$\mathrm{s}$ 1265 2002 209-219 209 DE ( Kenta Kobayashi ), (Hisashi Okamoto) (Research Institute for Mathematical Sciences, Kyoto Univ) (Jinghui Zhu) 1 Iiitroductioii (Xiamen Univ) $c$ (Fig 1) Levi-Civita sequation $\frac{d}{ds}(\frac{e^{2hv}}{2}+qe^{hv}\frac{dv}{ds})-pe^{-hv}\sin v=0$ $v(s)$ $H$ : Hilbert transform $p=gl/(2\pi c^{2})$ $q=2\pi T/(mc^{2}L)$ $L$, $g$,, $T$ $m$ Fig 1: Progressive water-wave Nekrasov[l] $q=0$, Levi-Civita s equation $\mu=3p\exp(-3hv(0))$ $v(s)= \frac{1}{37\gamma}\int_{-\pi}^{\pi}\log( \frac{1}{ ^{\underline{y}}\sin((s-t)/\underline{9}) })\frac{\mu\sin v(t)}{1+\mu/_{0}^{\backslash t}\sin v(w)dw}dt,$ $s\in[-\pi, \pi]$, (11) $\{$ $v(-s)=-v(s)$, $v(s+2\pi)=v(s)$, $v(s)\geq 0$, $s\in[0, \pi]$ Fig 2 $\mu$ $s=0$
210 Fig $\underline{9}_{:}$ The shape of the numerical solution when $\mu=40$ Chandler and Graham [4] $[0, \pi]$ 3,,, 2DE,, DE $\arctan$ $\mu$,, Chandler and Graham,, $10^{-4}\sim 10^{-5}$ 3DE FFT,,, DE, FFT,, DE ( ) (31) $v(s)= \frac{1}{6\pi}\int_{0}^{2\pi}\frac{g(t)-c_{7}(s)}{\tan((t-\mathrm{l}\backslash ^{\neg})/\underline{)})}dt $ $s\in[0,2\pi]$ $G( \theta)=\log(\mu^{-1}+\int_{0}^{\theta}\sin v(w)dw)$
$\backslash$ 211 $\Gamma v(\overline{s})$ (31) $v^{(0)}(s)= \frac{\pi-s}{6}$, $v^{(p+1)}(s)=\gamma v^{(p)}(s)$, $s\in[0,2\pi]$ DE $\varphi(y)$ $\varphi(y)=\pi\sum_{n=-(\infty}$ $\tanh \cdot\backslash \cdot\sinh(y+2n)l -\alpha(n)$ $( \backslash 2 \backslash \vee \cdot )$ 1 $( \mathrm{z}>0)$, ( ) $=\{$ -1 $(n\leq 0)$ $\varphi(y)$ Fig 3, DE, $G(t)$ FFT Fourier $n=\underline{9}m$ $yj,$ $tj,$ $wj$ $y_{j}= \frac{j}{m }$ $t_{j}=\varphi(y_{j})$, $w_{j}=\varphi (yj)$, $j=-m,$ $\cdots,$ $m$, $v^{(p)}(tj)$ $v_{j}^{(p)}$ $v_{j}^{(0)},$ $v_{j}^{(1)},$ $v_{j}^{(2)},$ $\cdots$ $v_{-m}^{(0)}=v_{m}^{(0)}=0$ $v_{j}$ (0) $\pi-t_{j}$ $=\overline{6}$ $-m+1\leq j\leq m-1$ $a_{k}^{(p)}= \frac{1}{n}\sum^{m-1}\sin v_{j}^{(p)}j=-m+1^{\cdot}wj\exp(-i\frac{kj\pi}{m})$, $-m\leq k\leq m-1$ $b_{j}^{(p)}= \sum_{k=-m+1,k\neq 0}^{m-1}\frac{a_{k}^{(p)}}{ik\pi}\exp(i\frac{kj\pi}{m})$, $-7n\leq j\leq m-1$ $c_{j}^{(p)}=b_{j}^{(p)}-b_{-m}^{(p)}$, $G_{j}^{(p)}=\log(\mu^{-1}+c_{j}^{(p)})$, $-rn+1\leq j\leq m-1$ $v_{q}^{(p+1)}= \frac{1}{6m\pi}\{\sum_{j=-m+1,j\neq q}^{m-1}\frac{g_{j}^{(p)}-g_{q}^{(p1}}{\tan((t_{j}-t_{q})/\underline{?})}w_{j}+2\frac{\sin v_{q}^{(p)}}{\mu^{-\mathrm{l}}+c_{q}^{(p)}}w_{q}\}$, $-,n+1\leq q\leq-1$ $v_{q}^{(p+1)}=-v_{-q}^{(p+1)}$, $1\leq q\leq m-1$ $v_{-m}^{(p+1)}=v_{m}^{(p+1)}=0$ $a_{k}^{(p)},$ $b_{j}^{(p)}$ FFT FFT
212 Fig $\cdot$3: The shape of $\varphi(y)\iota \mathrm{v}\mathrm{h}\mathrm{e}\mathrm{n}l=32$ Fig 4: The relative error between two successive steps when $n=256$ and $l=32$ 4 (Nekrasov s equation) $E=\mathrm{m}\mathrm{a}\mathrm{x}j v_{j}^{(p+1)}-v_{j}^{(p)} $, $n=256,\mathit{1}=32,$ $\mu=10^{5}$ Fig4 $l,$ $n,$ $\mu$, 24\sim 33 026 $l=32$ Table 1
$\mathrm{f}\mathrm{i}\mathrm{g}5$ $\mathrm{f}\mathrm{i}\mathrm{g}7$ 213 Fig 5: Relation between $\mu$ and the relative error when $n=256$ and $l=32$ $\mu$ $\mu$, $\mu=10^{10}$ $l$ $\mathrm{f}\mathrm{i}\mathrm{g}6$ $l$ $n$, $l=32$ ( $n=128$ ) 0 $\log(s)=-40$, $n=512$ $n$ $s$, Nekrasov s equation Table 1: Numerical results of Nekrasov s eq
214 Fig 6: Relation between $l$ and the relative error
215 Fig 7: The relative error between different $n$ when $l=32$
216 5Yalnada s equation HYamada[3], solitary wave $\frac{d}{ds}(\frac{e^{2hv}}{\underline{7}}+q \cos(s/\underline{)}) e^{hv}\frac{dv}{ds})-p\frac{e^{-hv}\sin v}{ \cos(s^{\tau}/\underline{)}) }=0$ Yamada s equation $q=0$ $q=0$, Nekrasov s equation $v(s)= \frac{1}{6\pi}\int_{0}^{2\pi}\frac{\sin(t-\grave{\mathrm{s}})}{1-\cos(t-s)}\{g(t)-g(s)\}clt$, $s\in[0,2\pi]$ $G( \theta)=\log(\mu^{-1}+\int_{0}^{\theta}\frac{\sin v(w)}{ \cos(w/\underline{?}) }dw)$ $\Gamma v(s)$ $v^{(0\}}(s)$ $= \frac{\pi-s}{6}$, $v^{(p+1)}(s)=\gamma v^{(p)}(s)$, $s\in[0,2\pi]$ Fig 8 $s=\pi$ Nekrasov s equation DE $\varphi(y)=\pi\{\frac{\exp((\pi/2)\sinh(2y-1)l_{1})}{\exp((\pi/2)\sin \mathrm{h}(2y-1)l_{1})+\exp(-(\pi/2)\sinh(2y-1)l_{2})}$ $- \frac{\exp(-(\pi/2)\sin \mathrm{h}(2y+1)l_{1})}{\exp((\pi/2)\sinh(2y+1)l_{2})+\exp(-(\pi/2)\sinh(2y+1)l_{1})}\}$ $l_{1}$ $\varphi(y)$ Fig 9, 12 $\pm\pi$ 6 (Yamada s equation) Nekrasov s equation, FFT $l=32$, Fig 10, Table 2, Fig ( $n=128$ ), Nekrasov s equation, $n=1024$ $10^{-11}$,
217 Fig 8: The shape of the numerical solufion when $\mu=40$ Fig 9: The shape of $\varphi(y)$ when $l_{1}=l_{2}=32$ 7 Nekrasov s equation Yamada s equation, DE FFT,,
218 Fig 10: The relative error between two successive steps when $n=256$ and $l_{1}=l_{2}=32$ Table $\underline{9}_{:}$ Numerical results of Yamada s eq [1] AINekrasov, On waves of permanent type $\mathrm{i}$ $3(1921),$ pp52-65, Izv IvanovO-Voznesensk Polit Inst, [2] HTakahasi and MMori, Double exponential formulas for numerical integration, Publ ${\rm Res}$ Inst Math Soc, $9(1974)$, pp721-741 [3] HYamada, On the highest solitary wave, Report ${\rm $5(1957),$ pp53-67 Res}$ Inst Appl Mech, Kyushu Univ, [4] $\mathrm{g}\mathrm{a}$ Chandler and IGGraham, The computation of water waves modelled by Nekrasov s equation, SIAM J Numer Anal, 30(1993), pp1041-1065
219 Fig 11: The relative error between different $n$ when $l_{1}=l_{2}=32$