Piecewise constant solution of waves

161 Views Asked by At

Please let me know if my title or notation should be changed to something more precise.

In the study of waves, I am considering eigenfunction expansions of waves on a finite interval $0 \leq x \leq L$ as products

$$y_n(x,t) = C_n \sin(kx)\sin(\omega t)$$ with a dispersion relation, $\omega = f(k)$, and where $C_n$ is the amplitude that is determined from initial conditions $y(x,0) = y_0(x)$.

Suppose $y_0(x)$ is piecewise constant on $[0,L]$, and consider the solution to our equation, $y(x,t)$. Sometimes, there exist future times $t_i >0$ where $y(x,t_i)$ is again piecewise constant, but not equal to $y_0(x)$.

For example, consider the equation on $0 \leq x \leq L$, $$\frac{\partial^2 \Psi}{\partial x^2} = i \frac{1}{c^2} \frac{\partial \Psi}{\partial t}$$ This has dispersion relation $\omega = c^2 k^2$. Playing around with solutions to this equation, one can see that a piecewise constant initial condition will produce piecewise constant profiles at certain later times. For example, if $$y_0(x)= \begin{cases} 0 & x\leq \frac{L}{4} \\ 1 & \frac{L}{4} \leq x\leq \frac{L}{2}\\ 0 & \frac{L}{2} \leq x \leq L \end{cases}$$

we see at later times some of the following solution profiles:

enter image description here

enter image description here

The solutions here are not exactly piecewise constant because I came across this accidentally, so could only approximate the correct times. And, of course, we see Gibbs oscillations, which prevents the solution from ever being truly constant. However, I hope the phenomenon I am describing is clear.

I have several questions here:

(1) Why does this occur??

(2) For what dispersion relations $\omega = f(k)$ does this occur?

(3) Given that $y_0(x)$ is piecewise constant, suppose there exists a time $t_1 > 0$ such that $y(t_1,x)$ is also piecewise constant. What will be the properties of $y(t_1, x)$? Is there a relationship between $\int_0^L y(t_1,x) dx$ and $\int_0^L y_0(x) dx$?

(4) Given that $y_0(x)$ is piecewise constant, suppose there exists a time $t_1 > 0$ such that $y(t_1,x)$ is also piecewise constant. Does this guarantee that there will exist infinitely many further times $t_2 < t_3 < ...$ with $y(x,t_i)$ piecewise constant? If so, what are these times?

I appreciate any and all help on this question!