Next: Threshold violations
Up: Overcoming Numerical Difficulties
Previous: Divergence in the nonlinear
The bottoming out may be due to:
- truncation error in finite precision arithmetic. If there are
reactive terms in the PDEs and these reactive terms are nearly equal
and have a magnitude of about
larger than the residual
when it bottoms out, then the residual being calculated is dominated
by truncation error. - an incorrect system of equations. To many boundary conditions
may lead to an overconstrained system of nonlinearx equations in
which you can only hope to minimize the residual.
Suggestions:
- If truncation error is the problem, rather than relaxing the
relative and/or absolute tolerances, it is probably better to turn
on the update norm (-useUpdateNorm).
Dan Yergeau
Thu Nov 7 17:12:57 PST 1996