[Getdp] mesh adaptation

Michal Hajek hajek1 at karlov.mff.cuni.cz
Mon Sep 19 18:59:58 CEST 2005


Hello :)

I am trying to figure out how to make adaptation of mesh. 
First I solve simple electrostatics problem and I get correct scalar
potential on some mesh. Than I compute flux resulting from the potential
gradient. So far so good. 

Than I wanted in PostOperation :

{ Name Tar; NameOfPostProcessing MyPostProcessing;
     Operation {
            Print [ Flux, OnElementsOf Domain_Elektrika, Dimension 2,
            Adapt P1, Target 300, Value 3, Format Gmsh, File
            "refined.msh" ];
}}

I thought that I make a very simple approximation of the error - just
the flux itself. Unfortunately, if Target value is too small (eg.
smaller than 40 in my case) I receive 
this error:

GetDP   : Error     : MIN1D not converged: f(3.25927) = -71350.2


or if Target value is too large (eg. >40 in my case) I receive another
error:
GetDP   : Error     : GSL: function not continuous (fsolver.c, line 40)


I am using getdp version 1.0.1 compiled with gsl-1.6 on a debian amd64
mashine. 

If someone is interested, I have put .geo, .msh, .pro and solver.par
files here:
http://material.karlov.mff.cuni.cz/people/hajek/getdp/

Thank you for any kind of help :)