[CIG-SHORT] early benchmarking results
Brad Aagaard
baagaard at usgs.gov
Fri Aug 25 17:10:07 PDT 2006
Hi all:
I have started tabulating the results for the Reverse Slip without
Gravity benchmark. PyLith and GeoFEST produce nearly identical
results, as expected. The integrated L2 errors in the elastic
displacements for the 1000m resolution simulations are 1.6791e+05 and
1.6793e+05 for PyLith and GeoFEST, respectively. So... PyLith "wins"
by a micron. The integrated L2 norm difference between the two
elastic solutions from PyLith and GeoFEST is 1.0587e+03. At this
point, about all we can say is that PyLith and GeoFEST agree >100
times better with each other than they do with the analytical
solution.
Next steps: Comparing the displacements and velocities for the
viscoelastic relaxation, examining (plotting) the local error, and
calculating the errors for the 500m resolution simulations.
Brad
More information about the CIG-SHORT
mailing list