I've been using -ts_type bdf for some cases, and I'm seeing odd behavior I
haven't seen previously; BDF is doing two nonlinear solves per step instead
of one. Not just on the first iteration where I understand it needs two
solves to do crank-nicolson. Output pasted below.
I recently added
Mark,
> The machine, compiler and MPI version should not matter.
I might have missed something earlier in the thread, but parmetis has a
dependency on the machine's glibc srand, and it can (and does) create
different partitions with different srand versions. The same mesh on the
same code on