Whether the flux-derivative damping term can be omitted

Determine whether the flux-derivative contribution in the damping coefficient of Scheme D can be set to zero while retaining convergence to the correct entropy solution for non-convex conservation laws.

Background

The practical damping coefficient DjD_j contains an interior flux-derivative term weighted by ϵ1\epsilon_1 and an interface-jump term weighted by ϵ2\epsilon_2. The flux-derivative term is analytically important in the paper's entropy and compensated-compactness arguments, but numerical experiments suggest that it may sometimes be unnecessary, especially for certain systems and multidimensional tests.

For non-convex scalar examples, however, setting ϵ1=0\epsilon_1=0 can produce incorrect solutions when the interface damping is moderate. The paper therefore leaves unresolved the general conditions under which the flux-derivative term may be omitted.

References

This section discusses the effects of $\epsilon_1$ on numerical results and the question of whether it can be set to zero.