APIC Solver numerical dissipation? about fluid-engine-dev

[doyubkim]

@aneeshvaidya, thanks for reporting this. It could be a test issue if I see other APIC test results which show low dissipation. Let me take a look.

[aneeshvaidya]

In the Apic Solver 2d rotation test, I commented out the following line:

solver->setPressureSolver(nullptr);

And there was significantly reduced dissipation. Not sure what this means, but I think it may have to do with the pressure projection correcting the velocities, allowing momentum conservation?

[doyubkim]

@aneeshvaidya, you're right. So the intention of that piece of code was to eliminate everything else except APIC transfer (grid <-> particles) and put position constraint (line 91 to 95 of apic_solver2_tests.cpp) to apply rigid body constraint. That position constraint part can cause a projection error, and as you observed, solving pressure step will make the particles to stay in the circular course and thus less projection error. Let me remove that line.