\R^2$ have one of the two forms from (a).
\end{enumerate}
\end{proof}
% Problem 3
\begin{problem}{3}
Suppose $T\in\Hom(\R^3)$ has an upper-triangular matrix with respect to the basis $(1, 0, 0)$, $(1, 1, 1)$, $(1, 1, 2)$. Find an orthonormal basis of $\R^3$ (use the usual inner product on $\R^3$) with respect to which $T$ has an upper-triangular matrix.
\end{problem}
\begin{proof}
Let $v_1 = (1, 0, 0), v_2 = (1, 1, 1)$, and $v_3 = (1, 1, 2)$. By the proof of 6.37, $T$ has an upper-triangular matrix with respect to the the basis $e_1, e_2, e_3$ generated by applying the Gram-Schmidt Procedure to $v_1, v_2, v_3$. Since $\norm{v_1} = 1$, $e_1 = v_1$. Next, we have
\begin{align*}
e_2 &= \frac{v_2 - \inp{v_2}{e_1} e_1}{\norm{ v_2 - \inp{v_2}{e_1} e_1}}\\
&=\frac{(1, 1, 1) - \inp{(1, 1, 1)}{(1, 0, 0)}(1, 0, 0)}{\norm{(1, 1, 1) - \inp{(1, 1, 1)}{(1, 0, 0)}(1, 0, 0)}}\\
&= \frac{(1, 1, 1) - (1, 0, 0)}{\norm{(1, 1, 1) - (1, 0, 0)}}\\
&= \frac{(0, 1, 1)}{\norm{(0, 1, 1)}}\\
&= \left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)
\end{align*}
and
\begin{align*}
e_3 &= \frac{v_3 - \inp{v_3}{e_1} e_1 - \inp{v_3}{e_2}e_2}{\norm{v_3 - \inp{v_3}{e_1} e_1 - \inp{v_3}{e_2}e_2}}\\
&= \frac{(1, 1, 2) - \inp{(1, 1, 2)}{(1, 0, 0)}(1, 0, 0) - \left\langle(1, 1, 2), \left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)\right\rangle \left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)}{\norm{(1, 1, 2) - \inp{(1, 1, 2)}{(1, 0, 0)}(1, 0, 0) - \left\langle(1, 1, 2), \left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)\right\rangle \left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)}}\\
&= \frac{(1, 1, 2) - (1, 0, 0) - \frac{3}{\sqrt{2}}\left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)}{\norm{(1, 1, 2) - (1, 0, 0) - \frac{3}{\sqrt{2}}\left(0, \frac{1}{\sqrt{2}}, \frac{1}{\sqrt{2}}\right)}}\\
&= \frac{\left(0, -\frac{1}{2}, \frac{1}{2}\right)}{\norm{\left(0, -\frac{1}{2}, \frac{1}{2}\right)}}\\
&= \left(0, -\frac{\sqrt{2}}{2}, \frac{\sqrt{2}}{2}\right),
\end{align*}
and we're done.
\end{proof}
% Problem 4
\begin{problem}{4}
Suppose $n$ is a positive integer. Prove that
\begin{align*}
\frac{1}{\sqrt{2\pi}}, \frac{\cos x}{\sqrt{\pi}}, \frac{\cos 2x}{\sqrt{\pi}}, \dots, \frac{\cos nx}{\sqrt{\pi}}, \frac{\sin x}{\sqrt{\pi}}, \frac{\sin 2x}{\sqrt{\pi}}, \dots, \frac{\sin nx}{\sqrt{\pi}}
\end{align*}
is an orthonormal list of vectors in $\mathcal{C}[-\pi, \pi]$, the vector space of continuous real-valued functions on $[-\pi, \pi]$ with inner product
\begin{align*}
\inp{f}{g} = \int_{-\pi}^\pi f(x)g(x)dx.
\end{align*}
\end{problem}
\begin{proof}
First we show that all vectors in the list have norm $1$. Notice
\begin{align*}
\norm{\frac{1}{\sqrt{2\pi}}} &= \sqrt{\int_{-\pi}^\pi\frac{1}{2\pi}dx}\\
&=\sqrt{ \frac{1}{2\pi}\int_{-\pi}^\pi dx}\\
&= 1.
\end{align*}
And for $k\in\Z^+$, we have
\begin{align*}
\norm{\frac{\cos(kx)}{\sqrt{\pi}}} &= \sqrt{\frac{1}{\pi}\int_{-\pi}^\pi \cos(kx)^2dx}\\
&= \sqrt{\frac{1}{\pi}\left[\frac{\sin(2kx)}{4k} + \frac{x}{2}\right]_{-\pi}^\pi}\\
&= \sqrt{\frac{1}{\pi}\left[\frac{\pi}{2} - \left(-\frac{\pi}{2}\right)\right]}\\
&= 1,
\end{align*}
and
\begin{align*}
\norm{\frac{\sin(kx)}{\sqrt{\pi}}} &= \sqrt{\frac{1}{\pi}\int_{-\pi}^\pi \sin(kx)^2dx}\\
&= \sqrt{\frac{1}{\pi}\left[\frac{x}{2} - \frac{\cos(2kx)}{4k}\right]_{-\pi}^\pi}\\
&= \sqrt{\frac{1}{\pi}\left[\frac{\pi}{2} - \left(-\frac{\pi}{2}\right)\right]}\\
&= 1,
\end{align*}
so indeed all vectors have norm $1$. Now we show them to be pairwise orthogonal. Suppose $j,k\in\Z$ are such that $j \neq k$. It follows from basic calculus
\begin{align*}
\left\langle \frac{\sin(jx)}{\sqrt{\pi}}, \frac{\sin(kx)}{\sqrt{\pi}}\right\rangle &= \frac{1}{\pi}\int_{-\pi}^\pi\sin(jx)\sin(kx)dx\\
&= \frac{1}{\pi}\left[\frac{k\sin(jx)\cos(kx) + j\cos(jx)\sin(kx)}{j^2 - k^2}\right]_{-\pi}^\pi\\
&= 0,
\end{align*}
\begin{align*}
\left\langle \frac{\sin(jx)}{\sqrt{\pi}}, \frac{\cos(kx)}{\sqrt{\pi}}\right\rangle &= \frac{1}{\pi}\int_{-\pi}^\pi\sin(jx)\cos(kx)dx\\
&= -\frac{1}{\pi}\left[\frac{k\sin(jx)\sin(kx) + j\cos(jx)\cos(kx)}{j^2 - k^2}\right]_{-\pi}^\pi\\
&= -\frac{1}{\pi}\left[\left(\frac{j\cos(j\pi)\cos(k\pi)}{j^2- k^2} \right) - \left(\frac{j\cos(-j\pi)\cos(-k\pi)}{j^2- k^2}\right) \right]\\
&= 0,
\end{align*}
\begin{align*}
\left\langle \frac{\cos(jx)}{\sqrt{\pi}}, \frac{\cos(kx)}{\sqrt{\pi}}\right\rangle &= \frac{1}{\pi}\int_{-\pi}^\pi\cos(jx)\cos(kx)dx\\
&= \frac{1}{\pi}\left[\frac{j\sin(j x)\cos(k x) - k\cos(j x)\sin(k x)}{j^2 - k^2}\right]_{-\pi}^\pi\\
&= 0,
\end{align*}
\begin{align*}
\left\langle \frac{\sin(jx)}{\sqrt{\pi}}, \frac{\cos(jx)}{\sqrt{\pi}}\right\rangle &= \frac{1}{\pi}\int_{-\pi}^\pi\sin(jx)\cos(jx)dx\\
&= \left[-\frac{\cos^2(jx)}{2j}\right]_{-\pi}^\pi\\
&= 0,
\end{align*}
\begin{align*}
\left\langle \frac{1}{\sqrt{2\pi}}, \frac{\cos(jx)}{\sqrt{\pi}}\right\rangle &= \frac{1}{\sqrt{2}\pi}\int_{-\pi}^\pi\cos(jx)dx\\
&= \frac{1}{\sqrt{2}\pi}\left[\frac{\sin(jx)}{j}\right]_{-\pi}^\pi\\
&= 0,
\end{align*}
and
\begin{align*}
\left\langle \frac{1}{\sqrt{2\pi}}, \frac{\sin(jx)}{\sqrt{\pi}}\right\rangle &= \frac{1}{\sqrt{2}\pi}\int_{-\pi}^\pi\sin(jx)dx\\
&= \frac{1}{\sqrt{2}\pi}\left[-\frac{\cos(jx)}{j}\right]_{-\pi}^\pi\\
&= \frac{1}{\sqrt{2}\pi}\left[-\frac{\cos(j\pi) - \cos(-j\pi)}{j}\right]\\
&= 0.
\end{align*}
Thus the list is indeed an orthonormal list in $\mathcal{C}[-\pi, \pi]$.
\end{proof}
% Problem 5
\begin{problem}{5}
On $\poly_2(\R)$, consider the inner product given by
\begin{equation*}
\inp{p}{q} = \int_0^1p(x)q(x)\,dx.
\end{equation*}
Apply the Gram-Schmidt Procedure to the basis $1, x, x^2$ to produce an orthonormal basis of $\poly_2(\R)$.
\end{problem}
\begin{proof}
First notice $\norm{1} = 1$, hence $e_1 = 1$. Next notice
\begin{align*}
v_2 - \inp{v_1}{e_1}e_1 &= x - \inp{x}{1}\\
&= x - \int_0^1x\,dx\\
&= x - \frac{1}{2}
\end{align*}
and
\begin{align*}
\norm{x - \frac{1}{2}} &= \sqrt{\left\langle x - \frac{1}{2}, x - \frac{1}{2}\right\rangle}\\
&= \sqrt{\int_0^1\left(x - \frac{1}{2}\right)\left(x - \frac{1}{2}\right)\,dx}\\
&= \sqrt{\int_0^1\left(x^2 - x + \frac{1}{4}\right)\,dx}\\
&= \sqrt{\frac{1}{3} - \frac{1}{2} + \frac{1}{4}}\\
&= \frac{1}{2\sqrt{3}},
\end{align*}
and therefore we have
\begin{equation*}
e_2 = 2\sqrt{3}\left(x - \frac{1}{2}\right).
\end{equation*}
To compute $e_3$, first notice
\begin{align*}
v_3 - \inp{v_3}{e_1}e_1 - \inp{v_3}{e_2}e_2 &= x^2 - \int_0^1x^2\,dx - \left[2\sqrt{3}\int_0^1x^2\left(x - \frac{1}{2}\right)\,dx\right]e_2\\
&= x^2 - \frac{1}{3} - \left[2\sqrt{3}\int_0^1\left(x^3 - \frac{x^2}{2}\right)\,dx\right]\left[2\sqrt{3}\left(x - \frac{1}{2}\right)\right]\\
&= x^2 - \frac{1}{3} - 12\left(\frac{1}{4} - \frac{1}{6}\right)\left(x - \frac{1}{2}\right)\\
&= x^2 - \frac{1}{3} - \left(x - \frac{1}{2}\right)\\
&= x^2 - x + \frac{1}{6}
\end{align*}
and
\begin{align*}
\norm{ x^2 - x + \frac{1}{6}} &= \sqrt{\left\langle x^2 - x + \frac{1}{6}, x^2 - x + \frac{1}{6}\right\rangle}\\
&= \sqrt{\int_0^1\left( x^2 - x + \frac{1}{6}\right)\left( x^2 - x + \frac{1}{6}\right)\,dx}\\
&= \sqrt{\int_0^1\left(x^4 - 2x^3 + \frac{4}{3}x^2 - \frac{x}{3} + \frac{1}{36}\right)\,dx}\\
&= \sqrt{\frac{1}{5} - \frac{1}{2} + \frac{4}{9} - \frac{1}{6} + \frac{1}{36}}\\
&= \frac{1}{\sqrt{180}}\\
&= \frac{1}{6\sqrt{5}}.
\end{align*}
Thus
\begin{equation*}
e_3 = 6\sqrt{5}\left(x^2 - x + \frac{1}{6}\right),
\end{equation*}
and we're done.
\end{proof}
% Problem 7
\begin{problem}{7}
Find a polynomial $q\in\poly_2(\R)$ such that
\begin{equation*}
p\left(\frac{1}{2}\right)=\int_0^1p(x)q(x)\,dx
\end{equation*}
for every $p\in\poly_2(\R)$.
\end{problem}
\begin{proof}
Consider the inner product $\inp{p}{q} = \int_0^1p(x)q(x)\,dx$ on $\poly_2(\R)$. Define $\varphi\in\Hom(\poly_2(\R))$ by $\varphi(p) = p\left(\frac{1}{2}\right)$ and let $e_1, e_2, e_3$ be the orthonormal basis found in Problem 5. By the Riesz Representation Theorem, there exists $q\in\poly_2(\R)$ such that $\varphi(p) = \inp{p}{q}$ for all $p\in\poly_2(\R)$. That is, such that
\begin{equation*}
p\left(\frac{1}{2}\right) = \int_0^1p(x)q(x)\,dx.
\end{equation*}
Equation 6.43 in the proof of the Riesz Representation Theorem fashions a way to find $q$. In particular, we have
\begin{align*}
q(x) &= \widebar{\varphi(e_1)}e_1 + \widebar{\varphi(e_2)}e_2 + \widebar{\varphi(e_3)}e_3\\
&= e_1 + 2\sqrt{3}\left(\frac{1}{2} - \frac{1}{2}\right)e_2 + 6\sqrt{5}\left(\frac{1}{4} - \frac{1}{2} + \frac{1}{6}\right)e_3\\
&= 1 + 6\sqrt{5}\left(\frac{-1}{12}\right)\left[6\sqrt{5}\left(x^2 - x + \frac{1}{6}\right)\right]\\
&= -15(x^2 - x) - \frac{3}{2},
\end{align*}
as desired.
\end{proof}
% Problem 9
\begin{problem}{9}
What happens if the Gram-Schmidt Procedure is applied to a list of vectors that is not linearly independent?
\end{problem}
\begin{proof}
Suppose $v_1,\dots,v_m$ are linearly dependent. Let $j$ be the smallest integer in $\{1,\dots, m\}$ such that $v_j \in \Span(v_1,\dots, v_{j - 1})$. Then $v_1,\dots, v_{j - 1}$ are linearly independent. The first $j - 1$ steps of the Gram-Schmidt Procedure will produce an orthonormal list $e_1,\dots, e_{j - 1}$. At step $j$, however, notice
\begin{align*}
v_j - \inp{v_j}{e_1}e_1 - \dots - \inp{v_j}{e_{j - 1}}e_{j-1} = v_j - v_j = 0,
\end{align*}
and we are left trying to assign $e_j$ to $\frac{0}{0}$, which is undefined. Thus the procedure cannot be applied to a linearly dependent list.
\end{proof}
% Problem 11
\begin{problem}{11}
Suppose $\inp{\cdot}{\cdot}_1$ and $\inp{\cdot}{\cdot}_2$ are inner products on $V$ such that $\inp{v}{w}_1 = 0$ if and only if $\inp{v}{w}_2 = 0$. Prove that there is a positive number $c$ such that $\inp{v}{w}_1 = c\inp{v}{w}_2$ for every $v,w\in V$.
\end{problem}
\begin{proof}
Let $v, w\in V$ be arbitrary. By hypothesis, if $v$ and $w$ are orthogonal relative to one of the inner products, they're orthogonal relative to the other. Hence any choice of $c\in \R$ would satisfy $\inp{v}{w}_1 = c\inp{v}{w}_2$. So suppose $v$ and $w$ are not orthogonal relative to either inner product. Then both $v$ and $w$ must be nonzero (by Theorem 6.7, parts b and c, respectively). Thus $\inp{v}{v}_1$, $\inp{w}{w}_1$, $\inp{v}{v}_2$, and $\inp{w}{w}_2$ are all nonzero as well. It now follows
\begin{align*}
0 &= \inp{v}{w}_1 - \frac{\inp{v}{w}_1}{\inp{v}{v}_1}\inp{v}{v}_1\\
&= \inp{v}{w}_1 - \left\langle v, \widebar{\left(\frac{\inp{v}{w}_1}{\inp{v}{v}_1}\right)}v \right\rangle_1\\
&= \left\langle v, w - \widebar{\left(\frac{\inp{v}{w}_1}{\inp{v}{v}_1}\right)}v \right\rangle_1\\
&= \left\langle v, w - \widebar{\left(\frac{\inp{v}{w}_1}{\inp{v}{v}_1}\right)}v \right\rangle_2\\
&= \inp{v}{w}_2 - \left\langle v, \widebar{\left(\frac{\inp{v}{w}_1}{\inp{v}{v}_1}\right)}v \right\rangle_2\\
&= \inp{v}{w}_2 - \frac{\inp{v}{w}_1}{\inp{v}{v}_1}\inp{v}{v}_2\\
&= \inp{v}{w}_2 - \frac{\inp{v}{v}_2}{\inp{v}{v}_1}\inp{v}{w}_1,
\end{align*}
where the fifth equality follows by our hypothesis. Thus
\begin{equation}
\inp{v}{w}_1 = \frac{{\norm{v}_1}^2}{{\norm{v}_2}^2}\inp{v}{w}_2. \label{eq4}
\end{equation}
By a similar computation, notice
\begin{align*}
0 &= \inp{v}{w}_1 - \frac{\inp{v}{w}_1}{\inp{w}{w}_1}\inp{w}{w}_1\\
&= \inp{v}{w}_1 - \left\langle \frac{\inp{v}{w}_1}{\inp{w}{w}_1}w, w \right\rangle_1\\
&= \left\langle v - \frac{\inp{v}{w}_1}{\inp{w}{w}_1}w, w \right\rangle_1\\
&= \left\langle v - \frac{\inp{v}{w}_1}{\inp{w}{w}_1}w, w \right\rangle_2\\
&= \inp{v}{w}_2 - \left\langle \frac{\inp{v}{w}_1}{\inp{w}{w}_2}w, w \right\rangle_2\\
&= \inp{v}{w}_2 - \frac{\inp{v}{w}_1}{\inp{w}{w}_2}\inp{w}{w}_2\\
&= \inp{v}{w}_2 - \frac{\inp{w}{w}_2}{\inp{w}{w}_1}\inp{v}{w}_1,
\end{align*}
and thus
\begin{equation}
\inp{v}{w}_1 = \frac{{\norm{w}_1}^2 }{{\norm{w}_2}^2}\inp{v}{w}_2 \label{eq5}
\end{equation}
as well. By combining Equations \eqref{eq4} and \eqref{eq5}, we conclude
\begin{equation*}
\frac{\inp{v}{v}_1}{\inp{v}{v}_2} = \frac{\inp{w}{w}_1}{\inp{w}{w}_2}.
\end{equation*}
Since $v$ and $w$ were arbitrary nonzero vectors in $V$, choosing $c = {\norm{u}_1}^2/{\norm{u}_2}^2$ for any $u\neq 0$ guarantees $\inp{v}{w}_1 = c\inp{v}{w}_2$ for every $v,w\in V$, as was to be shown.
\end{proof}
% Problem 13
\begin{problem}{13}
Suppose $v_1,\dots, v_m$ is a linearly independent list in $V$. Show that there exists $w\in V$ such that $\inp{w}{v_j} > 0$ for all $j \in \{1, \dots, m\}$.
\end{problem}
\begin{proof}
Let $W = \Span(v_1,\dots, v_m)$. Given $v\in W$, let $a_1,\dots, a_m\in\F$ be such that $v = a_1v_1 + \dots + a_mv_m$. Define $\varphi\in\Hom(W)$ by
\begin{align*}
\varphi(v) = a_1 + \dots + a_m.
\end{align*}
By the Riesz Representation Theorem, there exists $w\in W$ such that $\varphi(v) = \inp{v}{w}$ for all $v\in W$. But then $\varphi(v_j) = 1$ for $j \in\{1,\dots, m\}$, and indeed such a $w\in V$ exists.
\end{proof}
% Problem 15
\begin{problem}{15}
Suppose $C_\R([-1, 1])$ is the vector space of continuous real-valued functions on the interval $[-1, 1]$ with inner product given by
\begin{equation*}
\inp{f}{g} = \int_{-1}^1f(x)g(x)dx
\end{equation*}
for $f,g\in C_\R([-1, 1])$. Let $\varphi$ be the linear functional on $C_\R([-1, 1])$ defined by $\varphi(f) = f(0)$. Show that there does not exist $g\in C_\R([-1,1])$ such that
\begin{equation*}
\varphi(f) = \inp{f}{g}
\end{equation*}
for every $f\in C_\R([-1, 1])$.
\end{problem}
\begin{proof}
Suppose not. Then there exists $g\in C_\R([-1,1])$ such that
\begin{equation*}
\varphi(f) = \inp{f}{g}
\end{equation*}
for every $f\in C_\R([-1, 1])$. Choose $f(x) = x^2 g(x)$. Then $f(0) = 0$, and hence
\begin{equation*}
\int_{-1}^1 f(x) g(x)dx = \int_{-1}^1 [xg(x)]^2dx = 0.
\end{equation*}
Now, let $h(x) = xg(x)$. Since $h$ is continuous on $[-1, 1]$, there exists an interval $[a, b]\subseteq [-1, 1]$ such that $h(x)\neq 0$ for all $x \in [a, b]$. By the Extreme Value Theorem, $h(x)^2$ has a minimum at some $m\in[a,b]$. Thus $h(m)^2> 0$, and we now conclude
\begin{align*}
0 = \int_{-1}^1h(x)^2dx = \int_a^bh(x)^2dx \geq (b-a)h(m)^2 > 0,
\end{align*}
which is absurd. Thus it must be that no such $g$ exists.
\end{proof}
% Problem 17
\begin{problem}{17}
For $u\in V$, let $\Phi_u$ denote the linear functional on $V$ defined by
\begin{equation*}
(\Phi_u)(v) = \inp{v}{u}
\end{equation*}
for $v \in V$.
\begin{enumerate}[(a)]
\item Show that if $\F = \R$, then $\Phi$ is a linear map from $V$ to $V'$.
\item Show that if $\F = \C$ and $V\neq \{0\}$, then $\Phi$ is not a linear map.
\item Show that $\Phi$ is injective.
\item Suppose $\F = \R$ and $V$ is finite-dimensional. Use parts (a) and (c) and a dimension-counting argument (but without using 6.42) to show that $\Phi$ is an isomorphism from $V$ to $V'$.
\end{enumerate}
\end{problem}
\begin{proof}
\begin{enumerate}[(a)]
\item Suppose $\F=\R$. Let $u, u'\in V$ and $\alpha\in\R$. Then, for all $v\in V$, we have
\begin{equation*}
\Phi_{u + u'}(v) = \inp{v}{u + u'} = \inp{v}{u} + \inp{v}{u'} = \Phi_u(v) + \Phi_{u'}(v)
\end{equation*}
and
\begin{equation*}
\Phi_{\alpha u}(v) = \inp{v}{\alpha u} = \overline{\alpha}\inp{v}{u} = \alpha\inp{v}{u} = \alpha\Phi_u(v).
\end{equation*}
Thus $\Phi$ is indeed a linear map.
\item Suppose $\F = \C$ and $V \neq \{0\}$. Let $u\in V$. Then, given $v\in V$, we have
\begin{equation*}
\Phi_{iu}(v) = \inp{v}{iu} = \overline{i}\inp{v}{u},
\end{equation*}
whereas
\begin{equation*}
i\Phi_u(v) = i\inp{v}{u}.
\end{equation*}
Thus $\Phi_{iu} \neq i\Phi_u$, and indeed $\Phi$ is not a linear map, since is is not homogeneous.
\item Suppose $u, u'\in V$ are such that $\Phi_u = \Phi_{u'}$. Then, for all $v\in V$, we have
\begin{align*}
&\Phi_u(v) = \Phi_{u'}(v)\\
\implies &\inp{v}{u} = \inp{v}{u'}\\
\implies &\inp{v}{u} - \inp{v}{u'} = 0\\
\implies &\inp{v}{u - u'} = 0.
\end{align*}
In particular, choosing $v = u - u'$, the above implies $\inp{u - u'}{u - u'}=0$, which is true iff $u - u' = 0$. Thus we conclude $u = u'$, so that $\Phi$ is indeed injective.
\item Suppose $\F = \R$ and $\dim V = n$. Notice that since $\Phi: V\hookrightarrow V'$, we have
\begin{align*}
\dim V = \dim\Null\Phi + \dim\Range\Phi = \dim\Range\Phi.
\end{align*}
Thus $\Phi$ is surjective as well, and we have $V \cong V'$, as was to be shown. \qedhere
\end{enumerate}
\end{proof}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SECTION C
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{Orthogonal Complements and Minimization Problems}
% Problem 1
\begin{problem}{1}
Suppose $v_1,\dots, v_m \in V$. Prove that
\begin{equation*}
\{v_1, \dots, v_m\}^\perp = (\Span(v_1,\dots, v_m))^\perp.
\end{equation*}
\end{problem}
\begin{proof}
Suppose $v\in \{v_1,\dots, v_m\}^\perp$. Then $\inp{v}{v_k}=0$ for $k=1,\dots,m$. Let $u\in \Span(v_1,\dots, v_m)$ be arbitrary. We want to show $\inp{v}{u} = 0$, since this implies $v \in (\Span(v_1,\dots, v_m))^\perp$ and hence $\{v_1, \dots, v_m\}^\perp \subseteq (\Span(v_1,\dots, v_m))^\perp$. To see this, notice
\begin{align*}
\inp{v}{u} &= \inp{v}{\alpha_1v_1 + \dots + \alpha_m v_m}\\
&= \alpha_1\inp{v}{v_1} + \dots + \alpha_m\inp{v}{v_m}\\
&= 0,
\end{align*}
as desired. Next suppose $v'\in (\Span(v_1,\dots, v_m))^\perp$. Since $v_1,\dots,v_m$ are all clearly elements of $\Span(v_1,\dots, v_m)$, clearly $v'\in \{v_1, \dots, v_m\}^\perp$, and thus $(\Span(v_1,\dots, v_m))^\perp \subseteq \{v_1, \dots, v_m\}^\perp$. Therefore we conclude $\{v_1, \dots, v_m\}^\perp = (\Span(v_1,\dots, v_m))^\perp$.
\end{proof}
% Problem 3
\begin{problem}{3}
Suppose $U$ is a subspace of $V$ with basis $u_1,\dots, u_m$ and
\begin{equation*}
u_1, \dots, u_m, w_1, \dots, w_n
\end{equation*}
is a basis of $V$. Prove that if the Gram-Schmidt Procedure is applied to the basis of $V$ above, producing a list $e_1,\dots, e_m, f_1,\dots, f_n$, then $e_1,\dots, e_m$ is an orthonormal basis of $U$ and $f_1,\dots, f_n$ is an orthonormal basis of $U^\perp$.
\end{problem}
\begin{proof}
By 6.31, $\Span(u_1,\dots, u_m) = \Span(e_1,\dots, e_m)$. Since $e_1,\dots, e_m$ is an orthonormal list by construction (and linearly independent by 6.26), $e_1,\dots, e_m$ is indeed an orthonormal basis of $U$. Next, since each of $f_i$ is orthogonal to each $e_j$, so too is each $f_i$ orthogonal to any element of $U$. Thus $f_k\in U^\perp$ for $k=1,\dots,n$. Now, since $\dim U^\perp = \dim V - \dim U = n$ by 6.50, we conclude $f_1,\dots, f_n$ is an orthonormal list of length $\dim U^\perp$ and hence an orthonormal basis of $U^\perp$.
\end{proof}
% Problem 5
\begin{problem}{5}
Suppose $V$ is finite-dimensional and $U$ is a subspace of $V$. Show that $P_{U^\perp}= I - P_U$, where $I$ is the identity operator on $V$.
\end{problem}
\begin{proof}
For $v\in V$, write $v = u + w$, where $u\in U$ and $w \in U^\perp$. It follows
\begin{align*}
P_{U^\perp}(v) &= w\\
&= (u + w) - u\\
&= Iv - P_Uv,
\end{align*}
and therefore $P_{U^\perp}= I - P_U$, as desired.
\end{proof}
% Problem 7
\begin{problem}{7}
Suppose $V$ is finite-dimensional and $P\in\Hom(V)$ is such that $P^2 = P$ and every vector in $\Null P$ is orthogonal to every vector in $\Range P$. Prove that there exists a subspace $U$ of $V$ such that $P = P_U$.
\end{problem}
\begin{proof}
By Problem 4 of Chapter 5B, we know $V = \Null P \oplus \Range P$. Let $v\in V$. Then there exist $u\in \Null P$ and $w\in \Range P$ such that $v = u + w$ and hence
\begin{align*}
Pv &= P(u + w)\\
&= Pu + Pw\\