Topic: Deprecating nltransienttransport and nonstationarytransport

Hi,

As a preparation for the next set of changes (related to changes to DOFs and BCs), I started looking at the TM-module.
I think we are ready to remove nltransienttransport and nonstationarytransport. They are blockers for fixing the DOFs and BCs. We could do massive rewrites of them, which would only make them work like TransientTransport already does anyway.

The tests that give different output with TransientTransport   (for example tmpatch11.in) is because TransientTransport is actually more accurate/consistent when dealing with input files that have sudden nonzero BCs.

Tests should be switched to TransientTransport, and the CHECK-values should be changed.

2

Re: Deprecating nltransienttransport and nonstationarytransport

Mikael, I have been working on BCs unification recently, see my post here: http://www.oofem.org/forum/viewtopic.php?pid=6498#p6498
We should definitely coordinate the effort to avoid any cross changes.

Re: Deprecating nltransienttransport and nonstationarytransport

I don't think we are at risk of conflicting changes here. I'm not actually doing any real changes yet, just checking to see what breaks when i the Dirichlet-bc handling, and estimate how hard they are to fix.


I wrote TransientTransport to superseed "nonstationarytransportproblem" and "nltransienttransportproblem".

Many tests that still aren't using TransientTransport give different result because of the difference from asking the BC for VM_Velocity compared to computing VM_Velocity using the time-stepping method used.

I have started to switch as many input files as possible over to TransientTransport, adjusting errorchecks whereever necessary. A few tests remain and require additional investigation why TransientTransport isn't working (well), e.g. tmpatch16.in