Simpler Self gravity implementation seems to work

So, I gave up on trying to integrate the potential due to particles with that due to the gas to get a better gas potential at coarse fine boundaries. Now, only and are stored in q. Gas that is accreted during a finest level step will not be reflected in the boundary values to the gas potential at the coarse boundaries, until after the coarser grid updates. Fortunately, since the accretion is incremental this should not pose a significant problem. Additionally if the accretion is steady, then the time derivative of the gas potential should include this effect. I have performed a suite of 2D simulations with various levels of refinement and all seem to handle the collapse nicely.

movie​

And the 1 3D TrueLove problem I ran also seems to handle the collapse well. In the absence of accretion or particles, the scheme should be second order in time - although I have yet to verify this.

Also, it was necessary to reduce the softening length to 1 cell, instead of 4 cells, since the particle would form at a cell corner, but the gas would collect in a cell center. Due to the softening the gas cell would exert a greater pull than the particle which would cause the gas to wonder away from the particle.

Attachments (2)

Comments

No comments.