[CIG-SHORT] Pylith Green's functions

[Demian Gomez]

Hi Charles,

I thought so. What I'm doing is running a forward simulation for each
patch. I estimated the times and even though Pylith has to load the
matrices each time it starts a new patch (twice actually, once for strike
slip and another for reverse) it takes much much less than running the
Green's function problem. I didn't think of the time dependent problem that
you suggested. This sounds like a very good solution. I'll try to implement
this method. Do you have by chance an old slip time spatialdb where you did
this, as an example?

Thanks,
Demián

On Thu, Apr 7, 2016 at 6:17 PM, Charles Williams <willic3 at gmail.com> wrote:

> At present, I don’t think PyLith is set up to do this sort of thing.
> Depending on your situation, I can think of two options:
>
> 1.  Generate your Green’s functions in batches that can be done
> simultaneously.  You could define several subregions on your fault and run
> each of those separately.  This would work if you have a cluster or several
> workstations to run on.  Once all the functions are generated you can
> combine them however you like.
> 2.  Create a time-dependent problem where each patch uses a different slip
> history (you could write a script to generate this).  At t=0 you slip patch
> 0 by a positive amount, at t=1 you slip patch 0 by a negative amount (to
> cancel out the slip from the previous time step) and patch 1 by a positive
> amount, etc.  This was the sort of method we used before we had Green’s
> functions officially included in PyLith.
>
> Cheers,
> Charles
>
>
> On 8/04/2016, at 9:45 AM, Demian Gomez <ddgomez at memphis.edu> wrote:
>
> Hello,
>
> I have a model with a fault that is made of 16k nodes. I need the impulse
> responses to invert for the slip at the fault, but I've grouped the nodes
> into square slip patches (20 x 20 km). Therefore, I have ~ 200 patches that
> are formed by multiple nodes on my fault interface. Is there a way to run
> the Green's function problem in Pylith (2.1.0) with more than one impulse
> at the time? Running the problem with 16k nodes will take too much time and
> I don't need each individual node's impulse response, only the impulse
> response of each slip patch made of multiple nodes.
>
> Thanks,
> --
> *Demián D. Gómez*
> Graduate Research Assistant & PhD Candidate
> Center for Earthquake Research and Information
> 3890 Central Ave, Memphis TN 38152
> Tel: +1 (901) 678-4809
> Cell: +1 (901) 900-7324
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>
>
> Charles A. Williams
> Scientist
> GNS Science
> 1 Fairway Drive, Avalon
> PO Box 30368
> Lower Hutt  5040
> New Zealand
> ph (office): 0064-4570-4566
> fax (office): 0064-4570-4600
> C.Williams at gns.cri.nz
>
>
> _______________________________________________
> 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/20160407/2fc1d560/attachment-0001.html>