abarret / fluid-solver Goto Github PK

The boundary condition handling is, in my opinion, sub-par. Periodic boundaries need to be reworked and physical boundary conditions need to be added.

Note that for side centered velocities, in the periodic domain case, only one side of the boundary in each direction is a true degree of freedom. Currently, the data is just removed when a solver is used. There should be a better way to handle that.

Need an efficient multigrid preconditioner for pcg for the side centered velocity.

There's a bug in the advection code causing first order accuracy. This bug should be found and terminated as an example to others as soon as possible.