Skip to content

Commit 09f2ea7

Browse files
committed
Fourier 1 : finished section 1 (i.e. 1.2 and 1.3 today)
1 parent 0ae93a8 commit 09f2ea7

File tree

1 file changed

+292
-20
lines changed

1 file changed

+292
-20
lines changed

_posts/2024-10-05-Fourier_1.md

Lines changed: 292 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -6,9 +6,7 @@ tags: [analysis, Fourier analysis, Fourier Series]
66
use_math: true
77
publish: false
88
author_profile: false
9-
toc:
10-
enabled: True
11-
level: 2
9+
toc: true
1210
---
1311

1412
약 일년 반 전, 확률 및 통계 스터디를 하는 과정에서 Fourier analysis에 대해 잠시 공부한 적이 있었다.
@@ -143,15 +141,6 @@ $$
143141

144142
We analyze and solve the standard one and simply apply change of variables for the general cases.
145143

146-
### 1.2 Solution to the wave equation
147-
148-
<!-- There are two kinds of solution to the wave equation ; -->
149-
We can get two kinds of solution to the wave equation ;
150-
(1) using traveling waves
151-
(2) using the superposition of standing waves
152-
153-
#### A solution using traveling waves
154-
155144
Consider a physical problem of a string where $0\le x\le \pi$ :
156145

157146
$$
@@ -168,6 +157,15 @@ u(0,t)=u(\pi,t)=0
168157
\tag a
169158
$$
170159

160+
### 1.2 Solution to the wave equation
161+
162+
<!-- There are two kinds of solution to the wave equation ; -->
163+
We can get two kinds of solution to the wave equation ;
164+
(1) using traveling waves
165+
(2) using the superposition of standing waves
166+
167+
#### A solution using traveling waves
168+
171169
We claim that (a) has the general solution of the form
172170

173171
$$
@@ -211,11 +209,8 @@ Then
211209

212210
$$
213211
\begin{align*}
214-
u(0,t)
215-
&=\frac12\left[f(t)+f(-t)\right]+\frac12\int_{-t}^tg(y)\,dy\\
216-
&=0\\
217-
u(\pi,t)&=\frac12\left[f(\pi+t)+f(\pi-t)\right]+\frac12\int_{\pi-t}^{\pi+t}g(y)\,dy\\
218-
&=0.
212+
u(0,t)&=\frac12\left[f(t)+f(-t)\right]+\frac12\int_{-t}^tg(y)\,dy=0\\
213+
u(\pi,t)&=\frac12\left[f(\pi+t)+f(\pi-t)\right]+\frac12\int_{\pi-t}^{\pi+t}g(y)\,dy=0.
219214
\end{align*}
220215
$$
221216

@@ -320,15 +315,255 @@ Note that $C_1'+C_2'=0$ since $f(x)=F(x)+G(x)$.
320315
Therefore,
321316

322317
$$
323-
\begin{align*}
318+
\begin{aligned}
324319
u(x,t)
325320
&=F(x+t)+G(x-t)\\
326321
&=\frac12f(x+t)+\frac12\int_0^{x+t}g(y)\,dy+C_1'
327322
+\frac12f(x-t)-\frac12\int_0^{x-t}g(y)\,dy+C_2'\\
328323
&=\frac12\left[f(x+t)+f(x-t)\right]+\frac12\int_{x-t}^{x+t}g(y)\,dy.
329-
\end{align*}
324+
\end{aligned}
325+
$$
326+
327+
#### A solution using superposition of standing waves
328+
329+
We claim that (a) has the general solution of the form
330+
331+
$$u(x,t)=\sum_{m=1}^\infty\sin mx\left(A_m\cos mt+B_m\sin mt\right)\tag c$$
332+
333+
where
334+
335+
$$
336+
\begin{aligned}
337+
A_n&=\frac2\pi\int_0^\pi f(x)\sin nx\,dx\\
338+
B_n&=\frac2{n\pi}\int_0^\pi g(x)\sin nx\,dx\\
339+
\end{aligned}
340+
$$
341+
342+
343+
Separating variables $x$ and $t$, let
344+
345+
$$u(x,t)=\phi(x)\psi(t).$$
346+
347+
Then, the wave equation becomes
348+
349+
$$\phi(x)\psi''(t)=\phi''(x)\psi(t).$$
350+
351+
This implies, if $\phi(x)\ne0$ and $\psi(t)\ne0$,
352+
353+
$$\frac{\psi''(t)}{\psi(t)}=\frac{\phi''(x)}{\phi(x)}.$$
354+
355+
Since LHS is independent of $x$ and RHS is independent of $t$, it is independent of both of them.
356+
Let this constant be $\lambda$.
357+
Then
358+
359+
$$
360+
\begin{aligned}
361+
\psi''(t)-\lambda\psi(t)&=0\\
362+
\phi''(x)-\lambda\phi(x)&=0
363+
\end{aligned}\tag3
364+
$$
365+
366+
As in the simple harmonic motion in (1), these ODEs has general solutions
367+
368+
$$
369+
\begin{aligned}
370+
\psi(t)&=A\cos mt + B\sin mt\\
371+
\phi(x)&=\tilde A\cos mx + \tilde B\sin mx
372+
\end{aligned}
330373
$$
331374

375+
by 3.6, where $m^2=-\lambda$.
376+
Apply the fourth condition of (a).
377+
Since the string has zero displacement at $x=0$,
378+
379+
$$0=\tilde A\left(A\cos mt + B\sin mt\right)$$
380+
381+
for every $t$.
382+
Thus, $\tilde A=0$.
383+
At $x=\pi$,
384+
385+
$$0=\tilde B\sin m\pi\left(A\cos mt + B\sin mt\right)$$
386+
387+
for every $t$.
388+
Avoiding the trivial case when $\tilde B=0$, $m$ is an integer.
389+
For every integer $m$,
390+
391+
$$u_m(x,t)=\sin mx\left(A\cos mt + B\sin mt\right)$$
392+
393+
is a solution for the wave equation.
394+
395+
<!-- Let
396+
397+
$$u_m(x,t)=\sin mx\left(A_m\cos mt + B_m\sin mt\right)$$ -->
398+
399+
<div class="notice--danger">
400+
<b>Remark </b> <br>
401+
Letting, for example, $A_m=1$, $B_m=0$,
402+
$$u_1(x,t)=\cos t\sin x$$
403+
is called the fundamental tone or the first harmonic.
404+
$$u_2(x,t)=\cos2t\sin2x$$
405+
is called the first overtone or the second harmonic.
406+
$$u_3(x,t)=\cos3t\sin3x$$
407+
is called second overtone or the third harmonic, and so on.
408+
409+
For fixed $t$, $u_m(x,t)=0$ if $x=\frac\pi mk$ for all $k\in\mathbb Z$.
410+
For each $k$, $x=\frac\pi mk$ is called a node, while $x=\frac\pi m\left(k+\frac12\right)$ is called an antinode.
411+
</div>
412+
413+
The wave equation is linear in that any linear combination of solutions for the wave equation solves the equation too.
414+
Thus,
415+
416+
$$u(x,t)=\sum_{m=-\infty}^\infty\sin mx\left(A_m\cos mt+B_m\sin mt\right)$$
417+
418+
also satisfy the wave equation.
419+
Suppose that this solution is a general one (we omit the proof here.)
420+
Since $u_m=0$ for $m=0$ and
421+
422+
$$
423+
\begin{aligned}
424+
u_{-m}(x,t)
425+
&=\sin(-mx)\left(A_m\cos(-mt)+B_m\sin(-mt)\right)\\
426+
&=\sin mx\left((-A_m)\cos mt+B_m\sin mt\right),
427+
\end{aligned}
428+
$$
429+
430+
we can write, alternatively as
431+
432+
$$u(x,t)=\sum_{m=1}^\infty\sin mx\left(A_m\cos mt+B_m\sin mt\right)\tag4$$
433+
434+
for different $A_m$ and $B_m$.
435+
By the initial displacement condition $u(x,0)=f(x)$ and by the initial velocity condition $u_t(x,0)=g(x)$
436+
437+
$$
438+
\begin{aligned}
439+
f(x)&=\sum_{m=1}^\infty A_m\sin mx\\
440+
g(x)&=\sum_{m=1}^\infty mB_m\sin mx.
441+
\end{aligned}
442+
$$
443+
444+
Multiplying $\sin nx$ and integrating over $[0,\pi]$ for $n\ge1$,
445+
446+
$$
447+
\begin{aligned}
448+
\int_0^\pi f(x)\sin nx\,dx
449+
&=\sum_{m=1}^\infty A_m\int_0^\pi\sin mx\sin nx\,dx\\
450+
&=\sum_{m=1}^\infty\frac{A_m}2\int_0^\pi-\cos(m+n)x+\cos(m-n)x\,dx\\
451+
&=\frac{A_n}2\int_0^\pi\,dx\\
452+
&=A_n\times\frac\pi2\\
453+
\int_0^\pi g(x)\sin nx\,dx
454+
&=\sum_{m=1}^\infty mB_m\int_0^\pi\sin mx\sin nx\,dx\\
455+
&=\sum_{m=1}^\infty\frac{mB_m}2\int_0^\pi-\cos(m+n)x+\cos(m-n)x\,dx\\
456+
&=\frac{nB_n}2\int_0^\pi\,dx\\
457+
&=nB_n\times\frac\pi2.
458+
\end{aligned}
459+
$$
460+
461+
Thus, (Fourier sine coeffcient of $f$)
462+
463+
$$
464+
\begin{aligned}
465+
A_n&=\frac2\pi\int_0^\pi f(x)\sin nx\,dx\\
466+
B_n&=\frac2{n\pi}\int_0^\pi g(x)\sin nx\,dx.
467+
\end{aligned}
468+
$$
469+
470+
And the claim is now proved.
471+
472+
Note that this argument assumes that a *reasonable* function $f$ on $[0,\pi]$ can be expressed as a linear combination of sine functions ;
473+
474+
$$
475+
f(x)=\sum_{m=1}^\infty A_m\sin mx\tag{$\ast$}
476+
$$
477+
478+
Suppose further that this class of function can also be expressed as cosine series too ;
479+
480+
$$
481+
g(x)=\sum_{m=1}^\infty A'_m\cos mx
482+
$$
483+
484+
If an odd function $f$ on $[-\pi,\pi]$ is given, the restriction on $[0,\pi]$ can be expressed as $(\ast)$ and $(\ast)$ is also valid for $f$ on $[-\pi,\pi]$ too;
485+
486+
$$
487+
f(-x)=-f(x)=-\sum_{m=1}^\infty A_m\sin mx=\sum_{m=1}^\infty A_m\sin(-mx).
488+
$$
489+
490+
If an even function $g$ on $[-\pi,\pi]$ is given, the restriction on $[0,\pi]$ can be expressed as cosine series.
491+
The equation is valid for $g$ on $[-\pi, \pi]$ too, by the similar fashion.
492+
493+
Now suppose that a function $F$ on $[-\pi, \pi]$ is given.
494+
Since $F$ can be expressed as $f+g$,
495+
496+
$$
497+
\begin{aligned}
498+
F(x)
499+
&=f(x)+g(x)\\
500+
&=\sum_{m=1}^\infty\left(A_m\sin mx + A'_m\cos mx\right).
501+
\end{aligned}
502+
$$
503+
504+
By Euler identity, (we can choose $a_m=\frac{A_m'-A_mi}2$ and $a_{-m}=\frac{A_m'+A_mi}2$)
505+
506+
$$
507+
F(x) = \sum_{m=-\infty}^\infty a_me^{imx}.
508+
$$
509+
510+
Multiplying $e^{-inx}$ and integrating over $[-\pi, \pi]$,
511+
512+
$$
513+
\begin{aligned}
514+
\int_{-\pi}^\pi F(x)e^{-inx}\,dx
515+
&=\sum_{m=-\infty}^\infty a_m\int_{-\pi}^\pi e^{i(m-n)x}\,dx\\
516+
&=a_n\int_{-\pi}^\pi\,dx\\
517+
&=a_n\times2\pi.
518+
\end{aligned}
519+
$$
520+
521+
Thus,
522+
523+
$$
524+
a_n = \frac1{2\pi}\int_{-\pi}^\pi F(x)e^{-inx}\,dx.
525+
$$
526+
527+
Such $a_n$ is called the *Fourier coefficient* of $F$.
528+
529+
### 1.3 Example : The plucked string
530+
531+
Consider the case when the initial string is located as the two line segments
532+
533+
$$
534+
f(x)
535+
=\begin{cases}
536+
\frac{hx}p&(0\le x\le p)\\
537+
\frac{h(\pi-x)}{\pi-p}&(p\le x\le\pi)
538+
\end{cases}
539+
$$
540+
541+
and when there is no initial movement ; $g(x)=0$.
542+
Following (c),
543+
544+
$$
545+
f(x)=\sum_m A_m\sin mx,\quad
546+
A_m=\frac{2h\sin mp}{m^2p(\pi-p)}
547+
$$
548+
549+
by 3.9.
550+
Since $g(x)=0$, $B_m=0$ for all $m$.
551+
Thus (4) now reduces to
552+
553+
$$
554+
\begin{aligned}
555+
u(x,t)
556+
&=\sum_{m=1}^\infty A_m\sin mx\cos mt\\
557+
&=\frac12\sum_{n=1}^\infty A_m\left[\sin m(x+t)+\sin m(x-t)\right]\\
558+
&=\frac{\sum_{n=1}^\infty A_m\sin m(x+t)+\sum_{n=1}^\infty A_m\sin m(x-t)}2\\
559+
&=\frac{f(x+t)+f(x-t)}2
560+
\end{aligned}
561+
$$
562+
563+
The above equation also corresponds to (b) since $g(x)=0$.
564+
Note that the function $f$ in the last expression is the extended version (on $\mathbb R$) of the original function $f$ on $[0,\pi]$.
565+
566+
332567
## 2. The Heat Equation
333568

334569
## 3. Exercises
@@ -972,4 +1207,41 @@ and
9721207
$$
9731208
\lim_{h\to0}\frac{F(x+h)+F(x-h)-2F(x)}{h^2}
9741209
=\lim_{h\to0}\left(F''(x)+\phi(h)+\phi(-h)\right)=F''(x)
975-
$$
1210+
$$
1211+
1212+
### 3.9
1213+
1214+
Consider the case of plucked string ;
1215+
1216+
$$
1217+
f(x)
1218+
=\begin{cases}
1219+
\frac{hx}p&(0\le x\le p)\\
1220+
\frac{h(\pi-x)}{\pi-p}&(p\le x\le\pi)
1221+
\end{cases}
1222+
$$
1223+
1224+
Calculate the Fourier sine coefficient sine coefficient $A_m$.
1225+
For what position of $p$ are the second, fourth, $\cdots$ harmonics missing?
1226+
For what position of $p$ are the third, sixth, $\cdots$ harmonics missing?
1227+
1228+
(proof)
1229+
1230+
$$
1231+
\begin{aligned}
1232+
A_m
1233+
&=\frac2\pi\int_0^\pi f(x)\sin mx\,dx\\
1234+
&=\frac{2h}{\pi p}\int_0^px\sin mx\,dx
1235+
+\frac{2h}{\pi(\pi-p)}\int_p^{\pi}(\pi-x)\sin mx\,dx\\
1236+
&=\frac{2h}{\pi p}\left(-\frac1mp\cos np+\frac1{m^2}\sin mp\right)
1237+
+\frac{2h}{\pi(\pi-p)}\left(\frac1m(\pi-p)\cos mp+\frac1{m^2}\sin mp\right)\\
1238+
&=\frac{2h\sin mp}{m^2p(\pi-p)}
1239+
\end{aligned}
1240+
$$
1241+
1242+
$A_2=0$ iff $p=\frac\pi2k$ and $A_4=0$ iff $p=\frac\pi4k$ for an integer $k$.
1243+
Note also that neither $p=0$ nor $p=\pi$ as long as $h\ne0$.
1244+
So if $p=\frac\pi2$, then $A_2=A_4=\cdots=0$ and the even harmonics are missing.
1245+
1246+
$A_3=0$ iff $p=\frac\pi3k$ and $A_6=0$ iff $p=\frac\pi6k$ for an integer $k$.
1247+
If $p=\frac\pi3$ or $p=\frac23p$, then $A_3=A_6=\cdots=0$ and the harmonics of $3k$ are missing.

0 commit comments

Comments
 (0)