Author |
Message |
jorge pullin (pullin)
New member Username: pullin
Post Number: 1 Registered: 10-2003
| Posted on Friday, October 03, 2003 - 03:34 pm: | |
Hi I'm trying to use flexpde to solve three nasty coupled second order elliptic PDEs. It seems to be having trouble (I give it a simple constant exact solution as initial data, and it does not seem to realize it, it keeps on running, adding a lot of new elements along an axis, making very small corrections). Is there a way to get some more detailed information about what is going on when flexpde runs? In particular, to get the points of the grid and for which particular equation in the system does the "MAX Error" happen? Thanks Regards Jorge |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 45 Registered: 06-2003
| Posted on Sunday, October 05, 2003 - 06:53 pm: | |
There is no mechanism for extracting detailed error measures, but since the regridder is driven by cell errors, you can assume that the error is high where the mesh is dense. The behavior you describe is frequently encountered when a boundary condition is in conflict with the PDE. FlexPDE struggles to find a compromise between the conflicting demands of the PDE and the boundary condition. Try reanalyzing the meaning of your boundary conditions, and the consistency with the PDE, paying particular attention to the sign of natural BC's.
|
|