I have a question about the following paper:


My question is on the end of the second page, when the author solves the transport equation.

"Now let $S$ be the integral of surface formed from a union of these characteristic curves. In doing so, we see that $z(x,t)$ is constant along the lines $x - at = x_{0}$. That is, $z(x,t) = f(x - at)$. Letting $u(x,t) = z(x,t) = f(x - at)$ for any (smooth) function $f$..."

I don't understand why

"$z(x,t)$ is constant along the lines $x - at = x_{0}$" implies "$z(x,t) = f(x - at)$".

Maybe it's a dumb question, but could anyone help me?


The intuition is as follows: $z(x,t)$ is constant along the lines $x-at=x_0$, so $z$ only "sees" which of these lines you are on. This means that we can write it as $z(x,t)=f(x-at)$ in a well defined way. Another common example in pde is functions from $u:\mathbb{R}^2\to \mathbb{R}$ invariant under rotations are those that may be written as $$ u(r,\theta)=f(r) $$ for a function $f$ depending only on the radius, the only thing the original function $u$ cared about.

To see this more algebraically, note that for $(x,t)$ satisfying $x-at=x_0$, we have $$ z(x,t)=z(x_0,0) $$ by plugging in $t=0$. So, $$ z(x,t)=z(x-at,0) $$ Name $$ f(x-at)=z(x-at,0)=z(x,t) $$ This process works for any given initial position $x_0$ and has the nice visualization of tracing back along each of these lines $x-at=x_0$ in the $(x,t)$ plane to find the value of $z$ at a point $(x_0,0)$ where a value has been prescribed by initial data.

  • 1
    $\begingroup$ Great explanation! $\endgroup$ – Greg Oct 6 '18 at 16:54

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.