$Xโ€™ = F(X)$์˜ ํ•ด๊ฐ€ ์กด์žฌํ•˜๊ณ  ์œ ์ผํ•จ์„ ๋ณด์ด๋Š” ๊ณผ์ •.

10 minute read

๋ณต์ˆ˜์ „๊ณตํ•˜๊ณ  ์žˆ๋Š” ์ˆ˜ํ•™๊ณผ์˜ ์กธ์—…์‹œํ—˜์„ ์œ„ํ•ด ํ•™๋ถ€ ์ˆ˜ํ•™ ๊ณผ๋ชฉ๋“ค์„ ๋‹ค์‹œ ๊ณต๋ถ€ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค๋งŒโ€ฆ ๋ฏธ๋ถ„๋ฐฉ์ •์‹์€ ์กธ์—…์‹œํ—˜ ๋Œ€์ƒ ๊ณผ๋ชฉ์ด ์•„๋‹ˆ๋ผ๋Š” ๊ฑธ ๋‚˜์ค‘์— ์•Œ๊ฒŒ ๋˜์—ˆ์Šต๋‹ˆ๋‹คโ€ฆ OTLโ€ฆ ๊ทธ๋ž˜๋„ ์ด์™• ์‹œ์ž‘ํ•œ ๊ฑฐ ๋‹ค์‹œ ๋ณต์Šต ์ข€ ํ•ด๋ด…์‹œ๋‹ค! ๐Ÿƒ ๋ฏธ๋ถ„๋ฐฉ์ •์‹ ํฌ์ŠคํŠธ ์ „์ฒด ๋ณด๊ธฐ

๋“ค์–ด๊ฐ€๋ฉฐ

๊ฒฝ๊ณ ํ•˜๋Š”๋ฐ ์—ฌ๊ธฐ์„œ๋ถ€ํ„ฐ ์ง„์งœ ์™„์ „ํžˆ ์ƒˆ๋กœ์šด ๋‚ด์šฉ์ž…๋‹ˆ๋‹คโ€ฆ;; ์ง€๊ธˆ๊นŒ์ง€๋Š” ๋ฏธ๋ถ„๋ฐฉ์ •์‹์˜ ์‹ฌํ™” ๋ฒ„์ „์„ ํ•˜๋Š” ๋Š๋‚Œ์ด์—ˆ๋‹ค๋ฉด, ์—ฌ๊ธฐ์„œ๋ถ€ํ„ฐ ์ง„์งœ MATH4xx ๊ณผ๋ชฉ์˜ ์œ„์—„์ด ๋ญ”์ง€ ์ž‘์‚ด๋‚˜๊ฒŒ ๋Š๋‚„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค ใ…‹ใ…‹

์ด ์ฑ•ํ„ฐ์˜ ๋ชฉํ‘œ๋Š” ODE์˜ solution์ด ์กด์žฌ(Existence)ํ•˜๊ณ  ๊ทธ๋ฆฌ๊ณ  ์œ ์ผ(Uniqueness)ํ•˜๋‹ค๋Š” ๊ฒƒ์„ ๋ณด์ด๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค. ๊ทธ๋Ÿฐ๋ฐ ์ €๋Š” ๊ฐ์ž(๐Ÿฅ”)๋‹ˆ๊นŒ ๊ทธ ์ฃผ๋ณ€ ๊ณ๋‹ค๋ฆฌ๋ถ€ํ„ฐ ๋‹ค๊ฐ€๊ฐ€๋ณด๋„๋ก ํ•˜๊ฒ ์Šต๋‹ˆ๋‹ค.

[Existence and Uniqueness์˜ ๊ณ๋‹ค๋ฆฌ๋“ค]

์ˆœ์„œ๋Š” ์ƒ๊ด€์—†์Šต๋‹ˆ๋‹ค.

Picard Iteration

1์ฐจ์› ๋ฏธ๋ถ„๋ฐฉ์ •์‹์˜ ๊ฐ€์žฅ ์ผ๋ฐ˜์ ์ธ ํ˜•ํƒœ๋Š” ์•„๋ž˜์™€ ๊ฐ™์Šต๋‹ˆ๋‹ค.

\[x' = f(x)\]

์ด๋–„, $f(x)$๋Š” ์ž„์˜์˜ ํ•จ์ˆ˜์ด๊ณ , $x(0) = x_0$์ž…๋‹ˆ๋‹ค. ์šฐ๋ฆฌ์˜ ๋ชฉํ‘œ๋Š” ์ด๋Ÿฐ ๋ฏธ๋ถ„๋ฐฉ์ •์‹์— ํ•ด๊ฐ€ ์กด์žฌํ•˜๊ณ , ๋˜ ๊ทธ ํ•ด๊ฐ€ ์œ ์ผํ•ด๋ผ๋Š” ๊ฒƒ์„ ๋ณด์ด๊ณ ์ž ํ•ฉ๋‹ˆ๋‹ค.

[์ ๋ถ„๋ฐฉ์ •์‹ Form]

\[x(t) = x_0 + \int_{t_0}^t f(x(s)) \, ds\]

๋ฏธ๋ถ„๋ฐฉ์ •์‹์ด ๋ฐฉ์ •์‹์— ๋„ํ•จ์ˆ˜๊ฐ€ ์žˆ์–ด์„œ ๋ฏธ๋ถ„๋ฐฉ์ •์‹์ด๋‹ˆ ์œ„์˜ ์‹์€ โ€œ์ ๋ถ„โ€์ด ์žˆ์–ด์„œ โ€œ์ ๋ถ„๋ฐฉ์ •์‹โ€๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ์œ„์˜ ์‹์€ ๊ทธ๋ƒฅ $xโ€™ = f(x)$์—์„œ ์–‘๋ณ€์„ ์ ๋ถ„ํ•œ ๊ฒƒ์— ๋ถˆ๊ณผํ•ฉ๋‹ˆ๋‹ค.

์—ฌ๊ธฐ์—์„œ Iterative ํ•œ ๋ฐฉ์‹์œผ๋กœ solution $x(t)$๋ฅผ ๊ตฌํ•  ๊ฒƒ ์ž…๋‹ˆ๋‹ค!! ๋ฐฉ์‹์€ ์˜ˆ์ œ๋ฅผ ํ†ตํ•ด ์„ค๋ช…ํ•˜๊ฒ ์Šต๋‹ˆ๋‹ค.

Example

\[x' = 2t (1 + x), \qquad x(0) = 0\]

์ธ ํ˜•ํƒœ์˜ ๋ฏธ๋ถ„๋ฐฉ์ •์‹์ด ์žˆ๋‹ค๊ณ  ๊ฐ€์ •ํ•ฉ์‹œ๋‹ค. ์ด๊ฒƒ์„ Picard ๋ฐฉ์‹์œผ๋กœ ํ’€์–ด๋ด…์‹œ๋‹ค. ํ’€์ด๋ฅผ ํ•˜๊ธฐ ์ „์— ๊ณต๋ฐ์…๋‹˜์˜ ๊ธ€์ด ์ดํ•ด์— ๋งŽ์€ ๋„์›€์ด ๋˜์—ˆ์Œ์„ ๋ฐํž™๋‹ˆ๋‹ค.

์ผ๋‹จ ์ ๋ถ„ ๋ฐฉ์ •์‹๋ถ€ํ„ฐ ์„ธ์›Œ๋ด…๋‹ˆ๋‹ค. ์ด๋•Œ, ์‹ค์ œ solution $x(t)$๊ณผ ๊ตฌ๋ถ„ํ•˜๊ธฐ ์œ„ํ•ด $\phi(t)$๋ผ๋Š” ํ•จ์ˆ˜๋กœ ํ‘œ๊ธฐ๋ฅผ ๋ฐ”๊พธ์—ˆ์Šต๋‹ˆ๋‹ค.

\[\phi(t) = \int_0^t 2s (1 + \phi (s)) \, ds\]

์ผ๋‹จ ์ดˆ๊ธฐ ์กฐ๊ฑด $x(0) = 0$์„ ์ด์šฉํ•ด ์ดˆ๊ธฐ ํ•จ์ˆ˜๋ฅผ $\phi_0(t) = 0$์œผ๋กœ ์„ค์ •ํ•˜๊ณ  ์ฒซ๋ฒˆ์งธ Iteration์„ ์ˆ˜ํ–‰ํ•ฉ๋‹ˆ๋‹ค.

\[\phi_1(t) = \int_0^t 2s (1 + 0) \, ds = t^2\]

๊ฐ™์€ ๋ฐฉ๋ฒ•์œผ๋กœ $\phi_2(t)$๋ฅผ ๊ตฌํ•ฉ๋‹ˆ๋‹ค.

\[\phi_2(t) = \int_0^t 2s (1 + s^2) \, ds = t^2 + \frac{t^4}{2}\]

๋˜ ๊ตฌํ•ฉ๋‹ˆ๋‹ค.

\[\phi_3(t) = \int_0^t 2s (s^2 + \frac{s^4}{2}) \, ds = t^2 + \frac{t^4}{2} + \frac{t^6}{2\cdot 3}\]

์ด์ œ๋Š” ์กฐ๊ธˆ ํŒจํ„ด์ด ๋‚˜์˜ค๋Š”๋ฐ, ์•”ํŠผ ๋ฐ˜๋ณตํ•˜๋ฉด ์•„๋ž˜์™€ ๊ฐ™์€ ํ˜•ํƒœ๊ฐ€ ๋ฉ๋‹ˆ๋‹ค.

\[\begin{aligned} \phi_n(t) &= t^2 + \frac{t^4}{2} + \frac{t^6}{2\cdot 3} + \frac{t^8}{2\cdot 3 \cdot 4} + \cdots + \frac{t^{2n}}{n!} \\ &= \sum_{k=1}^{n} \frac{t^{2k}}{k!} \end{aligned}\]

์šฐ๋ฆฌ๋Š” $\phi(t) = \lim_{n\rightarrow \infty} \phi_n(t)$๋ฅผ ์ˆ˜ํ–‰ํ•˜์—ฌ ์ด๊ฒƒ์ด ODE์˜ solution์ž„์„ ์ฃผ์žฅํ•˜๊ณ  ์‹ถ๋‹ค. ๊ทธ๋Ÿฐ๋ฐ, ์ด๋ฅผ ์ฃผ์žฅํ•˜๋ ค๋ฉด ์ผ๋‹จ ์ด ๊ธ‰์ˆ˜๊ฐ€ ์ˆ˜๋ ดํ•˜๋Š”์ง€๋ฅผ ๋จผ์ € ๋ณด์—ฌ์•ผ ํ•œ๋‹ค.


[๊ธ‰์ˆ˜์˜ ์ˆ˜๋ ด ํŒ์ •]

๋น„์œจํŒ์ •๋ฒ•์„ ์ˆ˜ํ–‰ํ•œ๋‹ค.

\[\lim_{n\rightarrow \infty} \frac{a_{n+1}}{a_n} = \frac{t^{2n + 2}}{(n+1)!} \cdot \frac{n!}{t^{2n}} = \frac{t^2}{n+1} = 0 < 1\]

๋”ฐ๋ผ์„œ ๊ธ‰์ˆ˜๊ฐ€ ์ˆ˜๋ ดํ•œ๋‹ค. ์ด๋•Œ, ์œ„์˜ ๊ทนํ•œ์‹์€ $t$ ๊ฐ’์— ์ƒ๊ด€์—†์ด ์ˆ˜๋ ดํ•˜๋ฏ€๋กœ Convergence radius๋Š” ์ „์ฒด ์˜์—ญ์ด๋‹ค.


์‚ฌ์‹ค ์š” ๊ธ‰์ˆ˜๋Š” ์ง€์ˆ˜ ํ•จ์ˆ˜์˜ ๊ผด๋กœ ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ๋Š”๋ฐ, $e^t$์ด

\[e^t = 1 + \frac{t}{1} + \frac{t^2}{2!} + \cdots + \frac{t^n}{n!} + \cdots\]

์ด๋ฏ€๋กœ ๊ธ‰์ˆ˜๋ฅผ $\phi(t) = e^{t^2} - 1$๋กœ ์ •๋ฆฌํ•  ์ˆ˜ ์žˆ๋‹ค.


[ํ•ด์˜ ์œ ์ผ์„ฑ ์ฆ๋ช…]

์ฆ๋ช…์„ ์œ„ํ•ด ํ•ด๊ฐ€ ์œ ์ผํ•˜์ง€ ์•Š๊ณ , ๋˜ ๋‹ค๋ฅธ ํ•ด $\psi(t)$๊ฐ€ ์กด์žฌํ•œ๋‹ค๊ณ  ๊ฐ€์ •ํ•œ๋‹ค. ๋‘ ํ•จ์ˆ˜๊ฐ€ ๋‹ค๋ฅธ ํ•จ์ˆ˜์ด๋ฏ€๋กœ ์˜์—ญ ์œ„์˜ ์ ์–ด๋„ ํ•œ ์  $t$์—์„œ $\phi(t) - \psi(t) \ne 0$์ด๋‹ค.

$\psi(t)$๋„ ์ ๋ถ„๋ฐฉ์ •์‹์˜ ํ•ด์ด๋ฏ€๋กœ ์•„๋ž˜์˜ ๋“ฑ์‹์ด ์„ฑ๋ฆฝํ•œ๋‹ค.

\[\psi(t) = \int_0^t 2s(1 + \psi(s))\, ds\]

์ด๋•Œ, $\phi(t) - \psi(t)$์— ๋Œ€ํ•œ ์‹์„ ์„ธ์›Œ ์ด๊ฒƒ์ด ๋ชจ๋“  $t$์— ๋Œ€ํ•ด non-zero์ธ์ง€ ํ™•์ธํ•ด๋ณด์ž.

\[\phi(t) - \psi(t) = \int_0^{t} 2s(\phi(s) - \psi(s))\, ds\]

์œ„์˜ ์ ๋ถ„๋ฐฉ์ •์‹์˜ ์–‘๋ณ€์— ์ ˆ๋Œ“๊ฐ’์„ ์”Œ์šฐ๋ฉด ์•„๋ž˜์™€ ๊ฐ™์€ ๋ถ€๋“ฑ์‹์ด ์„ฑ๋ฆฝํ•œ๋‹ค.

\[\|\phi(t) - \psi(t)\| = \left\|\int_0^{t} 2s(\phi(s) - \psi(s))\, ds\right\| \le \int_0^{t} \| 2s (\phi(s) - \psi(s))\| \, ds\]

์ด๊ฒŒ ์„ฑ๋ฆฝํ•˜๋Š” ์ด์œ ๋Š”, ์–‘์ˆ˜-์Œ์ˆ˜ ๋‘˜๋‹ค ๊ฐ€๋Šฅํ•œ ํ•จ์ˆ˜๋ฅผ ์ ๋ถ„ํ•œ ๋„“์ด๋ณด๋‹ค ์–‘์ˆ˜๋งŒ ๊ฐ€๋Šฅํ•œ ์ ˆ๋Œ“๊ฐ’์„ ์ ๋ถ„ํ•œ ๊ฒƒ์˜ ๋„“์ด๊ฐ€ ๋” ํฌ๊ธฐ ๋•Œ๋ฌธ์ด๋‹ค.

์ ๋ถ„ ๋‚ด๋ถ€์— ์žˆ๋Š” $2s (\phi(s) - \psi(s))$์—์„œ $2s$๋ฅผ ๋ฐ–์œผ๋กœ ๊บผ๋‚ด๋ ค๊ณ  ํ•œ๋‹ค. $0 < t < A$์ธ ์–ด๋–ค ์ƒ์ˆ˜ $A$๋ฅผ ์žก๋Š”๋‹ค๋ฉด, ์•„๋ž˜์˜ ๋ถ€๋“ฑ์‹์ด ๋งŒ์กฑํ•œ๋‹ค.

\[\int_0^{t} \| 2s (\phi(s) - \psi(s))\| \, ds \le 2A \cdot \int_0^{t} \| (\phi(s) - \psi(s))\| \, ds\]

์•ž์œผ๋กœ์˜ ๊ณผ์ •์—์„œ ํ‘œ๊ธฐ์˜ ํŽธ์˜๋ฅผ ์œ„ํ•ด $B = 2A$๋กœ ๋Œ€์ฒดํ•˜์—ฌ ํ‘œ๊ธฐํ•œ๋‹ค. ๋‹ค์‹œ ๋ถ€๋“ฑ์‹์„ ์„ธ์›Œ๋ณด๋ฉด

\[\|\phi(t) - \psi(t)\| \le B \cdot \int_0^{t} \| (\phi(s) - \psi(s))\| \, ds\]

$U(t) = |\phi(t) - \psi(t)|$๋กœ ๋‘๊ณ  ์‹์„ ๋‹ค์‹œ ์“ฐ๋ฉด ์•„๋ž˜์™€ ๊ฐ™์€๋ฐ,

\[U(t) \le B \cdot \int_0^{t} U(s) \, ds\]

$U(0) = 0$๋Š” ์ ๋ถ„๋ฒ”์œ„์— ๋”ฐ๋ฅธ ๊ฒฐ๊ณผ์ด๊ณ , $U(t)$ ํ•จ์ˆ˜๊ฐ€ ์ ˆ๋Œ“๊ฐ’ ํ•จ์ˆ˜์ด๊ธฐ ๋•Œ๋ฌธ์— $U(t) \ge 0$์ด ์„ฑ๋ฆฝํ•œ๋‹ค. ์ด์ œ ๋ฏธ์ ๋ถ„ํ•™์˜ ๊ธฐ๋ณธ์›๋ฆฌ์— ์˜ํ•ด ์–‘๋ณ€์„ ๋ฏธ๋ถ„ํ•˜๋ฉด

\[U'(t) \le B \cdot U(t)\]

๊ฐ€ ๋˜๊ณ , RHS๋ฅผ LHS๋กœ ์˜ฎ๊ธฐ๋ฉด

\[U'(t) - B \cdot U(t) \le 0\]

์ด๋•Œ, ์–‘๋ณ€์— $e^{-Bt}$๋ฅผ ๊ณฑํ•ด์ค€๋‹ค. ์ด๋•Œ, $e^{-Bt}$๋Š” Integrating Factor์ด๋‹ค.

\[(e^{-Bt} U'(t) - B e^{-Bt} U(t)) = (e^{-Bt} \cdot U(t))' \le 0\]

์ด ๋œ๋‹ค!! ๋ถ€๋“ฑ์‹์„ ์ ๋ถ„ํ•˜๋ฉด

\[\begin{aligned} e^{-Bt} \cdot U(t) &\le 0 \\ U(t) &\le 0 \end{aligned}\]

๋ผ๋Š” ๊ฒฐ๊ณผ๊ฐ€ ๋‚˜์˜จ๋‹ค. ์ด๊ฒƒ์€ ์ฒ˜์Œ์— ๊ด€์ฐฐํ•œ $U(t) = |\phi(t) - \psi(t)| \ge 0$์™€ ์ผ์น˜ํ•˜๋ ค๋ฉด $U(t) = 0$์ด ์„ฑ๋ฆฝํ•ด์•ผ ํ•˜๊ณ , ์ด๋Š” $\phi(t) = \psi(t)$๋ผ๋Š” ๊ฒƒ์„ ๋งํ•œ๋‹ค. ์ด๊ฒƒ์€ ์ฒ˜์Œ ๊ฐ€์ •์— ๋ชจ์ˆœ๋˜๋ฏ€๋กœ, $\psi(t)$๋Š” ์กด์žฌํ•˜์ง€ ์•Š๊ณ , $\phi(t)$๊ฐ€ ์œ ์ผํ•œ solution์ด๋‹ค. $\blacksquare$

Picard Iterations: Theorem statement

์˜ˆ์ œ์˜ ํ’€์ด๊ฐ€ ๊ธธ์–ด์กŒ๋Š”๋ฐ, ์•”ํŠผ Picard Iteration์œผ๋กœ solution์„ ์–ป์„ ์ˆ˜ ์žˆ์„๊ฑฐ๋ผ๊ณ  ๋”์šฑ ๋ฏฟ๊ฒŒ ๋˜์—ˆ๋‹ค!! ๐Ÿ™‚ ์ด์ œ๋Š” ์ด ๊ณผ์ •์„ ์—„๋ฐ€ํžˆ ์ •์˜ํ•ด๋ณด์ž.

Given ODE $xโ€™(t) = f(t, x(t))$, and initial value $x(0) = x_0$, we can get a solution by iteration as follows

  • start as $\phi_0(t) = x_0$
  • generate sequence of functions as
\[\phi_{k+1}(t) = x_0 + \int_0^t f(s, \phi_k(s)) \, ds\]

this sequence of functions converges to the solution of the given ODE.

์ฆ๋ช…์€โ€ฆ ํŒจ์Šคํ•œ๋‹ค!! ๋‚˜๋Š” ์ปด๊ณต๊ณผ๋‹ˆ๊นŒ!! ์–ด๋–ป๊ฒŒ ๋ณด๋ฉด ๋ฏธ๋ถ„๋ฐฉ์ •์‹์„ โ€œNumerical Methodโ€๋กœ ํ‘ธ๋Š” ๋ฐฉ์‹์ด๋‹ค. ์ง€๊ธˆ๊นŒ์ง€ ํ–ˆ๋˜ ๋ฏธ๋ฐฉ ํ’€์ด์™€๋Š” ์ ‘๊ทผ ๋ฐฉ์‹์ด ์ข€ ๋‹ฌ๋ผ์„œ ์–ด์ƒ‰ ํ–ˆ๋˜ ๊ฒƒ ๊ฐ™๋‹ค.

Picard Iteration on ODE System

๋ฏธ๋ถ„๋ฐฉ์ •์‹์˜ ๊ฐ€์žฅ ์ผ๋ฐ˜์ ์ธ ํ˜•ํƒœ๋Š” $Xโ€™ = F(X)$ ์ž…๋‹ˆ๋‹ค. ์ด๋•Œ, $F(X)$๋Š” ์ž„์˜์˜ ๋ฒกํ„ฐ ํ•„๋“œ ์ž…๋‹ˆ๋‹ค. ๋‹น์žฅ $Xโ€™ = F(X)$์ธ ์‹œ์Šคํ…œ์„ ํ‘ธ๋ ค๊ณ  ํ•˜๋ฉด ๋จธ๋ฆฌ๊ฐ€ ์•„ํ”„๋‹ˆ๊นŒ ๐Ÿ˜ตโ€๐Ÿ’ซ ์ผ๋‹จ $xโ€™ = f(x)$์ธ 1์ฐจ์›์—์„œ Picard Iteration์œผ๋กœ ๋ฏธ๋ถ„๋ฐฉ์ •์‹์˜ ํ•ด๊ฐ€ ์กด์žฌํ•˜๊ณ , ์œ ์ผํ•œ์ง€๋ฅผ ๋ฐํ˜”์Šต๋‹ˆ๋‹ค.

Examples

์‹ ๊ธฐํ•˜๊ฒŒ๋„ ODE System์—์„œ๋„ Picard Iteration์„ ์“ธ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์•„๋ž˜์™€ ๊ฐ™์€ ODE System์„ Picard๋กœ ํ’€์–ด๋ด…์‹œ๋‹ค.

\[X' = F(X) = \begin{pmatrix} 0 & 1 \\ -1 & 0 \end{pmatrix} X\]

๊ทธ๋ฆฌ๊ณ  Initial Value๋Š” $X(0) = (1, 0)$์ž…๋‹ˆ๋‹ค. ์ง€๊ธˆ๊ฐ€์ง€ 4ํ•™๋…„ ๋ฏธ๋ฐฉ์—์„œ ๋งŽ์ด ๋ณธ ๋…€์„์œผ๋กœ ์‹ค์ œ ์†”๋ฃจ์…˜์€ $X(t) = (\cos t, -\sin t)$๋กœ ๋‚˜์˜ต๋‹ˆ๋‹ค.

Picard Iteration์„ ์ˆ˜ํ–‰ํ•ฉ์‹œ๋‹ค.

\[U_0(t) = \begin{pmatrix} 1 \\ 0 \end{pmatrix}\] \[U_1(t) = \begin{pmatrix} 1 \\ 0 \end{pmatrix} + \int_0^t \begin{pmatrix} 0 & 1 \\ -1 & 0 \end{pmatrix} \begin{pmatrix} 1 \\ 0 \end{pmatrix} ds = \begin{pmatrix} 1 \\ 0 \end{pmatrix} + \int_0^t \begin{pmatrix} 0 \\ -1 \end{pmatrix} ds = \begin{pmatrix} 1 \\ -t \end{pmatrix}\] \[U_2(t) = \begin{pmatrix} 1 \\ 0 \end{pmatrix} + \int_0^t \begin{pmatrix} -s \\ -1 \end{pmatrix} ds = \begin{pmatrix} 1 - t^2/2 \\ - t \end{pmatrix}\] \[U_3(t) = \begin{pmatrix} 1 \\ 0 \end{pmatrix} + \int_0^t \begin{pmatrix} -s \\ -1 + s^2/2 \end{pmatrix} ds = \begin{pmatrix} 1 + -t^2/2 \\ -t + t^3/3! \end{pmatrix}\] \[U_4(t) = \begin{pmatrix} 1 \\ 0 \end{pmatrix} + \int_0^t \begin{pmatrix} -s+s^3/3! \\ -1 + s^2/2 \end{pmatrix} ds = \begin{pmatrix} 1 - t^2/2! + t^4/4! \\ -t + t^3/3! \end{pmatrix}\]

์ด ๊ณผ์ •์„ ๋ฐ˜๋ณตํ•˜๋‹ค๋ณด๋ฉด, $U_n(t)$๋Š” ์•„๋ž˜์™€ ๊ฐ™์ด ๊ฒฐ์ •๋œ๋‹ค.

\[U_n(t) = \begin{pmatrix} 1 - t^2/2! + t^4/4! + \cdots \\ - (t - t^3/3! + t^5/5! - \cdots) \end{pmatrix}\]

ํ•จ์ˆ˜์—ด์„ ๊ทนํ•œ์œผ๋กœ ๋ณด๋‚ด๋ฉด ์•„๋ž˜์™€ ๊ฐ™์ด ์ˆ˜๋ ดํ•œ๋‹ค.

\[U(t) = \lim_{n\rightarrow \infty} U_n(t) = \begin{pmatrix} \cos (t) \\ - \sin (t) \end{pmatrix}\]

Reference