Asymptotic expansion for the solution of linear KDV eq.

167 Views Asked by At

Hi,

The question arises from the book Solitons by P. G. Drazin about the linearized KDV eq. $$ u_t+u_{xxx}=0 $$ My first step was to take a Fourier transform of the equation, find that the dispersion relation was $-\alpha^3$ and then to have a Fourier expansion of the form $$u(x,t)=\frac{1}{2\pi}\int_{-\infty}^\infty e^{i(kx+k^3t)}\int_{-\infty}^\infty e^{-iky}dydk $$ then by playing around with $k=\alpha/(3t)^{1/3}$ I get the following terms

\begin{align} u(x,t)&=\frac{1}{2\pi}\int_{-\infty}^\infty \frac{u_0(y)}{(3t)^{1/3}}dy\int_{-\infty}^\infty e^{i(\alpha\frac{x-y}{(3t)^{1/3}}+\frac{\alpha^3}{3})}d\alpha\\ &=\frac{1}{2\pi}\int_{-\infty}^\infty \frac{u_0(y)}{(3t)^{1/3}}\Bigg[\int_{-\infty}^\infty e^{i(\alpha\frac{x}{(3t)^{1/3}}+\frac{\alpha^3}{3})}d\alpha-\frac{y}{(3t)^{1/3}}\int_{-\infty}^\infty\alpha e^{i(\alpha\frac{x}{(3t)^{1/3}}+\frac{\alpha^3}{3})}d\alpha\Bigg]dy+\dots \end{align} where, to arrive at the second expression, we need to use taylor expansion for $exp(-iky)$. Now, I'm trying to find an asymptotic expansion in the area $x/t^{1/3}\approx O(1)$, for $x>0$ and $x<0$ separately... Drazin's book states that $x<0$ should be "a slowly decaying wave train" and $0<x$ should be "a steeply rising wave front" But i don't know how to arrive at that.

Could anyone lend a hand with the task?

edit:

After awhile trying to find the asymptotics using the asymptotics or $Ai(x)$ I've come to the conclusion that it is not the way since they are based on $\frac{x}{(3t)^{1/3}}\to\infty$ and are so useless to my effort. thus focusing on \begin{align} u(x,t)&=\int_{-\infty}^\infty \frac{u_0(y)}{(3t)^{1/3}}dy\cdot Ai(\frac{x}{(3t)^{1/3}})-\int_{-\infty}^\infty y\frac{u_0(y)}{(3t)^{2/3}}dy\cdot Ai'(\frac{x}{(3t)^{1/3}})+\dots \end{align} I've decided to look for the asymptotics for fixed $v=\frac{x}{(3t)^{1/3}}=o(1)$ which leads to: \begin{align} u(v(3t)^{1/3},t)&=\frac{v}{x}\int_{-\infty}^\infty u_0(y)dy\cdot Ai(v)-\frac{v^2}{x^2}\int_{-\infty}^\infty y\cdot u_0(y)dy\cdot Ai'(v)+\dots \end{align}

does anyone have an idea how to continue?