[Getfem-users] Penalized contact between nonmatching meshes

[logari81]

Hi all,

I have recently extended the penalized continuous contact brick to 
nonmatching meshes. However it doesn't seem to be very robust at all, 
especially when the global system contains other bricks that are based 
on Lagrange multipliers, so I am trying to find out if I have some bug 
somewhere or if it is just a theoretically expected behaviour.

In the attached files there is a quite minimal example demonstrating one 
case with good and one case with bad convergence. The only difference is 
the kind of the dirichlet conditions, penalized in the first case, with 
multipliers in the second one.

The examples can be run with
./test_contact input_simple
and
./test_contact_fail input_simple
respectively.

It would be nice to have a second opinion before spending more time on 
debugging this issue, so that I can proceed later with the 
implementation of the Alart Curnier contact bricks for nonmatching meshes.

Best regards

Kostas

input_simple
Description: Text document

plane_50.dat
Description: Netscape Proxy Auto Config

simple_2.dat
Description: Netscape Proxy Auto Config

test_contact.cc
Description: Text Data

test_contact_fail.cc
Description: Text Data