On Tue, Apr 10, 2012 at 4:39 PM, wang yunbo <[email protected]> wrote:
> Hi Dr. Wheeler, > > I drew a circle and then passed it to computeDistanceFunction2d through > cython wrapper from you. I found out if the resolution (dL) of the mesh is > smaller than 1e-8, there is some error with computeDistanceFunction2d. > > For example, in the test.py file as attached, the plot would be a perfect > circle if "Magnification" variable is set to be bigger than 1e3. Otherwise, > the circle would be wavy on the boundary. However, when the > computeDistanceFunction2d was commented out, it would be a perfect circle > again. In other words, if I use meters as my base unit of length, the code > fails. But if I use mm as my base unit of length, the code is fine. > This doesn't make sense to me since the meshes are self similar. > > Is there any way to solve that? I don't want to rescale the mesh to > different units. > Thanks for noticing this. It is almost certainly a bug in LSMLIB or a limitation. I'll add this to the LSMLIB ticket. <http://matforge.org/fipy/ticket/432> If you notice any other weirdness feel free to add a comment in the ticket or any of your thoughts in general. Don't hesitate looking under the hood of LSMLIB if you feel that you might be able to debug these sorts of issues. I am in contact with the LSMLIB author and I also now have commit privileges to the repository so we can try and fix these issues as they arise. Cheers -- Daniel Wheeler
_______________________________________________ fipy mailing list [email protected] http://www.ctcms.nist.gov/fipy [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
