[CIG-SHORT] statevarsvertex dof error

[Brad Aagaard]

Birendra,

I have found two small bugs with easy workarounds.

1. The initial state variables are not being setup correctly resulting 
in the error message you get. Since you are setting the initial state 
variables to zero, you can omit this and get the behavior you want.

2. The field split preconditioner has a bug in it that causes the matrix 
preallocation to be wrong when there is a buried fault in which the 
buried edges have not been marked. The "workaround" (proper thing to do) 
is to mark the buried edges in the meshes (CUBIT) and use the edge 
property.

I will fix these bugs as soon as I can find time to work on PyLith.

Brad


On 07/04/2014 06:57 AM, Birendra jha wrote:
> Hi
>
> I am trying to run the attached test case with one fault on Pylith, both v1.9 and v2.0.2, using windows binary. On version 2, I get the following error:
>
> -- Initializing integrators.
> v: 1600, dof: 1, stateVarsVetex: 2
> assertion "stateVarsVertex.size() == dof" failed: file "friction/FrictionModel.c
> c", line 235, function: virtual void pylith::friction::FrictionModel::initialize
> (const pylith::topology::Mesh&, pylith::feassemble::Quadrature*)
>
> On v1.9, it runs but the right corner node of the fault gets huge slip displacements, order of 5-10 m, and even when the boundaries of the box are fixed at 0 displacement, one of interior nodes just shoots out of the box (see it using Warp in paraview). I thought it is the fault approximation in v1.9 and earlier that is causing this, and fault edges get large tractions. However, I never had slip there, just unreal tractions.
> I switched to v2 but I am not able to run it. Can you please suggest something?
>
> Thanks and regards
> Birendra
>
>
>
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>