[CIG-SHORT] Uniform time-stepping for long runs on viscoelastic domain (Brad Aagaard) (Matthew Knepley)

Mon Jun 8 12:26:53 PDT 2015

Dear Brad and Matt, 

Brad: The minimum relaxation time in my problem is 1 year. The dt that I am using is 0.1 years. The simulation is quasi-static with two different viscoelastic materials, one that relaxes 100x faster than the other. For longer runs, the simulation gets stuck and doesn't create and write to some .xmf files (the ones that usually contain the displacements and velocities) when the nondimensionalization time is set to the default value of 1 year. When I changed the nondimensionalization time to 0.1 years (my dt value), the simulation ran fully and output the correct number of files. It's odd that a factor of 10 in the nondimensionalization time (rather than a number closer to 10e+6 as Matt wrote) could have such a strong effect on the success of the run. Maybe the effect becomes more pronounced as total_time / dt becomes larger?

Matt: Flagging bad nondimensionalization would be a great idea! This is actually the second time I've struggled for a couple of days only to realize the my choice of nondimensionalization parameters (or rather my forgetting of the importance of setting such parameters) was the cause of my problems. 

Best, 

Michal 

____________
Michal Dichter
Department of Physics and Astronomy
University of New Mexico
(860) 459-0956

________________________________________
From: CIG-SHORT <cig-short-bounces at geodynamics.org> on behalf of cig-short-request at geodynamics.org <cig-short-request at geodynamics.org>
Sent: Monday, June 8, 2015 1:00 PM
To: cig-short at geodynamics.org
Subject: CIG-SHORT Digest, Vol 105, Issue 10

Send CIG-SHORT mailing list submissions to
        cig-short at geodynamics.org

To subscribe or unsubscribe via the World Wide Web, visit
        http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
or, via email, send a message with subject or body 'help' to
        cig-short-request at geodynamics.org

You can reach the person managing the list at
        cig-short-owner at geodynamics.org

When replying, please edit your Subject line so it is more specific
than "Re: Contents of CIG-SHORT digest..."

Today's Topics:

   1. Re: Uniform time-stepping for long runs on viscoelastic
      domain (Brad Aagaard) (Matthew Knepley)
   2. Re: Uniform time-stepping for long runs on viscoelastic
      domain (Brad Aagaard) (Brad Aagaard)

----------------------------------------------------------------------

Message: 1
Date: Mon, 8 Jun 2015 06:16:19 -0500
From: Matthew Knepley <knepley at mcs.anl.gov>
To: "cig-short at geodynamics.org" <cig-short at geodynamics.org>
Subject: Re: [CIG-SHORT] Uniform time-stepping for long runs on
        viscoelastic domain (Brad Aagaard)
Message-ID:
        <CAMYG4GkMVbe3mXxxsFod=8ct+9JjKxwZzMXoYP87kZ_GM=Sy7w at mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

On Mon, Jun 8, 2015 at 1:23 AM, Michal Dichter <mdichter at unm.edu> wrote:

>  Dear Brad,
>
> Please disregard my last response. It seems that the cause of the problem
> that I have been describing has been with the nondimensionalization of
> time. (I stumbled upon a PyLith FAQ sort-of page where the importance of
> nondimensionalization was emphasized.) I changed my "relaxation_time"
> parameter to be equal to dt rather than to the minimum relaxation time.
> This enabled my code to run to completion and output the expected number of
> files. Thank you again for your help in this matter.
>
> PyLith developers,

I think some sort of diagnostic thing, even if only in the log, for looking
at the non-dimensionalization would be useful. We have
been struggling with this in PETSc as well. What if we flag things that are
1e6 times more or less than the non-dimensionalization?

  Thanks,

     Matt

> All the best,
>
>
>  Michal
>
>
>   ____________
> Michal Dichter
> Department of Physics and Astronomy
> University of New Mexico
> (860) 459-0956
>
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>

--
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/cig-short/attachments/20150608/b0770d26/attachment-0001.html>

------------------------------

Message: 2
Date: Mon, 8 Jun 2015 10:02:24 -0700
From: Brad Aagaard <baagaard at usgs.gov>
To: <cig-short at geodynamics.org>
Subject: Re: [CIG-SHORT] Uniform time-stepping for long runs on
        viscoelastic domain (Brad Aagaard)
Message-ID: <5575CAA0.8050306 at usgs.gov>
Content-Type: text/plain; charset="utf-8"; format=flowed

Michal,

Thanks for determining that your problem was related to
nondimensionalization. What is the minimum relaxation time in your problem?

I would like to reproduce your hang so that we can generate an
appropriate error message, in addition to the more general trapping of
nondimensionalization problems Matt suggested.

Thanks,
Brad

On 06/07/2015 11:23 PM, Michal Dichter wrote:
>
> Dear Brad,
>
>
> Please disregard my last response. It seems that the cause of the
> problem that I have been describing has been with the
> nondimensionalization of time. (I stumbled upon a PyLith FAQ sort-of
> page where the importance of nondimensionalization was emphasized.) I
> changed my "relaxation_time" parameter to be equal to dt rather than
> to the minimum relaxation time. This enabled my code to run to
> completion and output the expected number of files. Thank you again
> for your help in this matter.
>
>
> All the best,
>
>
> Michal
>
>
> ____________ Michal Dichter Department of Physics and Astronomy
> University of New Mexico (860) 459-0956
>
>
>
> _______________________________________________ CIG-SHORT mailing
> list CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>

------------------------------

Subject: Digest Footer

_______________________________________________
CIG-SHORT mailing list
CIG-SHORT at geodynamics.org
http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short

------------------------------

End of CIG-SHORT Digest, Vol 105, Issue 10
******************************************