[CIG-SHORT] Pylith stuck in the final step "Formulation finalize."

[Matthew Knepley]

On Tue, Oct 2, 2018 at 3:01 AM <hkj at mail.ustc.edu.cn> wrote:

>
> Thank you for nice advice, I already used hdf5 output before, but turning
> off the volume output do help a lot!
>
> There are still two things bothering me:
>
> 1. Since I only need two time points: 30yr's and 40yr's output among the
> whole 40 yr's simulation, but the minimum time step is 0.01 yr. Except
> changing material properties, can pylith only simulate the model for  dt =
> 0.01 yr in the first few years, and than gradually get larger in the
> remaining time period?
>

Yes, you are in control of the timestep. You can use

[pylithapp.problem.formulation]
solver = pylith.problems.SolverNonlinear
time_step = pylith.problems.TimeStepAdapt

or TimeStepUser.


> 2. Is it possible make material properties change with time during the
> simulation?(For example, let viscosity to be 1.0e+20 during 0~10 year and
> to be 1.0e+21 during 10~20 year)
>

Brad and Charles will know how to do this. However, does this make physical
sense?

  Thanks,

     Matt



> Sincerely
> Kejing
>
> > -----原始邮件-----
> > 发件人: "Brad Aagaard" <baagaard at usgs.gov>
> > 发送时间: 2018-10-02 06:52:51 (星期二)
> > 收件人: cig-short at geodynamics.org
> > 抄送:
> > 主题: Re: [CIG-SHORT] Pylith stuck in the final step "Formulation
> finalize."
> >
> > Kejing Huang,
> >
> > When PyLith appears to be stuck, it usually means that at least one
> > process aborted but MPI didn't exit cleanly, so one or more processes
> > are left hanging. It sounds like there is a chance you are running out
> > of disk space.
> >
> > If you are not using HDF5 output, the first thing to do is to switch
> > from VTK output (default) to HDF5 output. Secondly, write out only the
> > data fields that you are interested in; if you do not care about volume
> > output, turn it off or only write the final values. See the examples
> > (such as examples/3d/subduction) in the PyLith manual for illustrations
> > of how to do this.
> >
> > Regards,
> > Brad
> >
> > On 09/28/2018 09:28 PM, hkj at mail.ustc.edu.cn wrote:
> > > Dear Sir,
> > > I'm now running a .exo model with 55000 nodes. In the pylith
> simulation
> > > I used total time = 50 yr with dt = 0.01yr(since I need viscosity to
> be
> > > small, 0.01yr is the largest time step as pylith suggested).Thus the
> > > output file is about 80GB.
> > >
> > > The simulation and iteration run correctly, but it has about 20%
> > > probability to stuck at the final step after it shows "Formulation
> > > finalize".
> > >
> > > I waited for about 10 hours(even longer than the whole simulation
> time),
> > > but it still stuck at the final step. The last few lines of output is
> > > showed below
> > >
> > > /
> > > /
> > > /``` [OUTPUT]/
> > > /......./
> > > /......./
> > > /......./
> > > /Linear solve converged due to CONVERGED_ATOL iterations 50/
> > > / >>
> > >
> /home/collyking/Software/Pylith/lib/python2.7/site-packages/pylith/problems/TimeDependent.py:208:run/
> > > / -- timedependent(info)/
> > > / -- Finishing advancing solution from t=1.57725e+09*s to
> t=1.57788e+09*s./
> > > / >>
> > >
> /home/collyking/Software/Pylith/lib/python2.7/site-packages/pylith/problems/Implicit.py:242:poststep/
> > > / -- implicit(info)/
> > > / -- Writing solution fields./
> > > / >>
> > >
> /home/collyking/Software/Pylith/lib/python2.7/site-packages/pylith/faults/Fault.py:191:writeData/
> > > / -- faultcohesivekin(info)/
> > > / -- Writing fault data./
> > > / >>
> > >
> /home/collyking/Software/Pylith/lib/python2.7/site-packages/pylith/faults/Fault.py:191:writeData/
> > > / -- faultcohesivekin(info)/
> > > / -- Writing fault data./
> > > / >>
> > >
> /home/collyking/Software/Pylith/lib/python2.7/site-packages/pylith/problems/Formulation.py:273:finalize/
> > > / -- implicit(info)/
> > > / -- Formulation finalize./
> > > ```
> > >
> > > Is it the normal case since the output file is too large?
> > >
> > > Sincerely
> > > Kejing Huang
> >
> > _______________________________________________
> > CIG-SHORT mailing list
> > CIG-SHORT at geodynamics.org
> > http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20181002/1f565a14/attachment-0001.html>