[petsc-users] questions about snes/ex14

[Xiangdong]

Hello everyone,

I have some questions about running ex14 with different options -snes_fd
-fdcoloring and -snes_mf.

I run ex14 with -snes_max_it 1 to see how many function evaluations and
linear solver iterations take.

In this example, there are 64 unknowns.

1) -fdcoloring.  For -fdcoloring, -snes_view says "total number of function
evaluations=2" while the real number of calls of FormFunction (by putting a
counter inside the function) is 30.  Only 2 function calls reported by
snes_view seems to be small. How does snes_view count the number of
function evaluations? (For snes_fd, the number of function evaluations is
67 (64+3), which is normal.)

If I use -snes_fd_color, it still says "total number of function
evaluations=2" while the counter inside the FormFunction tells 12 now.

2)  -snes_mf. For -snes_mf, "the total number of linear solver
iterations=1". Given that no preconditioner (pcnone) is used, why does the
gmres converges in just one iteration?

Thank you.

Xiangdong
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140404/b9bc89e7/attachment.html>