3D viscoelastic problem with uniform reverse slip within a triangle patch, does not converge in Pylith 4.2

PyLith
1

Hi there,

I recently met a new problem when trying to calculate GF for 3D viscoelastic problem. I use .spatialdb file to assign a uniform reverse slip on a triangle patch (and slip outside of it is zero). The material is domain uniform: elastic oceanic and continental crust of shear modulus 48GPa; viscoelastic oceanic mantle of viscosity 10^20 Pa.S and shear modulus 64GPa, and viscoelastic continental mantle of viscosity 10^19 Pa.S and shear modulus 64GPa. For most of slip patches the simulation of each time step converge normally, as shown below:

0 TS dt 0.3 time -0.3

0 SNES Function norm 1.375377494776e-01

Linear solve converged due to CONVERGED_ATOL iterations 98

1 SNES Function norm 1.804826184505e-12

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 1

1 TS dt 0.3 time 0.

0 SNES Function norm 1.357814352493e-02

Linear solve converged due to CONVERGED_ATOL iterations 88

1 SNES Function norm 2.052612424117e-12

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 1

2 TS dt 0.3 time 0.3

0 SNES Function norm 1.077117544985e-02

Linear solve converged due to CONVERGED_ATOL iterations 82

1 SNES Function norm 1.826949239838e-12

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 1

3 TS dt 0.3 time 0.6

0 SNES Function norm 8.853957182423e-03

Linear solve converged due to CONVERGED_ATOL iterations 74

1 SNES Function norm 1.388661047318e-12

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 1

But for simulation on some GF patches (26 out of 300), the simulation result is always zero, as shown below:

-- Solving problem.

0 TS dt 0.3 time -0.3

0 SNES Function norm 0.000000000000e+00

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 0

1 TS dt 0.3 time 0.

0 SNES Function norm 0.000000000000e+00

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 0

2 TS dt 0.3 time 0.3

0 SNES Function norm 0.000000000000e+00

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 0

3 TS dt 0.3 time 0.6

0 SNES Function norm 0.000000000000e+00

Nonlinear solve converged due to CONVERGED_FNORM_ABS iterations 0

with some error showing after the last simulation time step
WARNING! There are options you set that were not used!

WARNING! could be spelling mistake, etc!

There is one unused database option. It is:

Option left: name:-mg_fine_pc_type value: vpbjacobi source: code

**I wonder what might cause some simulation fails to converge from the first step. I checked the fault slip file .spatialdb and they look quite OK, with proper slip within the target zone, instead of zero slip.

Any suggestion and hint could be much appreciated!**

Thanks,

Qingjun

2

The solution is not failing to converge. The initial residual is zero, so the solution is zero and no iterations are required to solve the equations. I suspect that these simulations have zero fault slip, which you can verify by examining the fault info and data files.

The warning you are getting at the end of the simulation is simply a consequence of the solver not using the fine scale preconditioning because the residual is zero.

3

Hi Brad,

I double checked the GF cases with all zero output. You are right, no slip is applied into the active fault zone for these cases. The spatialdb file generation script has bug. Thank you very much for your time!

Best,

Qingjun