[CIG-SHORT] A problem with Green's function generation examples

[Brad Aagaard]

Zhou,

The problem is that in pylithapp.cfg you specified the fault to be 
FaultCohesiveKin:

[pylithapp.timedependent.interfaces]
fault = pylith.faults.FaultCohesiveKin

This sets up several defaults, one of which is a SimpleDB for a slip 
distribution (note the backtrace information in the error message). 
PyLith is complaining because there are parameters missing for some of 
these defaults. This is why you don't see a missing filename in your 
.cfg files.

There is also another error in your pylithapp file. When computing 
Green's functions, the problem changes from TimeDependent to GreensFns. 
As a result, you need to use the generic name of the component "problem" 
rather than "timedependent" or "greensfns" in pylithapp.cfg.

Example (there are other settings that need to be updated as well):

# NEXT LINE timedependent -> problem
[pylithapp.problem]
materials = [elastic]

# NEXT LINE timedependent -> problem
[pylithapp.problem.materials.elastic]
quadrature.cell = pylith.feassemble.FIATSimplex
quadrature.cell.dimension = 3
label = elastic lithosphere
id = 1
db_properties.label =  uniform elastic
db_properties.iohandler.filename = elastic.spatialdb

It looks like you have some other errors in your .cfg file as well 
related to BC and output on the ground surface. You might want to 
consult the .cfg files in examples/3d/hex8.

Regards,
Brad


On 9/11/12 8:53 PM, Zhou Yu wrote:
> Dear developers:
>           I'm so glad that pylith-1.7.1 can compute static Green's
> function and a 2D example in contained in the example directory of
> pylith release package. When I ran the 2D strikeslip example, it went so
> well. Then I ran a simple 3D example written by myself, the following
> error message prompted out after I ran "pylith
> --problem=pylith.problems.GreensFns"
>
>     Traceback (most recent call last):
>        File "/opt/pylith-1.7.1-linux-x86_64/bin/pylith", line 39, in
>     <module>
>          start(applicationClass=PyLithApp)
>        File
>     "/opt/pylith-1.7.1-linux-x86_64/lib/python2.7/site-packages/pythia-0.8.1.14-py2.7.egg/pyre/applications/__init__.py",
>     line 41, in start
>          shell.run(**kwds)
>        File
>     "/opt/pylith-1.7.1-linux-x86_64/lib/python2.7/site-packages/pythia-0.8.1.14-py2.7.egg/pyre/applications/Shell.py",
>     line 125, in run
>          app.applyConfiguration(context)
>        File
>     "/opt/pylith-1.7.1-linux-x86_64/lib/python2.7/site-packages/pythia-0.8.1.14-py2.7.egg/pyre/inventory/Configurable.py",
>     line 84, in applyConfiguration
>          raise ValueError("%s\nBacktrace - Component %s" % (err.message,
>     aliases))
>     ValueError: Error while configuring spatial database reader
>     (simpleioascii, iohandler):
>     Filename for spatial database not specified.
>     Backtrace - Component simpleioascii, iohandler
>     Backtrace - Component simpledb, slip
>     Backtrace - Component stepslipfn, slip_function
>     Backtrace - Component eqkinsrc, rupture
>     Backtrace - Component singlerupture, eq_srcs
>     Backtrace - Component faultcohesivekin, fault
>     Backtrace - Component interfaces
>     Backtrace - Component greensfns, problem
>     Backtrace - Component pylithapp
>
> I checked every line of my pylithapp.cfg and greenfns.cfg, but I
> couldn't find anything wrong. Then I looked up in the maillist,
> fortunately I found another example in CIG-SHORT Digest, Vol 70, Issue
> 18 in the attachment by Rowena Lohman. Her example seems all right
> according to her mail except for a minor output problem, but when I
> downloaded the attachment and ran it, the same error message prompt up
> again. I can't figure out the error message because I do have specified
> every spatialdb file name!
>
> I attached both my example and Lohman's in this mail. My pylith version
> is pylith-1.7.1-linux-i686 binary. Could you help me with this problem?
>
>
> _______________________________________________
> CIG-SHORT mailing list
> CIG-SHORT at geodynamics.org
> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>