On Mon, Jul 26, 2010 at 12:21 PM, Roy Stogner <[email protected]> wrote:
>
> On Mon, 26 Jul 2010, John Peterson wrote:
>
>> I'm just guessing but if you're doing refinement, you're probably
>> failing some of our built-in geometric tests which run when new points
>> are inserted.  These tests use a default absolute tolerance that may
>> be too small for the Mesh you are working with.
>
> This definitely *used* to be a libMesh bug, but I ran into it years
> ago and changed the tests to use a relative-to-element-size tolerance
> where possible.


Ah, I see: MeshRefinement::add_point calls LocationMap::find with a
passed tolerance, which is set to

const Real pointtol = this->hmin() * TOLERANCE;

by Elem::refine.  Nice.


> But it's been a long time; although the problem was fixed for me then,
> there may have been some place I wasn't able to change or some new
> code written since then that used absolute tolerances inappropriately.

Well the ex14 traceback appears to end in DofObject::n_comp.  There is
some extra debugging information that gets printed in that function
when -DDEBUG is defined so David, perhaps you could recompile with
METHOD=dbg and run it again.  At least we would get some extra print
statements, it's also possible you will hit an assert even before you
reach that point.

-- 
John

------------------------------------------------------------------------------
The Palm PDK Hot Apps Program offers developers who use the
Plug-In Development Kit to bring their C/C++ apps to Palm for a share 
of $1 Million in cash or HP Products. Visit us here for more details:
http://ad.doubleclick.net/clk;226879339;13503038;l?
http://clk.atdmt.com/CRS/go/247765532/direct/01/
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to