Dohyun Kim

I managed to resolve this issue by replacing `` keys b `` to `` keys ~b ``

> With this terminology in in mind, it makes more sense to me to call the "pre-assembled" option "element assembly" (for the divergence operator) and use `-ea`/`--element-assembly-divergence` (or something like...

> This is more of an overarching suggestion: Would it make more sense to have the Riemann solver be a subset of the FluxFunction class? Besides some of the simpler...

@pazner Thank you for your comments. I think I resolved all the comments. Would you please revise the updated version? Also, it seems that branch-history check fails because I have...

Sounds good to me. I appreciate your comments. I also agree with you on the filename, but I have no better idea. Any suggestions would be appreciated.

Thank you for your approval @tarikdzanic. I think `hyperbolic` is a good candidate. Do you have any other suggestions, @pazner?

@v-dobrev I also just noticed that the current code is slower than the previous. It seems that the main difference is coming from 1) higher order integration (can be controlled...

I made another [branch](https://github.com/mfem/mfem/tree/HCL-preassemble) and added an option to switch between linear/nonlinear weak divergence, see [here](https://github.com/mfem/mfem/compare/HCL-refactor...HCL-preassemble) to compare. If we use the linear weak divergence, then CPU time / iteration...

> Hi, @dohyun-cse, I checked the proposed update in the branch https://github.com/mfem/mfem/tree/HCL-preassemble and it looks like the slowdown is essentially gone in the serial case (`ex18`): I saw at most...

Thank you @v-dobrev! I'm wondering why the Sanitizer suddenly failed after I made some local changes in `hyperbolic.cpp`. The failed tests seem not related to `hyperbolic.cpp`. Should I worry about...