[Gmsh] Warning: jac < 0

Rene Schneider rene.schneider at feins.org
Sat Aug 22 09:38:42 CEST 2015


On 20.08.15 18:07, Aleksejs Fomins wrote:
> Dear GMSH,
>
> I have a relatively simple .geo geometry of a prism with curved edges
> with several thin layers of stuff on it. When I attempt to make 2nd
> order curvilinear mesh from it, it complains that several elements have
> negative jacobians. When I visualise this mesh with my own code, I see
> that those elements are self-intersecting.
>
> Could someone suggest what is the cause of such behaviour, and what
> steps in design I should follow in order to avoid it.

Hi Aleksejs,

this reminds me of a bug upon which I stumbled myself. This would be the 
case if you define the "curved edges" by b-splines.

Sebastian Eiser then pointed me to the bug report:
   https://geuz.org/trac/gmsh/ticket/184 (gmsh:gmsh)

As far as I know, this is still not fixed.

If you observe this with the latest svn version, you can add your 
example to the bug report, to give it more weight.

Rene


-- 
----------------------------------------------------

       Rene Schneider

       Hainichener Str. 17a, 09569 Oederan, Germany

       Tel.:  +49-37292-28596
       Mobil: +49-163-6415093
       rene.schneider at feins.org

       http://feins.org

----------------------------------------------------