<div class="gmail_quote">On Fri, Dec 30, 2011 at 10:34, Dominik Szczerba <span dir="ltr"><<a href="mailto:dominik@itis.ethz.ch">dominik@itis.ethz.ch</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div id=":600">Very useful hint that with -fp_trap. I get trapped at:<br>
<br>
Program received signal SIGFPE, Arithmetic<br>
exception.0x00000000010fad4e in hypre_BoomerAMGRelax (A=0x1b56e9a0,<br>
f=0x195487b0, cf_marker=0x692b900, relax_type=6, relax_points=1,<br>
relax_weight=1, omega=1, l1_norms=0x0, u=0xb0f8aa0,<br>
Vtemp=0x18fd1900, Ztemp=0x0) at par_relax.c:19251925<br>
u_data[i] = res / A_diag_data[A_diag_i[i]];<br>
with res = -2.4683099805590231e+303 and A_diag_data[A_diag_i[i]] =<br>
9.287739686222747e-06. The full backtrace goes through many boomeramg<br>
lines to point back at the end to my code where (in a custom<br>
preconditioner) I solve a linear system with KSPGMRES and<br>
PCHYPRE/boomeramg. I guess I will proceed now with inspecting the<br>
values by hand, but if there still are some useful Petsc tricks I am<br>
eager to hear.</div></blockquote></div><div><br></div><div>Is your system singular? Or perhaps the BoomerAMG smoother is not stable for your problem. You didn't show the full stack.</div>