Initial thought: could you be generating a NaN in your own code as the mesh is refined? E.g. if you're interpolating some expression that leads to a 0/0 at some points in the domain. On 24 March 2017 at 10:45, Buesing, Henrik <HBuesing@eonerc.rwth-aachen.de> wrote:

Dear all,

I am doing a refinement study for two-phase flow in porous media. The first five levels of refinement succeed and I get reasonable results. On level six I get an error (see [1]). I also attached the trace.

I tried the suggestion with –snes_converged_reason and –ksp_converged_reason, but this gives no output. Since the error seems to originate from the projection on the linear output space, I removed these parts. Then I get [2].

Do you have an idea what I am doing wrong? Why does the formulation seem to be correct for coarse discretizations, but gives NAN for finer ones?

Thank you!

Henrik

[1]

firedrake.exceptions.ConvergenceError: Nonlinear solve failed to converge after 0 nonlinear iterations.

Reason:

Inner linear solve failed to converge after 0 iterations with reason: unknown reason (petsc4py enum incomplete?), try with -snes_convered_reason and -ksp_converged_reason

%s""" % (snes.getIterationNumber(), msg))

[2]

Nonlinear solve did not converge due to DIVERGED_FNORM_NAN iterations 0

--

Dipl.-Math. Henrik Büsing

Institute for Applied Geophysics and Geothermal Energy

E.ON Energy Research Center

RWTH Aachen University

------------------------------------------------------

Mathieustr. 10 | Tel +49 (0)241 80 49907 <+49%20241%208049907>

52074 Aachen, Germany | Fax +49 (0)241 80 49889 <+49%20241%208049889>

------------------------------------------------------

http://www.eonerc.rwth-aachen.de/GGE

hbuesing@eonerc.rwth-aachen.de

------------------------------------------------------