On Mon, 2 Jun 2008, Derek Gaston wrote:

>> If this is the "rounding error when locating new points" bug again,
>> then what's triggering it isn't the file format, it's the node
>> locations.  Cubes of floating-point friendly dimensions are less
>> likely to trigger such a bug than a real unstructured mesh...
>> otherwise I would have found this problem myself months ago and I
>> wouldn't be bugging everyone with it today.
>
> My mesh is pretty well behaved.... no tiny elements or anything...

Triggering that bug didn't need tiny elements, or badly shaped
elements, just neighboring elements whose children's nodes end up with
rounding error that falls on opposite sides of the right arbitary
boundaries.  If you just do build_cube on a unit cube, for an extreme
counter-example, all your node coordinates end up being integer
multiples of powers of 2, exactly representable in binary, and there
may be no rounding error when refining at all.
---
Roy

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to