On 1/18/19 12:08 PM, [email protected] wrote: > > When I apply periodic boundary conditions only on velocities at all the four > sides of my unit square domain, then the rank of B (mxn) becomes rank(B) = > n-1, which I think is as you say due to the fact that the pressure is up to a > constant . However, when I try to impose periodic boundary conditions for > both > velocities AND pressure, then the rank of B becomes much less than the number > of its columns and there comes my problem with inverting the B^T (diag > M)^{-1} B. > > Any suggestions how to overcome this problem? I've tried to release some > nodes > on the boundaries (i.e. have periodic b.c. on all the sides except from 4 > nodes - 1 free node per side) but still not working.
Magda -- you are asking mathematical questions, not ones that are implementation details. Can you experimentally determine whether the rank of the matrix is, for example, n-4 or some such? Also, what would you expect from mathematical theory what the rank is or should be? Have you found statements in the literature that address this issue? Best W. -- ------------------------------------------------------------------------ Wolfgang Bangerth email: [email protected] www: http://www.math.colostate.edu/~bangerth/ -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
