/
sol-new.tex
2459 lines (1918 loc) · 98.4 KB
/
sol-new.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
\stepcounter{chapter}
\setcounter{eqtn}{0}
\raggedcolumns\setlength{\multicolsep}{-0mm}
\begin{multicols}{2}
\parbf{\ref{ex:ell-infty}.} Check all the conditions in the definition of metric, page \pageref{page:def:metric}.
\parbf{\ref{ex:B2inB1}}; \ref{SHORT.ex:B2inB1:a}.
Observe that $\dist{p}{q}{\spc{X}}\le 1$.
Apply the triangle inequality to show that $\dist{p}{x}{\spc{X}}\le 2$ for any $x\in B[q,1]$.
Make a conclusion.
\parit{\ref{SHORT.ex:B2inB1:b}.}
Take $\spc{X}$ to be the half-line $[0,\infty)$ with the standard metric; $p=0$ and $q=\tfrac45$.
\parbf{\ref{ex:shrt=>continuous}.}
Show that the conditions in \ref{def:continuous} hold for $\delta=\epsilon$.
\parbf{\ref{ex:close-open}.}
Suppose the complement $\Omega=\spc{X}\setminus Q$ is open.
Then for each point $p\in \Omega$ there is $\epsilon>0$ such that $\dist{p}{q}{\spc{X}}>\epsilon$ for any $q\in Q$.
It follows that $p$ is \textit{not} a limit point of any sequence $q_n\in Q$.
That is, any limit of a sequence in $Q$ lies in $Q$;
by the definition, $Q$ is closed.
Now suppose $\Omega=\spc{X}\setminus Q$ is not open.
Show that there is a point $p\in \Omega$ and a sequence $q_n\in Q$ such that $\dist{p}{q_n}{\spc{X}}\z<\tfrac 1n$ for any~$n$.
Conclude that $q_n\to p$ as $n\to \infty$;
therefore $Q$ is not closed.
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\stepcounter{chapter}
\setcounter{eqtn}{0}
\parbf{\ref{ex:9}}; \ref{SHORT.ex:9:compact}. Use that a continuous injection defined on a compact domain is an embedding (\ref{thm:Hausdorff-compact}).
\parit{\ref{SHORT.ex:9:9}.} The image of $\gamma$ might have the shape of the digit $8$ or $9$.
\parbf{\ref{aex:simple-curve}.}
Let $\alpha$ be a path, connecting $p$ to~$q$.
Passing to a subarc of $\alpha$,
we can assume that $\alpha(t)\ne p,q$ for $t\ne0,1$.
An open set $\Omega$ in $(0,1)$ will be called {}\emph{suitable}
if, for any connected component $(a,b)$ of $\Omega$, we have $\alpha(a)=\alpha(b)$.
Show that the union of nested suitable sets is suitable.
Therefore, we can find a maximal suitable set $\hat \Omega$.
Define $\beta(t)=\alpha(a)$ for any $t$ in a connected component $(a,b)\subset\hat \Omega$, and $\beta (t) = \alpha (t) $ for $t\notin\hat{\Omega}$.
Note that for any $x\in [0,1]$ the set $\beta^{-1}\{\beta(x)\}$ is connected.
It remains to show that $\beta$ is a reparametrization of a simple path.
In order to do that, we need to construct a non-decreasing surjective function $\tau\:[0,1]\z\to[0,1]$ such that
$\tau(t_1)\z=\tau(t_2)$ if and only if there is a connected component $(a,b)\subset\hat \Omega$ such that $t_1,t_2\z\in [a,b]$.
The required function $\tau$ can be constructed similarly to the so-called {}\emph{devil's staircase} --- learn this construction and modify it.
The simple arc we are looking for is $\beta \circ \sigma$, where $\sigma\: [0,1]\to [0,1]$ is any right inverse of $\tau$.
\parbf{\ref{ex:L-shape}.}
Denote the union of two half-axis by~$L$.
Observe that $f(t)\to\infty$ as $t\to \infty$.
Since $f(0)=0$, the intermediate value theorem implies that $f(t)$ takes all nonnegative values for $t\ge 0$.
Use it to show that $L$ is the range of~$\alpha$.
Further, show that the function $f$ is strictly increasing for $t> 0$.
Use this to show that the map $t\mapsto \alpha(t)$ is injective.
\begin{wrapfigure}{r}{20 mm}
\vskip-7mm
\centering
\includegraphics{mppics/pic-270}
\vskip0mm
\end{wrapfigure}
Summarizing, we get that $\alpha$ is a smooth parametrization of~$L$.
Now suppose $\beta\:t\z\mapsto (x(t),y(t))$ is a smooth parametrization of~$L$.
Without loss of generality, we may assume that $x(0)\z=y(0)=0$.
Note that $x(t)\ge 0$ for any $t$ therefore $x'(0)=0$.
The same way we get that $y'(0)\z=0$.
That is, $\beta'(0)=0$;
so $L$ does not admit a smooth \textit{regular} parametrization.
\parbf{\ref{ex:cycloid}.}
Apply the definitions.
For \ref{SHORT.ex:cycloid:regular} you need to check that $\gamma'_\ell\ne 0$.
For \ref{SHORT.ex:cycloid:simple} you need to check that $\gamma_\ell(t_0)\z=\gamma(t_1)$ only if $t_0=t_1$.
\parbf{\ref{ex:nonregular}.}
Note that the parametrization $t\mapsto (t,t^3)$ is smooth and regular.
Modify it so it has zero speed at a point.
\parbf{\ref{ex:y^2=x^3}.}
This is the so-called \index{semicubical parabola}\emph{semicubical parabola}; it is shown on the diagram.
Try to argue similarly to \ref{ex:L-shape}.
\parbf{\ref{ex:viviani}.}
For $\ell=0$ the system describes a pair of points $(0,0,\pm1)$, so we can assume that $\ell\ne 0$.
Note that the first equation describes the unit sphere centered at the origin, and the second equation describes a cylinder over the circle in the $(x,y)$-plane with center at $(-\tfrac\ell2,0)$ and radius~$|\tfrac\ell2|$.
\begin{Figure}
\centering
\vskip-0mm
\begin{lpic}[t(2mm),b(0mm),r(0mm),l(0mm)]{asy/viviani(1)}
\lbl[r]{-.5,18;$x$}
\lbl[l]{41,22;$y$}
\lbl[r]{18,54;$z$}
\end{lpic}
\end{Figure}
Find the gradients $\nabla f$ and $\nabla h$ for the functions
\begin{align*}
f(x,y,z)&=x^2+y^2+z^2-1,
\\
h(x,y,z)&=x^2+\ell\cdot x+y^2.
\end{align*}
Show that for $\ell\ne 0$,
the gradients are linearly dependent only on the $x$-axis.
Conclude that for $\ell\ne\pm 1$ each connected component of the set of solutions is a smooth curve.
Show that
\begin{itemize}
\item if $|\ell|<1$, then the set has two connected components with $z>0$ and $z<0$.
\item if $|\ell|\ge1$, then the set is connected.
\end{itemize}
Note that the linear independence of the gradients provides only a sufficient condition.
Therefore, the case $\ell=\pm1$ has to be checked by hand.
In this case, a neighborhood of $(\pm1,0,0)$ does not admit a smooth regular parametrization --- try to prove it.
The case $\ell=1$ is shown on the diagram.
\parit{Remark.}
In the case $\ell=\pm1$, the curve is called \index{Viviani's curve}\emph{Viviani's curve}.
It admits the following smooth regular parametrization with a self-intersection:
$t\mapsto(\pm(\cos t)^2,\cos t\cdot\sin t,\sin t)$.
\parbf{\ref{ex:open-curve}.}
Assume the contrary, then there is a sequence of real numbers $t_n\to \pm \infty$ such that $\gamma(t_n)$ converges;
denote its limit by~$p$.
Let $K$ be a closed ball centered at~$p$.
Observe that $\gamma^{-1}(K)$ is not compact.
Conclude that $\gamma$ is not proper.
\parbf{\ref{ex:proper-closed}.}
Show and use that a set $C\subset \mathbb{R}^3$ is closed if and only if the intersection $K\cap C$ is compact for any compact $K\subset \mathbb{R}^3$.
\parbf{\ref{ex:proper-curve}.}
Without loss of generality, we may assume that the origin does not lie on the curve.
Show that inversion of the plane $(x,y)\z\mapsto (\tfrac{x}{x^2+y^2},\tfrac{y}{x^2+y^2})$ maps our curve to a closed curve with the origin removed.
Apply Jordan's theorem for the obtained curve, and use the inversion again.
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\stepcounter{chapter}
\setcounter{eqtn}{0}
\parbf{\ref{ex:integral-length-0}.}
Show that if we take the least upper bound in \ref{def:length} for all sequences
$a=t_0\le t_1\le\z\dots\le t_k=b$, then the result is the same.
Suppose $\gamma_2$ is reparametrization of $\gamma_1$ by $\tau\:[a_1,b_1]\to [a_2,b_2]$;
without loss of generality, we may assume that $\tau$ is nondecreasing.
Set $\theta_i=\tau(t_i)$.
Observe that $a_2=\theta_0\z\le\theta_1\le \z\dots\le\theta_k=b_2$ if and only if
$a_1=t_0\le t_1\le\z\dots\le t_k=b_1$.
Make a conclusion.
\parbf{\ref{ex:length-chain}.}
Show that for any inscribed polygonal line $\beta$ and any $\epsilon>0$ we have
\[\length\beta_n>\length\beta-\epsilon.\]
for all large $n$.
Conclude that
\[\liminf_{n\to\infty}\length\beta_n\ge \length \gamma.\]
Use the definition of length to show that
\[\limsup_{n\to\infty}\length\beta_n\le \length \gamma.\]
Observe that the two obtained inequalities imply the required statement.
\parbf{\ref{ex:length-image}.}
Choose a partition $0=t_0<\dots <t_n=1$ of $[0,1]$.
Set $\tau_0=0$ and
\[\tau_i=\max\set{\tau \in[0,1]}{\beta(\tau_i)=\gamma(t_i)}\]
for $i>0$.
Show that $(\tau_i)$ is a partition of $[0,1]$;
that is, $0=\tau_0<\tau_1<\dots<\tau_n=1$.
By construction
\begin{align*}
|\gamma(t_0)&-\gamma(t_1)|+|\gamma(t_1)-\gamma(t_2)|+\dots
\\
&\qquad\dots+|\gamma(t_{n-1})-\gamma(t_n)|=
\\
&=
|\beta(\tau_0)-\beta(\tau_1)|+|\beta(\tau_1)-\beta(\tau_2)|+\dots
\\
&\qquad\dots+|\beta(\tau_{n-1})-\beta(\tau_n)|.
\end{align*}
Since the partition $(t_i)$ is arbitrary, we get
\[\length \beta\ge \length \gamma.\]
\parit{Remarks.}
It is instructive to compare this exercise with \ref{obs:S2-length}.
The inequality might be strict.
It happens if $\beta$ runs back and forth along $\gamma$.
In this case the partition $(\tau_i)$ above is \textit{not} arbitrary.
If one assumes only that $\beta([0,1])\supset\gamma([0,1])$, then the problem becomes harder.
It follows from the following useful fact \cite[2.6.1+2.6.2]{burago-burago-ivanov}:
\textit{Denote by $h$ the 1-dimensional Hausdorff measure of the image $\gamma([0,1])$.
Then $h\le \length\gamma$;
moreover the equality holds if $\gamma$ is simple.}
\parbf{\ref{ex:integral-length}.}
For \ref{SHORT.ex:integral-length>}, apply the fundamental theorem of calculus for each segment in a given partition.
For \ref{SHORT.ex:integral-length<} consider a partition such that the velocity vector $\alpha'(t)$ is nearly constant on each of its segments.
\parbf{\ref{adex:integral-length}.}
Use the theorems of Rademacher and Lusin (\ref{thm:rademacher} and \ref{thm:lusin}).
\parbf{\ref{ex:nonrectifiable-curve}}; \ref{SHORT.ex:nonrectifiable-curve:a}.
Look at the diagram, and guess the parametrization of an arc of the snowflake by $[0,1]$.
Extend it to the whole snowflake.
Show that it indeed describes an embedding of the circle into the plane.
\begin{Figure}
\vskip-0mm
\centering
\includegraphics{mppics/pic-226}
\vskip0mm
\end{Figure}
\parit{\ref{SHORT.ex:nonrectifiable-curve:b}.}
Suppose $\gamma\:[0,1]\to\mathbb{R}^2$ is a rectifiable curve and let $\gamma_k$ be a scaled copy of $\gamma$ with factor $k>0$;
that is, $\gamma_k(t)=k\cdot\gamma(t)$ for any~$t$.
Show that
\[\length\gamma_k=k\cdot\length \gamma.\]
Now suppose the arc $\gamma$ of the Koch snowflake shown on the diagram is rectifiable; denote its length by $\ell$.
Observe that $\gamma$ can be divided into 4 arcs such that each one is a scaled copy of $\gamma$ with factor $\tfrac13$.
It follows that $\ell=\tfrac43\cdot\ell$.
Evidently, $\ell>0$ --- a contradiction.
\parbf{\ref{ex:cont-length}.}
Use \ref{thm:length-semicont} to show that $s$ is lower semicontinuous;
that is, if $t_n\to t_\infty$ as $n\to\infty$, then
\[\liminf_{n\to\infty} s(t_n)\ge s(t_\infty).\]
Observe that
\[s(t)=s(b)-\length(\gamma|_{[t,b]}).\]
Apply \ref{thm:length-semicont} to show that $s$ is upper semicontinuous.
Conclude that $s$ is continuous.
Finally, show and use that $s$ is nondecreasing.
\parbf{\ref{ex:arc-length-helix}.}
We have to assume $a\ne 0$ or $b\ne0$;
otherwise, we get a constant curve.
Show that $|\gamma'(t)|\equiv \sqrt{a^2+b^2}$;
in particular, the velocity is constant.
Therefore, $s\z=t/\sqrt{a^2+b^2}$ is an arc-length parameter.
It remains to substitute $s\cdot \sqrt{a^2+b^2}$ for~$t$.
\parbf{\ref{ex:convex-hull}.}
Choose a closed polygonal line $p_1\dots p_n$ inscribed in $\beta$.
By \ref{cor:convex=>rectifiable}, we can assume that its length is arbitrarily close to the length of $\beta$;
that is, given $\epsilon>0$
\[\length (p_1\dots p_n)>\length\beta-\epsilon.\]
Show that we may assume in addition that each point $p_i$ lies on $\alpha$.
Since $\alpha$ is simple, the points $p_1,\dots,p_n$ appear on $\alpha$ in the same cyclic order;
that is, the polygonal line $p_1\dots p_n$ is also inscribed in $\alpha$.
In particular,
\[\length\alpha\ge \length (p_1\dots p_n).\]
It follows that
\[\length\alpha>\length\beta-\epsilon.\]
for any $\epsilon>0$.
Whence
\[\length\alpha\ge\length\beta.\]
\begin{wrapfigure}{r}{25 mm}
\vskip-6mm
\centering
\includegraphics{mppics/pic-275}
\vskip0mm
\end{wrapfigure}
If $\alpha$ has self-intersections, then the points $p_1,\dots, p_n$ might appear on $\alpha$ in a different cyclic order, say $p_{i_1},\dots,p_{i_n}$.
Apply the triangle inequality to show that
\[\length(p_{i_1}\dots p_{i_n})\ge \length (p_1\dots p_n)\]
and use it to modify the given proof.
\parbf{\ref{ex:convex-croftons}.}
Denote by $\ell_{\vec u}$ the line segment
obtained by orthogonal projection of $\gamma$ to the line in the direction ${\vec u}$.
Since $\gamma_{\vec u}$ runs along $\ell_{\vec u}$ back and forth, we get
\[\length\gamma_{\vec u}\ge 2\cdot\length\ell_{\vec u}.\]
By the Crofton formula,
\[\length\gamma\ge \pi\cdot \overline{\length\ell_{\vec u}}.\]
In the case of equality, the curve $\gamma_{\vec u}$ runs exactly back and forth along $\ell_{\vec u}$ without additional zigzags for almost all (and therefore for all) ${\vec u}$.
Let $K$ be a closed set bounded by~$\gamma$.
Observe that the last statement implies that every line may intersect $K$ only along a closed segment or a single point.
It follows that $K$ is convex.
\parbf{\ref{adex:more-croftons}.}
The proof is identical to the proof of the standard Crofton formula.
To find the coefficients it is sufficient to check it on a unit interval.
The latter can be done by integration:
\begin{align*}
\frac1{k_a}&=\frac{1}{\area \mathbb{S}^2}\cdot\iint_{\mathbb{S}^2} |x|;
\\
\frac1{k_b}&=\frac{1}{\area \mathbb{S}^2}\cdot\iint_{\mathbb{S}^2} \sqrt{1-x^2}.
\end{align*}
The answers are $k_a=2$ and $k_b=\tfrac4\pi$.
\parbf{\ref{ex:intrinsic-convex}.}
The only-if part is trivial.
To show the if part, assume $A$ is not convex;
that is, there are points $x,y\in A$ and a point $z\notin A$ that lies between $x$ and~$y$.
Since $A$ is closed, its complement is open.
That is, the ball $B(z,\epsilon)$ does not intersect $A$ for some $\epsilon>0$.
Show that there is $\delta>0$ such that any path of length at most $\dist{x}{y}{\mathbb{R}^3}+\delta$ passes thru $B(z,\epsilon)$.
It follows that $\dist{x}{y}A\ge \dist{x}{y}{\mathbb{R}^3}+\delta$;
in particular, $\dist{x}{y}A\ne \dist{x}{y}{\mathbb{R}^3}$.
\parbf{\ref{ex:antipodal}.}
The spherical curve shown on the diagram does not have antipodal pairs of points.
However, it has three points $x,y,z$ on one of its sides and their antipodal points $-x,-y,-z$ on the other.
\begin{wrapfigure}[6]{r}{23 mm}
\vskip-6mm
\centering
\includegraphics{mppics/pic-280}
\vskip0mm
\end{wrapfigure}
Show that this property is sufficient to conclude that the curve does not lie in any hemisphere.
\parbf{\ref{ex:bisection-of-S2}.}
Assume the contrary, then by the hemisphere lemma (\ref{lem:hemisphere}) $\gamma$ lies in an open hemisphere.
In particular, it cannot divide $\mathbb{S}^2$ into two regions of equal area --- a contradiction.
\parbf{\ref{ex:flaw}.}
The first sentence is wrong --- it is \textit{not} sufficient to show that the diameter is at most~2.
For example, if an equilateral triangle has circumradius slightly above $1$,
then its diameter (which is defined as the maximal distance between its points) is slightly bigger than $\sqrt3$, so it is smaller than $2$.
On the other hand, it is easy to modify the proof of the hemisphere lemma (\ref{lem:hemisphere}) to get a correct solution.
That is, (1) choose two points $p$ and $q$ on $\gamma$ that divide it into two arcs of the same length;
(2) set $z$ to be a midpoint of $p$ and $q$,
and (3) show that $\gamma$ lies in the unit disc centered at~$z$.
\parbf{\ref{adex:crofton}.}
For \ref{SHORT.adex:crofton:crofton}, modify the proof of the original Crofton formula
(Section~\ref{sec:crofton}).
\parit{\ref{SHORT.adex:crofton:hemisphere}.}
Assume $\length \gamma<2\cdot\pi$.
By \ref{SHORT.adex:crofton:crofton},
\[\overline{\length \gamma^*_{\vec u}}<2\cdot\pi.\]
Therefore, we can choose ${\vec u}$ so that
\[\length \gamma^*_{\vec u}<2\cdot\pi.\]
Observe that $\gamma^*_{\vec u}$ runs in a semicircle~$h$.
Therefore $\gamma$ lies in a hemisphere with $h$ as a diameter.
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\stepcounter{chapter}
\setcounter{eqtn}{0}
\parbf{\ref{ex:zero-curvature-curve}.}
For a unit-speed parametrization $\gamma$, we have $\gamma''\equiv 0$.
Therefore, the velocity $\vec v=\gamma'$ is constant.
Hence, $\gamma(t)=p+(t-t_0)\cdot \vec v$, where $p=\gamma(t_0)$.
\parbf{\ref{ex:scaled-curvature}.}
Observe that $\alpha(t)\df\gamma_{\lambda}(t/\lambda)$ is a unit-speed parametrization of the curve $ \gamma_{\lambda}$.
Apply the chain rule twice.
\parbf{\ref{ex:curvature-of-spherical-curve}.} Differentiate the identity $\langle\gamma(s),\gamma(s)\rangle=1$ a couple of times.
\parbf{\ref{ex:curvature-formulas}.}
Set $\tan=\tfrac{\gamma'}{|\gamma|}$.
Prove and use the following identities:
\begin{align*}
\gamma''-(\gamma'')^\perp&=\tan\cdot\langle\gamma'',\tan\rangle,
&
|\gamma'|&=\sqrt{\langle \gamma',\gamma'\rangle}.
\end{align*}
\parbf{\ref{ex:curvature-graph}.}
Apply \ref{ex:curvature-formulas:a} for the parametrization $t\z\mapsto (t,f(t))$.
\parbf{\ref{ex:approximation-const-curvature}.}
Without loss of generality, we may assume that $\gamma$ has a unit-speed parametrization.
Consider the tangent indicatrix $\tan(s)\z=\gamma'(s)$.
Note that $\tan$ is a spherical curve and $|\tan'|\le 1$.
Use it to construct a sequence of unit-speed spherical curves $\tan_n\:\mathbb{I}\to\mathbb{S}^2$ such that $\tan_n(s)\to \tan(s)$ as $n\to\infty$ for any~$s$.
Show that the following sequence of curves solves the problem:
\[\gamma_n(s)=\gamma(a)+\int_a^s\tan_n(t)\cdot dt.\]
\parbf{\ref{ex:no-parallel-tangents}.}
Reuse the construction in \ref{ex:antipodal} to get a closed smooth simple spherical curve $\tan\:\mathbb{S}^1\z\to\mathbb{S}^2$ that contains no pair of antipodal points and has zero average.
Then construct a curve with tangent indicatrix $\tan$.
(It was constructed by Beniamino Segre \cite{segre}.)
\parbf{\ref{ex:helix-curvature}.}
Show that $\gamma_{a,b}''\perp \gamma'_{a,b}$, and apply \ref{ex:curvature-formulas:a}.
\parbf{\ref{ex:length>=2pi}.} Apply Fenchel's theorem.
\parbf{\ref{ex:gamma/|gamma|}.}
We can assume that $\gamma$ is unit-speed.
Set $\theta(s)=\measuredangle(\gamma(s),\gamma'(s))$.
Since $\langle\tan,\tan\rangle=1$, we have $\tan'\perp \tan$.
Observe that $\langle \tan,\sigma\rangle=\cos\theta$;
therefore
\begin{align*}
\kur\cdot \sin\theta
&=|\tan'|\cdot \sin\theta\ge
-\langle \tan',\sigma\rangle=
\\
&=
\langle \tan,\sigma'\rangle-\langle \tan,\sigma\rangle'=
\\
&=
(|\sigma'|+\theta')\cdot \sin\theta.
\end{align*}
Whence $\kur\ge |\sigma'|+\theta'$
if $\theta\ne0,\pi$.
It remains to integrate this inequality and show that the set with $\theta\ne0$ or $\pi$ does not create a problem.
\parit{An alternative solution} can be built on \ref{prop:inscribed-total-curvature}.
\parbf{\ref{ex:DNA}}; \ref{SHORT.ex:DNA:c''c>=k}.
Since $|\gamma(s)|\le 1$, we have
\begin{align*}
\langle\gamma''(s),\gamma(s)\rangle&\ge -|\gamma''(s)|\cdot|\gamma(s)|\ge-\kur(s)
\end{align*}
for all~$s$.
\parit{\ref{SHORT.ex:DNA:int>=length-tc}.}
Since $\gamma$ is unit-speed, we have $\ell=\length\gamma$ and $\langle\gamma',\gamma'\rangle\equiv1$.
Therefore,
\[\int_0^\ell\langle\gamma(s),\gamma'(s)\rangle'\cdot ds
=\]
\[=\int_0^\ell\langle\gamma'(s),\gamma'(s)\rangle\cdot ds+\int_0^\ell\langle\gamma(s),\gamma''(s)\rangle\cdot ds\ge\]
\[\ge \length\gamma-\tc\gamma.\]
\parit{\ref{SHORT.ex:DNA:end}.}
By the fundamental theorem of calculus, we have
\begin{align*}
\int_0^\ell\langle\gamma(s),\gamma'(s)\rangle'\cdot ds
&=\langle\gamma(s),\gamma'(s)\rangle\bigg|_0^\ell.
\end{align*}
Since $\gamma(0)=\gamma(\ell)$ and $\gamma'(0)=\gamma'(\ell)$, the right-hand side vanishes.
Note that without loss of generality, we can assume the curve in \ref{thm:DNA} is described by a loop $\gamma\:[0,\ell]\to\mathbb{R}^3$ parametrized by arc-length.
We can also assume that the origin is the center of the ball; that is $|\gamma|\le 1$.
Since $\gamma$ is a smooth closed curve, we have
$\gamma'(0)=\gamma'(\ell)$ and $\gamma(0)=\gamma(\ell)$.
Therefore, \ref{SHORT.ex:DNA:int>=length-tc} and \ref{SHORT.ex:DNA:end} imply \ref{thm:DNA}.
\parbf{\ref{ex:tangent-support}.}
Show that no line distinct from $\ell$ can support $F$ at $p$.
Apply \ref{lem:separation} to show that there is a line supporting $F$ at $p$.
Conclude that it must be $\ell$.
{
\begin{wrapfigure}{r}{22 mm}
\vskip-6mm
\centering
\includegraphics{mppics/pic-255}
\vskip0mm
\end{wrapfigure}
\parbf{\ref{ex:anti-bow}.}
Start with the curve $\gamma_1$ shown on the diagram.
To obtain $\gamma_2$, slightly unbend (that is, decrease the curvature of) the dashed arc of $\gamma_1$.
\parbf{\ref{ex:length-dist}}; \ref{SHORT.ex:length-dist:>}.
Choose a value $s_0\in[a,b]$ that splits the total curvature of $\gamma$ into two equal parts.
Observe that $\measuredangle(\gamma'(s_0),\gamma'(s))\le \theta$ for any~$s$.
Use this inequality in the same way as in the proof of the bow lemma.
Part \ref{SHORT.ex:length-dist:self-intersection:>pi} is straightforward.
For part \ref{SHORT.ex:length-dist:=} consider the concatenation of two equal line segments with the external angle $2\cdot\theta$ and smooth its joint.
}
\parbf{\ref{ex:schwartz}.}
Let $\ell=\length\gamma$.
Suppose $\ell_1<\ell<\ell_2$.
Let $\gamma_1$ be an arc of the unit circle with length $\ell$.
Show that the distance between the endpoints of $\gamma_1$ is smaller than $|p-q|$, and apply the bow lemma (\ref{lem:bow}).
\parbf{\ref{ex:loop}.}
Suppose $\length\gamma<2\cdot\pi$.
Apply the bow lemma (\ref{lem:bow}) to $\gamma$ and an arc of the unit circle of the same length.
\parbf{\ref{ex:bow-upper}.}
Choose a smooth spherical curve $\alpha$ that runs near one point;
for example, a small spherical circle centered at this point.
Consider a curve $\tan$ that runs along $\alpha$ with speed $\kappa(s)$ at any $s\in [0,\ell]$.
Show that a curve with tangent indicatrix $\tan$ solves the problem.
\parbf{\ref{ex:gromov-twist}.}
We can assume that $\gamma$ is unit-speed.
Suppose the inequality fails at $t=0$.
We may assume that $\alpha(0)\le\tfrac\pi2$;
otherwise, revert the parametrization.
In the plane spanned by $\gamma(0)$ and~$\gamma'(0)$, choose a unit-speed circline arc $\sigma$ from $0$ to $\gamma(0)$ that comes to $\gamma(0)$ in the direction opposite to $\gamma'(0)$.
Consider a unit-speed semicircle $\tilde\gamma$ with curvature $2$
that starts at $\gamma(0)$ in the direction $\gamma'(0)$ so that that the concatenation $\sigma*\tilde\gamma$ is an arc of a convex plane curve; see the figure.
\begin{Figure}
\vskip-1mm
\centering
\includegraphics{mppics/pic-282}
\vskip-1mm
\end{Figure}
Show that if $|\gamma(0)|> \sin[\alpha(0)]$, then $\tilde\gamma$ leaves the unit ball; that is, $|\tilde\gamma(t_0)|>1$ for some~$t_0$.
The concatenations $\sigma*\gamma$ and $\sigma*\tilde\gamma$ fail to be smooth at the joint, but it is differentiable at this point.
Check that the proof of the bow lemma works in this more general case.
Apply the bow lemma for $\sigma*\gamma$ and $\sigma*\tilde\gamma$, to get $|\gamma(t_0)|\ge |\tilde\gamma(t_0)|$, and arrive at a contradiction.
\parit{Source:} This statement was used by the first author \cite{petrunin2023}.
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\parbf{\ref{ex:chord-lemma-optimal}.}
Set $\alpha=\measuredangle(\vec w,\vec u)$
and $\beta=\measuredangle(\vec v,\vec u)$.
Try to guess the example from the diagram.
\begin{Figure}
\vskip-1mm
\centering
\includegraphics{mppics/pic-285}
\vskip-1mm
\end{Figure}
The shown curve is divided into three arcs: I, II, and III.
Arc I turns from $\vec v$ to $\vec u$;
it has total curvature $\alpha$.
Similarly, arc III turns from $\vec u$ to $\vec w$ and has total curvature $\beta$.
Arc II goes very close and almost parallel to the chord $pq$, and its total curvature can be made arbitrarily small.
\parbf{\ref{ex:monotonic-tc}.}
Use that an exterior angle of a triangle equals the sum of the two remote interior angles.
For the second part apply induction on the number of vertices.
\parbf{\ref{ex:sef-intersection}};
\ref{SHORT.ex:sef-intersection:>pi}. Assume $x$ is a point of self-intersection.
Show that we may choose two points $y$ and $z$ on $\gamma$ so that the triangle $xyz$ is nondegenerate.
In particular,
$\measuredangle\hinge yxz
\z+
\measuredangle\hinge zyx
<\pi$, or, equivalently, for the \textit{nonclosed} polygonal line $xyzx$ we have $\tc{xyzx}\z>\pi$.
It remains to apply \ref{prop:inscribed-total-curvature}.
\begin{wrapfigure}{r}{23 mm}
\vskip-3mm
\centering
\includegraphics{mppics/pic-290}
\vskip-0mm
\end{wrapfigure}
\parit{\ref{SHORT.ex:sef-intersection:<2pi}.}
See the diagram.
\parbf{\ref{ex:quadrisecant}.}
Consider the closed polygonal line $acbd$.
Observe that $\tc{acbd}=4\cdot\pi$.
It remains to apply \ref{prop:inscribed-total-curvature}.
\parbf{\ref{ex:total-curvature=}.}
By \ref{prop:inscribed-total-curvature}, $\tc\gamma\ge \tc\beta$;
it remains to show that
$\tc\gamma\le\sup\{\tc\beta\}$.
In other words,
given $\epsilon>0$ and a polygonal line $\sigma=\vec u_0\dots \vec u_k$ inscribed in the tangent indicatrix $\tan$ of $\gamma$,
we need to construct a polygonal line $\beta$ inscribed in $\gamma$ such that
\[\length\sigma<\tc\beta+\epsilon.
\eqlbl{eq:tc=<tc}\]
Suppose $\vec u_i=\tan(s_i)$.
Choose an inscribed polygonal line $\beta=p_0\dots p_{2\cdot k+1}$ such that $p_{2\cdot i}$ and $p_{2\cdot i+1}$ lie sufficiency close to $\gamma(s_i)$; so we can assume that the direction of $p_{2\cdot i+1}-p_{2\cdot i}$ is sufficiency close to $\vec u_i$ for each~$i$.
Show that \ref{eq:tc=<tc} holds true for the constructed polygonal line~$\beta$.
\parbf{\ref{ex:tc-rectifiable}.}
Show that for any polygonal line $\beta$ in a ball of radius $R$, we have
\[\tc{\beta}+2\cdot \pi\ge\frac{\length\beta}R.\]
Observe that $\gamma$ lies in a ball and apply this inequality.
An example for the second question can be found among logarithmic spirals.
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\stepcounter{chapter}
\setcounter{eqtn}{0}
\parbf{\ref{ex:helix-torsion}.}
The arc-length parameter $s$ is already found in \ref{ex:arc-length-helix}.
It remains to find the Frenet frame and calculate curvature and torsion.
The latter can be done by straightforward calculations.
{
\begin{wrapfigure}{r}{25 mm}
\vskip-6mm
\centering
\begin{lpic}[t(-0mm),b(0mm),r(0mm),l(0mm)]{asy/helix(1)}
\lbl[br]{8,24;$\norm$}
\lbl[b]{2,26;$\bi$}
\lbl[wl]{15,25;$\tan$}
\end{lpic}
\vskip-0mm
\end{wrapfigure}
The answers are
\begin{align*}
\tan(t)&=\tfrac{(-a\cdot\sin t, a\cdot\cos t,b)}{\sqrt{a^2+b^2}},
\\
\norm(t)&=(-\cos t,-\sin t,0),
\\
\bi(t)&=\tfrac{(b\cdot \sin t,-b\cdot \cos t, a)}{\sqrt{a^2+b^2}},
\\
\kur &\equiv \tfrac{a}{a^2+b^2},
\\
\tor &\equiv \tfrac{b}{a^2+b^2}.
\end{align*}
It remains to show that the map $(a,b) \z\mapsto (\frac{a}{a^2+b^2}, \frac{b}{a^2+b^2})$ sends bijectively the half plane $a>0$ onto itself.
}
\parbf{\ref{ex:beta-from-tau+nu}.} By the product rule, we get
\begin{align*}
\bi'&=(\tan\times \norm)'=
\tan'\times \norm+\tan\times\norm'.
\end{align*}
It remains to substitute the values from \ref{eq:frenet-tau} and \ref{eq:frenet-nu} and simplify.
\parbf{\ref{ex:torsion=0}.}
This is a consequence of the equation $\bi' = - \tor\cdot \norm $.
\parbf{\ref{ex:+B}.}
We can assume that $\gamma_0$ is unit-speed.
Show that
$|\gamma_1'|=|\tan-\tor\cdot \norm|\ge 1$ and use it.
\parbf{\ref{ex:frenet}.}
Observe that $\tfrac{\gamma'\times\gamma''}{|\gamma'\times\gamma''|}$ is a unit vector perpendicular to the plane spanned by $\gamma'$ and $\gamma''$, so, up to sign, it has to be equal to $\bi$.
It remains to check that the sign is right.
\parbf{\ref{ex:moment-curve}.} Apply \ref{ex:curvature-formulas:b} and \ref{ex:frenet}.
\parbf{\ref{ex:torsion-indicatrix}.} Assume that the tangent indicatrix has no self-intersection.
Show that it lies in an open hemisphere and argue as in Fenchel's theorem.
\parbf{\ref{ex:lancret}}; \ref{SHORT.ex:lancret:a}.
Observe that
$\langle \vec w,\tan\rangle'=0$.
Show that it implies that $\langle \vec w, \norm\rangle =0$.
Further, show and use that $\langle \vec w, \tan\rangle^2+\langle \vec w, \bi\rangle^2=\langle \vec w, \vec w\rangle$.
To prove the last identity, apply the Frenet formula for~$\norm'$.
\parit{\ref{SHORT.ex:lancret:b}.}
Show that $\vec w'=0$;
it implies that $\langle \vec w,\tan\rangle\z=\tfrac\tor\kur$.
In particular, the velocity vector of $\gamma$ makes a constant angle with $\vec w$; that is, $\gamma$ has a constant slope.
\parbf{\ref{ex:evolvent-constant-slope}.}
Suppose $\alpha$ is an evolvent of $\gamma$, and $\vec w$ is a fixed vector.
Show that $\langle \vec w,\alpha\rangle$ is constant if $\gamma$ makes a constant angle with $\vec w$.
\parbf{\ref{ex:spherical-frenet}}.
Part \ref{SHORT.ex:spherical-frenet:tau} follows since $(\tan,\norm,\bi)$ is an orthonormal basis.
For \ref{SHORT.ex:spherical-frenet:nu}, take the first and second derivatives of the identity $\langle\gamma,\gamma\rangle=1$ and simplify them using the Frenet formulas.
Part \ref{SHORT.ex:spherical-frenet:beta} follows from \ref{SHORT.ex:spherical-frenet:nu} and the Frenet formulas.
By \ref{SHORT.ex:spherical-frenet:beta}, $\int\tfrac\tor\kur=0$, hence \ref{SHORT.ex:spherical-frenet:beta+} follows.
Part \ref{SHORT.ex:spherical-frenet:kur-tor} is proved by algebraic manipulations.
For \ref{SHORT.ex:spherical-frenet:f},
use the Frenet formulas to show that $(\gamma+\tfrac1\kur\cdot \norm+\tfrac{\kur'}{\kur^2\cdot\tor}\cdot\bi)'=0$.
(Part \ref{SHORT.ex:spherical-frenet:beta} implies that \textit{$\int\tfrac\tor\kur=0$ for any closed smooth spherical curve}.
In fact this property characterizes spheres and planes; the latter was proved by Beniamino Segre \cite{segre}.)
\parbf{\ref{ex:cur+tor=helix}.} Use the second statement in \ref{ex:helix-torsion}.
\parbf{\ref{ex:const-dist}.} Note that the function
\begin{align*}
\rho(\ell)&=|\gamma(t+\ell)-\gamma(t)|^2=
\\
&=\langle \gamma(t+\ell)-\gamma(t),\gamma(t+\ell)-\gamma(t)\rangle
\end{align*}
is smooth and does not depend on~$t$.
Express speed, curvature, and torsion of $\gamma$ in terms of the derivatives $\rho^{(n)}(0)$.
Be patient, you will need two derivatives for the speed,
four for the curvature,
and six for the torsion.
Once it is done, apply \ref{ex:cur+tor=helix}.
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\stepcounter{chapter}
\setcounter{eqtn}{0}
\parbf{\ref{ex:bike}.}
Without loss of generality, we may assume that $\gamma_0$ is parametrized by its arc-length.
Then
\begin{align*}
|\gamma_1'|&=|\gamma_0'+\tan'|=|\tan+\kur\cdot\norm|.
\end{align*}
It follows that
\[|\gamma_1'(t)|\ge|\gamma_0'(t)|
\quad\text{and}\quad
|\gamma_1'(t)|\ge|\kur(t)_{\gamma_0}|
\]
for any $t\in[a,b]$.
Integrate these inequalities and apply
\ref{ex:integral-length}.
\parbf{\ref{ex:trochoids}.}
Observe that
\[\gamma'_a(t)=(1+a\cdot \cos t, -a\cdot \sin t);\]
that is, $\gamma'_a$ runs clockwise along a circle with center at $(1,0)$ and radius $\vert a \vert$.
\parit{Case $|a|>1$.} Note that $\tan_a(t)=\gamma'_a/|\gamma'_a|$ runs clockwise and makes a full turn in time $2\cdot\pi$.
Therefore, $\tgc{\gamma_a}=-2\cdot\pi$ and $\tc\gamma_a\z=|\tgc{\gamma_a}|=2\cdot\pi$.
\parit{Case $|a|<1$.}
Set $\theta_a=\arcsin a$.
Show that $\tan_a(t)=\gamma'_a/|\gamma'_a|$ starts with the horizontal direction $\tan_a(0)=(1,0)$, turns monotonically to angle $\theta_a$, then monotonically to $-\theta_a$ and then monotonically back to $\tan_a(2\cdot\pi)=(1,0)$.
It follows that if $|a|>1$, then
$\tgc{\gamma_a}=0$ and $\tc{\gamma_a}=4\cdot\theta_a$.
\parit{Case $a=-1$.}
The velocity $\gamma'_{-1}(t)$ vanishes at $t=0$ and $2\cdot\pi$.
Nevertheless, the curve admits a smooth regular parametrization --- find it.
The answer is $\tc{\gamma_{-1}}=-\tgc{\gamma_{-1}}=\pi$.
\parit{Case $a=1$.}
The velocity $\gamma'_1(t)$ vanishes at $t=\pi$.
At $t=\pi$ the curve has a cusp;
that is, $\gamma_1$ turns exactly back at the time $\pi$.
So $\gamma_1(t)$ has undefined total signed curvature.
The curve is a joint of two smooth arcs with external angle $\pi$, and
the total curvature of each arc is $\tfrac\pi2$, so
$\tc{\gamma_{1}}=\tfrac\pi2+\pi+\tfrac\pi2=2\cdot\pi$.
\parbf{\ref{ex:zero-tsc}.}
The answers are shown.
In the last picture, we assume that the two marked points have parallel tangent lines.
\begin{Figure}
\begin{minipage}{.27\textwidth}
\centering
\includegraphics{mppics/pic-260}
\end{minipage}\hfill
\begin{minipage}{.42\textwidth}
\centering
\includegraphics{mppics/pic-261}
\end{minipage}
\hfill
\begin{minipage}{.27\textwidth}
\centering
\includegraphics{mppics/pic-262}
\end{minipage}
\end{Figure}
\parbf{\ref{ex:length'}}; \ref{SHORT.ex:length':reg}.
Show that
\[
\gamma_\ell'(t)=(1-\ell\cdot\skur(t))\cdot \gamma'(t).
\]
By regularity of $\gamma$, $\gamma'\ne0$.
So if $\gamma_\ell'(t)=0$, then $\ell\cdot \skur(t)= 1$.
\parit{\ref{SHORT.ex:length':formula}.} We can assume that $\gamma$ is parametrized by its arc-length, so $\gamma'(t)=\tan(t)$.
Suppose $|\ell|<\frac{1}{\kur(t)}$ for any~$t$.
Then
\[
|\gamma_\ell'(t)|=(1-\ell\cdot\skur(t)).
\]
Therefore,
\begin{align*}
L(\ell)
&=
\int_a^b(1-\ell\cdot\skur(t))\cdot dt=
\\&=
\int_a^b 1\cdot dt-\ell\cdot \int_a^b\skur(t)\cdot dt=
\\
&=
L(0)+\ell\cdot\tgc\gamma.
\end{align*}
\parit{\ref{SHORT.ex:length':antiformula}.}
Consider the unit circle $\gamma(t)=(\cos t,\sin t)$ for $t\in[0,2\cdot\pi]$ and $\gamma_\ell$ for $\ell=2$.
\parbf{\ref{ex:inverse}.}
Use the definition of osculating circle via order of contact, and that inversions send circles to circlines.
\parbf{\ref{ex:evolute}.}
We assume that $\gamma$ is unit-speed.
Show that $\omega'=\tfrac{\skur'}{\skur^2}\cdot \norm$.
Conclude that $\omega$ has Frenet frame either $(\norm,-\tan)$ or $(-\norm,\tan)$, and its curvature is $|\tfrac{\kur^3}{\kur'}|$.
\parbf{\ref{ex:3D-spiral}.}
Start with a plane spiral curve as shown on the diagram.
Applying \ref{thm:fund-curves}, increase the torsion of the dashed arc without changing the curvature until a self-intersection appears.
\begin{wrapfigure}{r}{19 mm}
\vskip-0mm
\centering
\includegraphics{mppics/pic-296}
\vskip-0mm
\end{wrapfigure}
You may think that the dashed arc is very short, so the tangent vector $\tan$ is nearly constant on this arc.
Note that increasing the torsion can rotate normal vector $\norm$ arbitrary around $\tan$.
An intersection appears if $\norm$ is rotated by certain angle near $\pi$.
(Compare to \ref{ex:approximation-const-curvature}.)
\parbf{\ref{ex:double-tangent}.}
Observe that if a line or circle is tangent to $\gamma$,
then it is tangent to the osculating circle at the same point.
Then apply the spiral lemma (\ref{lem:spiral}).
\parbf{\ref{ex:spherical-spiral}.}
Note that osculating circles of a spherical curve lie in the sphere.
Prove an analog of \ref{lem:spiral} for these circles.
(Compare to \ref{ex:spherical-frenet:beta+}.)
%\end{multicols}
%\par\noindent\rule{\textwidth}{0.4pt}
%\begin{multicols}{2}
\stepcounter{chapter}
\setcounter{eqtn}{0}
\parbf{\ref{ex:vertex-support}.}
Apply the spiral lemma (\ref{lem:spiral}).
\parit{Computational solution.}
We will assume that the curvature does not vanish at $p$, the remaining case is simpler.
We may assume that $\gamma$ is unit-speed, $p=\gamma(0)$,
and the center of curvature of $\gamma$ at $p$ is the origin;
in other words, $\kur(0)\cdot\gamma(0)+ \norm(0)=0$.
Consider the function $f\:t\mapsto \langle\gamma(t),\gamma(t)\rangle$.
Direct calculations show that
\begin{align*}
f'
&=\langle\gamma,\gamma\rangle'
=2\cdot\langle\tan,\gamma\rangle,
\\
f''
&=2\cdot\langle\tan,\gamma\rangle'
=2\cdot\kur\cdot\langle\norm,\gamma\rangle+2,
\\
f'''
&=2\cdot\kur'\cdot\langle\norm,\gamma\rangle
+2\cdot\kur\cdot\langle\norm',\gamma\rangle+\cancel{2\cdot\kur\cdot\langle\norm,\tan\rangle}.
\end{align*}
Observe that $\norm'(0)\perp\gamma(0)$.
Therefore, $f'(0)=0$, $f''(0)=0$, and $f'''(0)\z=-2\cdot\kur'/\kur$.
If the osculating circle supports $\gamma$ at $p$,
then the function $f$ has a local maximum or minimum at $0$.
Therefore $f'''(0)=0$ and $\kur'=0$.
\parbf{\ref{ex:support}.}
Consider the coordinate system with the origin at $p$ and the common tangent line to $\gamma_1$ and $\gamma_2$ as the $x$-axis.
We may assume that $\gamma_1$ and $\gamma_2$ are defined in $(-\epsilon,\epsilon)$ for small $\epsilon>0$,
and they run almost horizontally.
Given $t\in[0,1]$ consider the curve $\gamma_t$ that is tangent and cooriented to the $x$-axis at $\gamma_t(0)$ and has signed curvature defined by $\skur_t(s)\z\df(1\z-t)\cdot\skur_0(s)+t\cdot\skur_1(s)$.
It exists by \ref{thm:fund-curves-2D}.
Choose $s\approx 0$.
Consider the curve $\alpha_s\:t\z\mapsto \gamma_t(s)$.
Show that $\alpha_s$ moves almost vertically up.
Use that $\gamma_t$ moves almost horizontally to show that in a small neighborhood of $p$, the curve $\gamma_1$ lies above $\gamma_0$,
whence the statement follows.
\parbf{\ref{ex:in-circle}.}
Move the unit circle straight until its center gets to the base of the loop,
and then reduce its radius to zero.
Show that at the moment when the circle first touches the loop, it supports the loop at a point distinct from the base.
Apply~\ref{prop:supporting-circline}.
\parbf{\ref{ex:between-parallels-1} $\bm{+}$ \ref{ex:in-triangle} $\bm{+}$ \ref{ex:lens}.}
Observe that one of the arcs of curvature 1 in the families shown on the diagram supports $\gamma$, and apply \ref{prop:supporting-circline}.
\begin{Figure}
\begin{minipage}{.35\textwidth}
\centering
\includegraphics{mppics/pic-265}
\end{minipage}
\hfill
\begin{minipage}{.3\textwidth}
\centering
\includegraphics{mppics/pic-266}
\end{minipage}
\hfill
\begin{minipage}{.25\textwidth}
\centering
\includegraphics{mppics/pic-267}
\end{minipage}
\end{Figure}
The second part in \ref{ex:between-parallels-1} can be reduced to \ref{ex:in-circle} using the shown family and another family of arcs curved in the opposite direction.
\parit{Remark.} Compare to \ref{ex:moon-rad}.
\begin{wrapfigure}{r}{34 mm}
\vskip-6mm
\centering
\includegraphics{mppics/pic-268}
\vskip0mm
\end{wrapfigure}
\parbf{\ref{ex:convex small}.}
Note that we can assume that $\gamma$ bounds a convex figure~$F$.
Otherwise, by \ref{prop:convex} its curvature changes sign.
Therefore, $\gamma$ has zero curvature at some point.
Choose two points $x$ and $y$ surrounded by $\gamma$ such that $|x\z-y|\z>2$.
Look at the maximal lens bounded by two arcs with a common chord $xy$ that lies in~$F$.
Apply the supporting test (\ref{prop:supporting-circline}).
\parbf{\ref{ex:convex-lens}.}
Apply the lens lemma (\ref{lem:lens}) to show that $\gamma$ lies on one side from the line $pq$.
Conclude that the union of the arc $\gamma$ and its chord $[p,q]$ is a simple closed curve;
by Jordan's theorem, it bounds a figure, say~$F$.
Assume that $F$ is not convex and arrive at a contradiction as in \ref{prop:convex}.