[CIG-SEISMO] 1000s of receivers SPECFEM Cartesian

[Bernhard Schuberth]

Dear Dimitri, dear all,

I have now committed the new version of locate_receivers.f90 to SPECFEM3D_GLOBE. Stations are now read in in subsets to reduce memory requirements for large number of stations (NREC_SUBSET_MAX is currently set to 200 in setup/constants.h.in). The code is following the strategy used in locate_sources.f90 that I think Vala Hjörleifsdóttir implemented some years ago. I have run some tests with around 50000 stations that went fine, but further testing would be good.

Best wishes,

Bernhard 


On Jun 6, 2013, at 2:05 PM, Dimitri Komatitsch wrote:

> 
> Dear Andrea, dear all,
> 
> Bernhard Schuberth has solved this problem and implemented a nice source code patch for SPECFEM (GLOBE I think, but it would work for Cartesian as well with minor modifications).
> He sent it to me last week but unfortunately in my group in Marseille we do not have time to work on that in June, thus I can suggest two options:
> 
> 1/ Bernhard commits it to SVN now (he has already fully tested his modifications therefore that is safe to do)
> 
> 2/ we wait until July or August and then people in my group here could maybe do more tests
> 
> I think #1 is probably better.
> 
> Thanks,
> Dimitri.
> 
> On 06/05/2013 06:51 PM, Andrea Colombi wrote:
>> Hello,
>> when running simulation with SPECFEM cartesian I would like to have in
>> output many receivers, lets say ~10000 for relatively
>> long simulation (1000000 timestep), in parallel simulation that may
>> feature 100-1000 cpu. However the program stales because of unclear
>> memory reason (not a direct error of SPECFEM) related to the cluster.
>> 
>> My strategy is at the moment to dump the fields at each corner point of
>> the mesh through netcdf library that can be easily ported on other
>> machines. For a single simulation everything works smooth and by
>> adjusting the sampling rate the total size of the file remains under
>> control.
>> However, in this approach, if the mesh changes, the position of the
>> point also changes, complicating eventually the comparison of the results.
>> 
>> Having tested few configuration, I know that I/O of ASCII seismograms is
>> a kind of heavy task that slows down the time integration but I wonder
>> if a different strategy to do that is available. Not sure if generic
>> binary output would help if the files must be read from other
>> potentially different machines.
>> 
>> Thanks for the help
>> Andrea
>> 
>> 
>> _______________________________________________
>> CIG-SEISMO mailing list
>> CIG-SEISMO at geodynamics.org
>> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
>> 
> 
> -- 
> Dimitri Komatitsch
> CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics,
> UPR 7051, Marseille, France    http://komatitsch.free.fr