Inquiry Regarding Timestep 0 Stagnation in Thermo-Viscoelastic Subduction Simulation

ASPECT
1

Dear Aspect Development Team,

I hope this message finds you well. My name is Yuan, a novice Aspect user currently studying subduction dynamics through the Cookbook examples:

  1. Subduction initiation from Matsumoto and Tomoda (1983)
  2. Continental extension

While attempting to perform thermo-viscoelastic numerical simulations using 20 MPI cores, the computation persistently stagnates at the initial timestep (Timestep 0). I have carefully reviewed the .prm configuration parameters and verified MPI environment settings, but the issue remains unresolved.

For your reference, I have attached:

Complete parameter file: test-tem-13.prmSimulation log output:log.txt`

" The linear solver tolerance is set to 0.01.
Rebuilding Stokes preconditioner…
Solving Stokes system (AMG)… 0+3 iterations.
Newton system information: Norm of the rhs: 1.37694e+12, Derivative scaling factor: 0
Relative nonlinear residual (Stokes system) after nonlinear iteration 55: 2.16956

The linear solver tolerance is set to 0.01.
Rebuilding Stokes preconditioner…
Solving Stokes system (AMG)… 0+3 iterations.
Newton system information: Norm of the rhs: 1.37268e+12, Derivative scaling factor: 0
Relative nonlinear residual (Stokes system) after nonlinear iteration 56: 2.16285

The linear solver tolerance is set to 0.01.
Rebuilding Stokes preconditioner…
Solving Stokes system (AMG)… 0+2 iterations.
Newton system information: Norm of the rhs: 1.34657e+12, Derivative scaling factor: 0
Relative nonlinear residual (Stokes system) after nonlinear iteration 57: 2.12171

The linear solver tolerance is set to 0.01.
Rebuilding Stokes preconditioner…
Solving Stokes system (AMG)… 0+3 iterations.
Newton system information: Norm of the rhs: 1.34481e+12, Derivative scaling factor: 0
Relative nonlinear residual (Stokes system) after nonlinear iteration 58: 2.11894

The linear solver tolerance is set to 0.01.
Rebuilding Stokes preconditioner…
Solving Stokes system (AMG)… 0+2 iterations.
Newton system information: Norm of the rhs: 1.2977e+12, Derivative scaling factor: 0
Relative nonlinear residual (Stokes system) after nonlinear iteration 59: 2.0447

The linear solver tolerance is set to 0.01.
Rebuilding Stokes preconditioner…
Solving Stokes system (AMG)… 0+3 iterations.
Newton system information: Norm of the rhs: 1.28825e+12, Derivative scaling factor: 0
Relative nonlinear residual (Stokes system) after nonlinear iteration 60: 2.02982"

Please accept my sincere gratitude for your continued development of this outstanding geodynamic modeling tool. Any guidance you could provide would significantly advance my understanding of Aspect’s parallel computation workflow.

Thank you in advance for your time and expertise.

Kind regards,
Yuan

test-tem-13.prm (7.0 KB)

log.txt (20.3 KB)

2

Hi @YUAN,

Thank you for posting your question to the forum.

It will be difficult to diagnose exactly what the issue in this type of complicated simulation without knowing the full development history of your model.

Rather than diagnose the exactly why your model has poor convergence behavior at the first time step (there can be quite a few causes), my suggestion is to start from the PRM provided through these recent papers on subduction using ASPECT.

The ASPECT website provides an up to date and searchable list of papers that use ASPECT, and a few examples of recent studies using ASPECT to model subduction include:

There are quite a few more recent papers focusing on subduction, so again my advice would be to start with the PRM file from one of these studies that most closely matches the intended model design for your study.

Cheers,
John

3

Thank you so much for your assistance. Your guidance and the literature recommendations have been invaluable to my ongoing numerical simulation work. I truly appreciate the time and effort you put into supporting my research