> This is unexpected... I have (what I'm pretty sure is) a (PETSc-wrapper) > identity matrix and I am using PETScWrappers::SolverBicgstab to solve > it.. this may seem unbelievably stupid but it is for an advection > equation that I usually solve semi-implicitly but i want to be able to > solve it explicitly as well. The matrix is an identity (I have checked > this) and the RHS doesn't have any weird values in it. When I attempt to > "solve" the system I get a NoConvergence error. What worries me is if > bicgstab can't invert an identity matrix, how is it inverting one with > non-diagonal entries?
Yes, this would concern me as well :-) > Could this be one of those PETSc-related things where some of the > entries in the matrix are not initialized? But if that were the case > would I not get a different error? PETSc has functions to print a matrix (and the PETScWrappers::SparseMatrix has a conversion operator to get at the PETSc matrix handle). I'd take a look at the matrix entries and make sure that it really is the identity matrix. In particular, make sure that there are no rows that have no nonzero entries in them. Best W. -- ------------------------------------------------------------------------- Wolfgang Bangerth email: [email protected] www: http://www.math.tamu.edu/~bangerth/ _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
