4
$\begingroup$

I'm trying to show that a delta function $\delta(x - ct)$ is a distributional solution of the PDE $$ D_{(0,2)}u(x, t) = c^2 D_{(2,0)}u(x,t). $$ Here $D_{(i,j)}$ means $i$-th partial differentiation on the first variable and $j$-th partial differentiation on the second variable. I'm not strictly sure the "travling delta function" is actually a distributional solution. However, I think it could be intuitively.

The delta function may be expressed as a distribution: $\left< \delta_c, \phi \right> = \int_{-\infty}^\infty \phi(ct, t) dt$. This is motivated by the manipulations of the delta function in physics literatures like $$ \int\int \delta(x - ct) f(x, t) dx dt = \int f(ct, t) dt. $$

Now for the left hand side of the PDE, $$ \left< D_{(0,2)}\delta_c, \phi \right> = \left< \delta_c, D_{(0,2)} \phi \right> = \int_{-\infty}^\infty D_{(0,2)}\phi (ct, t) dt. $$ where $\phi$ is a test function defined on $\mathbb{R}^2$. For the right hand side, $$ c^2 \left< D_{(2,0)}\delta_c, \phi \right> = c^2 \left< \delta_c, D_{(2,0)} \phi \right> = c^2 \int_{-\infty}^\infty D_{(2,0)}\phi (ct, t) dt. $$ I'm stuck here. How can I show the two integrals are the same? Or is the delta function not a solution?

$\endgroup$
2
$\begingroup$

I think I understood the phenomenon :

$$\int_{-\infty}^\infty \phi(ct-h,t) dt = \frac{1}{c} \int_{-\infty}^\infty \phi(x,\frac{x+h}{c}) dx$$

by the variable change $x = ct-h$, so that :

$$\int_{-\infty}^\infty \frac{\partial^2}{\partial x^2} \phi(ct,t) dt = \lim_{h \to 0} \int_{-\infty}^\infty \frac{-\phi(ct-h,t)+2\phi(ct,t)-\phi(ct+h,t)}{h^2} dt$$ $$ = \frac{1}{c}\lim_{h \to 0} \int_{-\infty}^\infty \frac{-\phi(x,\frac{x+h}{c})+2\phi(x,\frac{x}{c})-\phi(x,\frac{x-h}{c})}{h^2} dx = \frac{1}{c^3}\int_{-\infty}^\infty \frac{\partial^2}{\partial t^2} \phi(x,x/c) dx$$ $$= \frac{1}{c^2}\int_{-\infty}^\infty \frac{\partial^2}{\partial t^2} \phi(c\tau,\tau) d\tau$$

as far as $\phi$ is a nice $C^\infty$ function with compact support allowing all these manipulations.

note : this is nothing else than saying $\delta''_{x=t}$ as $\delta_{x=t}$ are symmetric 2D distributions : $\langle \ \delta_{\displaystyle x=t}, \varphi_{\displaystyle x,t}\ \rangle = \langle \ \delta_{\displaystyle x=t}, \varphi_{\displaystyle t,x} \ \rangle$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.