GetDP, cyl.axis

[Christophe Geuzaine]

Samuel Kvasnica wrote:
> 
> > There is an elegant solution to the problem: you should use a slightly
> > modified jacobian matrix, namely "VolAxiSqu", instead of "VolAxi". The
> > idea is to use r^2 as the radial coordinate, instead of r. If you want a
> 
> That seems to be really a good solution to remove the singularity. Just a question
> about how it is implemented: does it mean, I'll have to generate the mesh also in r^2
> coordinates or there's some mechanism to convert r to r^2 and r^2 back to r implicitly
> during post- and preprocesing ?

Everything is implicitly done with the definition of the jacobian. Of
course, you have to specify this jacobian both in the processing _and_
in the postprocessing stage.

-- 
Christophe Geuzaine

Tel: +32-(0)4-366.37.10    mailto:Christophe.Geuzaine at ulg.ac.be
Fax: +32-(0)4-366.29.10    http://www.geuz.org