search

lucasmyers97 / maier-saupe-lc-hydrodynamics

Work at University of Minnesota using finite element methods to simulate hydrodynamics of liquid crystals with a Maier-Saupe field theory free energy
4 stars 0 forks source link

Freezing defects not working on >9 MPI processes #53

Closed lucasmyers97 closed 1 year ago

lucasmyers97 commented 1 year ago

For whatever reason, the scheme in which I freeze the defects is not converging to a solution when I use more than 9 MPI cores. The fact that it works otherwise makes it really hard to investigate. I suspect it has something to do with chopping up the grid so much, but I'm not sure how to diagnose the actual issue. May have to print out the fixed border points.

lucasmyers97 commented 1 year ago

The reason for this was because the internal boundaries which fixed degrees of freedom were not consistent across various processes. I fixed it in this commit d5ccce5 with one of the built in deal.II functions.