수학 (기타)2017. 12. 12. 12:45

Theorem. $\pi$ is irrational.


Assume that $\pi$ is rational. Then $\pi = q/p$ for some positive integer $p, q$. Let $n\in\mathbb{N}$ and define a polynomial $f_n(x)$ by

$$f_n(x) = \frac{p^{n}x^{n}(\pi-x)^{n}}{n!} = \frac{x^{n}(q-px)^{n}}{n!}.$$

Since the polynomial $x(q-px)$ has a maximum value at the point $x = \frac{q}{2p} = \frac{\pi}{2}$, $f_n(x) \leq f_n\left(\frac{\pi}{2}\right)$ for any $0\leq x\leq \pi$. Let $M_n$ be the maximum value of $f_n(x)$. Then

$$M_n = f_n\left(\frac{\pi}{2}\right) = \frac{p^{n}\pi^{2n}}{n!2^{2n}}.$$

Since

$$\lim_{n\rightarrow\infty}\frac{a^n}{n!} = 0$$

for any positive number $a$, we obtain $\lim_{n\rightarrow\infty}M_n = 0$. 

Define $I_n$ by

$$I_n = \int_0^\pi f_n(x)\sin xdx.$$

Since $f_n(x)>0$ and $\sin x > 0$ for any $x\in(0,\pi)$, we have $I_n>0$. Since $f_n(x)\leq M_n$ for any $n\in\mathbb{N}$ and $0\leq x\pi$, 

$$I_n \leq \int_0^\pi M_n\sin xdx = 2M_n.$$

Thus $0<I_n\leq 2M_n$ for any $n\in\mathbb{N}$. Since $M_n\rightarrow 0$ as $n\rightarrow\infty$, there is $N\in\mathbb{N}$ such that $M_n < 1/4$ for all $n\geq N$. Thus $0<I_n<1$ for any $n\geq N$.

On the other hand, Note that $f^{(k)}(x) = 0$ if $k > 2n$. Then

$$I_n = \int_0^\pi f_n(x)\sin xdx = \left[-f_n(x)\cos x + f'_n(x)\sin x - \cdots \pm f_n^{(2n)}(x)\cos x\right]_0^\pi.$$

Note that $f^{(k)}(0) = f^{(k)}(\pi) = 0$ if $0\leq k< n$, by the definition of the polynomial $f_n(x)$. If $n\leq k< 2n-1$, then

$$\begin{eqnarray}
f_n^{(k)}(x) & = & p^{n}n(n-1)\cdots(n-(k-n))\left[(-1)^{k-n}(\pi-x)^{n-(k-n)} + (-1)^nx^{n-(k-n)}\right] \\
& & +\sum_{i=k-(n-1)}^{n-1}c_ix^{n-i}(\pi-x)^{n-(k-i)}
\end{eqnarray}$$

for some numbers $c_i$. Thus

$$\begin{eqnarray}
f_n^{(k)}(0) & = & n(n-1)\cdots(2n-k)(-1)^{k-n}p^{k-n}q^{2n-k} \\
f_n^{(k)}(\pi) & = & n(n-1)\cdots(2n-k)(-1)^np^{k-n}q^{2n-k}
\end{eqnarray}$$

if $n\leq k<n-1$. Note that

$$f_n^{(2n-1)}(x) = p^nn!\left[(-1)^{n-1}(\pi-x) - x\right]$$

and so $f_n^{(2n-1)}(0) = n!(-1)^{n-1}p^{n-1}q$ and $f_n^{(2n-1)}(\pi) = n!p^{n-1}q^n$. Note also that

$$f^{(2n)}_n(x) = (-1)^np^{n}n!.$$

Thus $f^{(k)}(\pi)$ and $f^{(k)}(0)$ are integer for any $0\leq k\leq 2n$. Since

$$\begin{eqnarray}
I_n & = & \left[-f_n(x)\cos x + f'_n(x)\sin x - \cdots \pm f_n^{(2n)}(x)\cos x\right]_0^\pi \\
& = & \sum_{n\leq k\leq 2n, \ \textrm{$k$ is even}}(-1)^{k/2+1}\left[f_n^{(k)}(\pi)\cos\pi - f_n^{(k)}(0)\cos 0\right] \\
& = & \sum_{n\leq k\leq 2n, \ \textrm{$k$ is even}}(-1)^{k/2}\left[f_n^{(k)}(0) + f_n^{(k)}(\pi)\right],
\end{eqnarray}$$

$I_n$ must be an integer. But it contradicts that $0<I_n<1$ if $n\geq N$. $\Box$



이 글을 네이버 블로그에서 보기 (링크)


매우 오래 전에 쓴 글이라 어디서 참고를 한건지 모르겠네요... 아마 Wikipedia를 참고했을 가능성이 높습니다.


Wikipedia에 다양한 증명들이 있으니 참고해보셔도 될 것 같습니다 (링크).




Posted by CAMERAMAN
수학 (기타)2017. 12. 8. 21:05


:: Laplace Transform Table ::


Function

Laplace Transform

Remarks

$$f(x)$$

$$\mathcal{L}\{f\}(s) = \int_0^\infty f(x)e^{-sx}dx$$

Definition

$$af(x) +bg(x)$$

$$a\mathcal{L}\{f\}(s) + b\mathcal\{g\}(s)$$

Linearity

$$xf(x)$$

$$-\left(\mathcal{L}\{f\}\right)'(s)$$


$$x^nf(x)$$

$$(-1)^n\left(\mathcal{L}\{f\}\right)^{(n)}(s)$$

 

$$f'(x)$$

$$s\mathcal{L}\{f\}(s) - f(0)$$

 

$$f^{(n)}(x)$$

$$s^n\mathcal{L}\{f\}(s) - \sum_{k=1}^ns^{n-k}f^{(k-1)}(0)$$

 

$$\frac{f(x)}{x}$$

$$\int_s^\infty \mathcal{L}\{f\}(\sigma)d\sigma$$


$$\int_0^xf(t)dt$$

$$\frac{\mathcal{L}\{f\}(s)}{s}$$

 

$$e^{ax}f(x)$$

$$\mathcal{L}\{f\}(s-a)$$

 

$$f(x-a)u(x-a)$$

$$e^{-as}\mathcal{L}\{f\}(s)$$

 

$$f(ax)$$

$$\frac{1}{a}\mathcal{L}\{f\}(s)$$

 

$$(f*g)(x)$$

$$\mathcal{L}\{f\}(s)\mathcal{L}\{g\}(s)$$

 


Function

Laplace Transform

Remarks

$$f(x)$$

$$\mathcal{L}\{f\}(s) = \int_0^\infty f(x)e^{-sx}dx$$

Definition

$$\delta(x)$$

$$1$$

$\delta$ : Dirac Delta Function

$$u(x)$$

$$\frac{1}{s}$$

$u(x)$ : Step Function

$$\sqrt[n]{x}u(x)$$

$$\frac{1}{s^{\frac{1}{n}+1}}\Gamma\left(\frac{1}{n}+1\right)$$

 

$$\sin(a x)u(x)$$

$$\frac{a}{s^2+a^2}$$

 

$$\cos(ax)u(x)$$

$$\frac{s}{s^2+a^2}$$

 

$$\sinh(ax)u(x)$$

$$\frac{a}{s^2-a^2}$$

 

$$\cosh(ax)u(x)$$

$$\frac{s}{s^2-a^2}$$

 



Wikipedia의 Laplace Transform 항목을 참고하였습니다.





Posted by CAMERAMAN
수학 (기타)2017. 12. 6. 12:49

Fourier Transform Table

Function

Fourier Transform

Remarks

$$f(x)$$

$$\widehat{f}(\omega) = \int_{-\infty}^\infty f(x)e^{-i\omega x}dx$$

Definition

$$af(x) + bg(x)$$

$$a\widehat{f}(\omega) + b\widehat{g}(\omega)$$

Linearity

$$f(x-a)$$

$$e^{-ia\omega}\widehat{f}(\omega)$$

Shift in Time Domain

$$f(x)e^{iax}$$

$$\widehat{f}(x-a)$$

Shift in Frequency Domain

$$f(ax)$$

$$\frac{1}{|a|}\widehat{f}\left(\frac{\omega}{a}\right)$$

 

$$\widehat{f}(x)$$

$$\frac{1}{2\pi}f(-\omega)$$

Duality

$$f'(x)$$

$$i\omega\widehat{f}(\omega)$$

 

$$f^{(n)}(x)$$

$$(i\omega)^n\widehat{f}(\omega)$$

 

$$xf(x)$$

$$i\frac{d\widehat{f}}{d\omega}(\omega)$$

 

$$x^nf(x)$$

$$i^n\frac{d^n\widehat{f}}{d\omega^n}(\omega)$$

 

$$(f*g)(x)$$

$$\widehat{f}(\omega)\widehat{g}(\omega)$$

Convolution

$$f(x)g(x)$$

$$\left(\widehat{f}*\widehat{g}\right)(\omega)$$

 

$$\overline{f(x)}$$

$$\overline{\widehat{f}(-\omega)}$$

 

$$f(x)\cos(ax)$$

$$\frac{\widehat{f}(\omega-a) + \widehat{f}(\omega+a)}{2}$$

 

$$f(x)\sin(ax)$$

$$\frac{\widehat{f}(\omega-a) - \widehat{f}(\omega+a)}{2i}$$

 

$$\int_{-\infty}^xf(t)dt, \ \textrm{where} \ \int_{-\infty}^\infty f(x)dx = 0$$

$$\frac{\widehat{f}(\omega)}{i\omega}$$

 

$$\int_{-\infty}^xf(t)dt$$

$$\frac{\widehat{f}(\omega)}{i\omega} + \pi\widehat{f}(0)\delta(\omega)$$


※ 단, $\delta$는 Dirac delta function.


Function

Fourier Transform

Remarks

$$f(x)$$

$$\widehat{f}(\omega) = \int_{-\infty}^\infty f(x)e^{-i\omega x}dx$$

Definition

$$\mathrm{rect}(x)$$

$$\frac{\sin(\omega/2)}{\omega/2}$$

$$\mathrm{rect}(x) = \left\{\begin{array}{ll}
1 & \textrm{if $|x|\leq 1/2$} \\
0 & \textrm{otherwise}
\end{array}\right.$$

$e^{-ax}u(x)$, $a>0$

$$\frac{1}{a+i\omega}$$

$$u(x) = \left\{\begin{array}{ll}
1 & \textrm{if $x>0$} \\
0 & \textrm{otherwise}
\end{array}\right.$$

$$e^{-x^2}$$

$$\sqrt{\pi}e^{-\omega^2/4}$$

 

$$e^{-|x|}$$

$$\frac{2}{1+\omega^2}$$

 

$$1$$

$$2\pi\delta(\omega)$$


$$\delta(x)$$

$$1$$

 

$$\cos(x)$$

$$\pi\left[\delta(\omega-1) + \delta(\omega + 1)\right]$$

 

$$\sin(x)$$

$$-i\pi\left[\delta(\omega-1) - \delta(\omega+1)\right]$$

 

$$u(x)$$

$$\frac{1}{i\omega} + \pi\delta(\omega)$$

$$u(x) = \left\{\begin{array}{ll}
1 & \textrm{if $x>0$} \\
0 & \textrm{otherwise}
\end{array}\right.$$

$$\mathrm{sgn}(x)$$

$$\frac{2}{i\omega}$$

$$\mathrm{sgn}(x) = \left\{\begin{array}{ll}
1 & \textrm{if $x>0$} \\
-1 & \textrm{otherwise}
\end{array}\right.$$

※ 단, $\delta$는 Dirac delta function.


Wikipedia의 Fourier Transform 항목을 참고하였습니다.


네이버 블로그에서 보기 (링크)

Posted by CAMERAMAN
수학 (기타)2017. 12. 5. 21:02




:: Preliminaries ::


Definition. 함수 $f:\mathbb{R}\rightarrow\mathbb{R}$가 $\mathbb{R}$에서 적분가능하고 조각마다 연속(piecewisely continuous)일 때, $f$의 푸리에 변환(Fourier transform of $f$)을 기호로 $\widehat{f}$ 또는 $\mathfrak{F}(f)$으로 쓰고, 다음과 같은 함수로 정의한다.

$$\widehat{f}:\mathbb{R}\rightarrow\mathbb{R}, \quad \widehat{f}(\omega) = \int_{-\infty}^\infty f(x)e^{-i\omega x}dx$$


Theorem. 함수 $f,g:\mathbb{R}\rightarrow\mathbb{R}$가 $\mathbb{R}$에서 적분가능하고 조각마다 연속일 때, 다음이 성립한다.

(1) 임의의 실수 $a,b\in\mathbb{R}$에 대하여 $\mathfrak{F}(af+bg) = a\mathfrak{F}(f) + b\mathfrak{F}(g)$이다. 

(2) $\widehat{f'} = i\omega\widehat{f}$.

(3) $\widehat{f*g} = \widehat{f}\widehat{g}$.

(4) $a\in\mathbb{R}$에 대하여 함수 $f_a:\mathbb{R}\rightarrow\mathbb{R}$를 $f_a(x) = f(x+a)$으로 정의하자. 그러면 $\widehat{f_a}(\omega) = e^{ia\omega}\widehat{f}(\omega)$이다.

(5) 함수 $f$가 $x\in\mathbb{R}$에서 연속이면 다음이 성립한다.

$$f(x) = \frac{1}{2\pi}\int_{-\infty}^\infty \widehat{f}(\omega)e^{i\omega x}d\omega$$

만약 $f$가 $x\in\mathbb{R}$에서 불연속이면 다음이 성립한다.

$$\frac{f(x+0) + f(x-0)}{2} = \frac{1}{2\pi}\int_{-\infty}^\infty\widehat{f}(\omega)e^{i\omega x}d\omega$$

(6) $\widehat{f} = \widehat{g}$이면 $f = g$이다.



:: Solution ::


함수 $f_0, h$를

$$f_0(x) = \left\{\begin{array}{ll} 1-|x| & \textrm{if $|x|\leq 1$} \\ 0 & \textrm{otherwise}\end{array}\right.,$$

$$ h(x) = \left\{\begin{array}{ll} \cos(\pi x) & \textrm{if $k\leq x\leq k+8$} \\ 0 & \textrm{otherwise}\end{array}\right.$$

으로 정의하고, 함수 $H,K$를 다음과 같이 정의하자.

$$H(x) = \int_{-\infty}^xh(t)dt, \quad K(x) = \int_{-\infty}^xH(t)dt$$

그러면 임의의 $t\in\mathbb{R}$에 대하여

$$g(t) = \int_k^{k+8}f(x)\cos(\pi x)dx = \int_{-\infty}^\infty h(x)f_0(x-t)dx = (h*f_0)(t)$$

이다. 이때

$$\mathrm{Rect}(x) = \left\{\begin{array}{ll} 1 & \textrm{if $|x|\leq 1/2$} \\ 0 & \textrm{otherwise}\end{array}\right.$$

라고 하자. 그러면 $f_0 = \mathrm{Rect}*\mathrm{Rect}$이므로 $g = h*\mathrm{Rect}*\mathrm{Rect}$이다. 이때 함수 $g_0$를 $g_0 = h*\mathrm{Rect}$으로 정의하자. 그러면

$$\begin{eqnarray}\widehat{\mathrm{Rect}}(\omega) & = & \int_{-\infty}^\infty\mathrm{Rect}(x)e^{-i\omega x}dx = \int_{-1/2}^{1/2}e^{-i\omega x}dx \\& = & \frac{1}{-i\omega}\left[e^{-i\omega x}\right]^{1/2}_{-1/2} = \frac{e^{i\omega/2} - e^{-i\omega/2}}{i\omega}\end{eqnarray}$$

이므로

$$i\omega\widehat{g_0}(\omega) = i\omega\widehat{\mathrm{Rect}}(\omega)\widehat{h}(\omega) = e^{i\omega/2}\widehat{h}(\omega) - e^{-i\omega/2}\widehat{h}(\omega)$$

이다. 따라서

$$g_0'(t) = h\left(t+\frac{1}{2}\right) - h\left(t-\frac{1}{2}\right)$$

이다. 이때 임의의 실수 $t \leq k-1/2$에 대하여 $g_0(t) = 0$이므로

$$g_0(t) = \int_{-\infty}^tg_0'(\tau)d\tau = H\left(t+\frac{1}{2}\right) - H\left(t-\frac{1}{2}\right)$$

다. 그리고 $g = g_0*\mathrm{Rect}$이므로

$$i\omega\widehat{g}(\omega) = e^{i\omega/2}\widehat{g_0}(\omega) - e^{-i\omega/2}\widehat{g_0}(\omega)$$

이다. 따라서

$$g'(t) = g_0\left(t+\frac{1}{2}\right) - g_0\left(t-\frac{1}{2}\right) = H(t+1) + H(t-1) - 2H(t)$$

이다. 임의의 실수 $t\leq k-1$에 대하여 $g(t) = 0$이므로,

$$\begin{eqnarray}g(t) & = & \int_{-\infty}^tg'(\tau)d\tau = \int_{-\infty}^t\left[H(\tau+1) + H(\tau-1) - 2H(\tau)\right]d\tau \\ & = & K(t+1) + K(t-1) - 2K(t)\end{eqnarray}$$

이다. 이때

$$H(x) = \int_{-\infty}^xh(t)dt = \left\{\begin{array}{ll}\frac{1}{\pi}\sin(\pi x) & \textrm{if $k\leq x\leq k+8$} \\ 0 & \textrm{otherwise}\end{array}\right.,$$

$$K(x) = \int_{-\infty}^xH(t)dt = \left\{\begin{array}{ll}-\frac{1+\cos(\pi x)}{\pi^2} & \textrm{if $k\leq x\leq k+8$} \\ 0 & \textrm{otherwise}\end{array}\right.$$

임을 이용하면 함수 $g$를 구할 수 있다. 


원본 글 링크





Posted by CAMERAMAN