I'm looking for some methods that could be directly applied to the PDE $$ \frac{\partial u}{\partial t} + \frac{(u_x)^2}{2} = 0\tag{*} $$ without converting it by $v = u_x$ to the Hopf equation $$ \frac{\partial v}{\partial t} + \frac{\partial (v^2/2)}{\partial x} = 0. $$

Ideally, a WENO scheme is perfect, but I'm wondering how to apply a limiter or a numerical flux to $(*)$.

Simple centered scheme $$ \dot u_i(t) = -\frac{1}{2}\left(\frac{u_{i+1}(t) - u_{i-1}(t)}{2h}\right)^2 $$ expectedly oscillates when shock (on $v = u_x$) is formed. enter image description here

  • $\begingroup$ You have tagged this with "hyperbolic-pde", but it is really a Hamilton-Jacobi equation. Have you read anything about discretization of Hamilton-Jacobi equations? $\endgroup$ – David Ketcheson Oct 18 '15 at 9:45
  • $\begingroup$ @DavidKetcheson Thank you David, I'm going to read about HJ equation now. $\endgroup$ – uranix Oct 18 '15 at 9:47

This is a Hamilton-Jacobi equation. You can read about how to apply WENO to such equations in Section 4 of Chi-Wang Shu's 2009 WENO review paper, and references therein.


The simplest scheme you could employ is an upwind scheme. It's first-order and introduces artificial viscosity/diffusion, but doesn't require limiters. Probably the next simplest class would be Lax-Wendroff schemes, for which you can find a comprehensive explanation in LeVeque's book on Finite Volume schemes.

  • $\begingroup$ Can you please show me the upwind scheme for my equation? Not for the Hopf form, but for the original one. $\endgroup$ – uranix Oct 14 '15 at 18:54

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.