Petsc ILUDT requires further work - both algorithmic correctness and performance optimization. Current status is not ready for users yet.
Shall we take these options off until it is ready? Hong On Mar 22, 2010, at 2:51 PM, Jed Brown wrote: > On Mon, 22 Mar 2010 14:36:50 -0500, Barry Smith <bsmith at mcs.anl.gov> > wrote: >> >> You should consider that PETSc currently has no drop tolerance >> factorizations; what is in the source code is just partial code and >> partial place holders. >> We could add this to the manual pages. > > There is a man page for PCFactorSetDropTolerance, the user's manual > discusses it, and the option is available in -help. I'm just pointing > out that this is fairly confusing. The manual also mentions > PCFactorDTSetReuseFill which does not currently exist. > > There seems to be something available with -pc_type icc > -pc_factor_mat_solver_package bas, but this is actually undocumented > so > I'm not concerned about whether or not it works. > > Jed
