I've come across a PDE where I feel there should be some clever way of solving it, but I can't seem to find it. It has a form like this:

$$\frac{\partial}{\partial t}\left(y+cz\right)=\frac{\partial}{\partial x}\left(-by+acz\right)$$

Here $y$ and $z$ are the dependent variables and a, b, and c are all constants. Ideally, there would exist some other variable $u$ thats a function of $y, z$ such that we can rewrite the equation in an equivalent form like:

$$\frac{\partial}{\partial t}\left(u\right)=\frac{\partial}{\partial x}\left(u\right)$$

which could be solved with method of characteristics, though I can't seem to find such a variable. Assuming there isn't one, can you recommend any other approaches?

Edit: Here's the original two PDE's from which the above expression was obtained:

$$\frac{\partial y}{\partial t}=-a_1\frac{\partial y}{\partial x}-b_1\left(y-z\right)$$ $$\frac{\partial z}{\partial t}=a_2\frac{\partial z}{\partial x}+b_2\left(y-z\right)$$

We took a linear combination of these 2 equations to try eliminate the terms with $\left(y-z\right)$ in them and then tried to find an invariant variable to solve as I was talking about earlier. $a_1, a_2, b_1, b_2$ are all constants.

  • $\begingroup$ How about Fourier transforming it? $\endgroup$ – joriki Jul 18 '18 at 16:59
  • 1
    $\begingroup$ They are two unknowns $y(x,t)$ and $z(x,t)$. But there is only one equation. The problem is undetermined, or something is missing in the wording of the question, isn't it ? $\endgroup$ – JJacquelin Jul 18 '18 at 17:07
  • $\begingroup$ Yeah, this pde actually comes from the linear combination of two other pdes. One for y, and one for z. We ended up doing that combination to eliminate a term that was present in both of them. $\endgroup$ – Leif Ericson Jul 18 '18 at 17:15
  • 1
    $\begingroup$ @Leif Ericson. What are the two PDEs ? Why don't you write the whole problem ? This makes doubtful what you did and what really is the key issue.. $\endgroup$ – JJacquelin Jul 18 '18 at 17:51
  • $\begingroup$ Sorry about the lateness of the reply, I edited the original post to show the PDE's from which the combined expression came from. Thanks again for your help. $\endgroup$ – Leif Ericson Jul 19 '18 at 16:21

A good way to find such solutions is too assume that $$\frac{\partial}{\partial t}\left(y+cz\right)=\frac{\partial}{\partial x}\left(-by+acz\right)=f'(t)g'(x)$$therefore $$y+cz=f(t)g'(x)+h_1(x)\\-by+acz=f'(t)g(x)+h_2(t)$$therefore$$z=\dfrac{1}{ac+bc}(bf(t)g'(x)+f'(t)g(x)+bh_1(x)+h_2(t))\\y=\dfrac{1}{a+b}(af(t)g'(x)-f'(t)g(x)+ah_1(x)-h_2(t))$$


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.