[CIG-SHORT] pylith time stepping

[Brad Aagaard]

Chris-

This was indeed a bug.

I have written a test for this and pushed a fix to the development 
version. If you want to fix it in your code, in 
pylith/meshio/OutputManager.py change

   nsteps = int(numTimeSteps / (1+self.skip))

to

   nsteps = int(1 + (numTimeSteps-1) / (1+self.skip))

Thanks for reporting this error.

Brad


Christopher DiCaprio wrote:
> It looks like the problem with the data files occurs when
> total_time / max_dt  < output.skip
> 
> - Chris
> 
> 
> On May 12, 2009, at 6:45 AM, Brad Aagaard wrote:
> 
>> Chris-
>>
>> As Charles noted, the time step for the Maxwell model is independent 
>> of the deformation rate, so the adaptive time step is uniform. It does 
>> depend on the Maxwell time. The time step for an elastic material is 
>> infinite (actually 1.0e+30); any time step gives a numerically stable 
>> solution. If you use an elastic material with adaptive time stepping, 
>> the time step is 1.0e+30, so the number of time steps is being rounded 
>> off to 0. The output manager currently uses the number of time steps 
>> as a flag to decide whether it is writing "info" or "data". If the 
>> number of time steps is 0, it thinks it is writing "info".
>>
>> The adaptive time stepping correctly writes "data" files when using a 
>> Maxwell model. See examples/3d/hex8/savageprescott.cfg.
>>
>> Brad
>>
>>
>> Charles Williams wrote:
>>> At present, the automatic time stepping is controlled purely by the 
>>> Maxwell time (for viscoelastic materials).  Since the current 
>>> viscoelastic materials are all linear, the time step size is 
>>> independent of the deformation rate.  Once we include nonlinear 
>>> materials (e.g, power-law rheology), this will no longer be true.
>>> We will have to look at the second thing you mention.  I don't think 
>>> we had noticed that before.
>>> Thanks,
>>> Charles
>>> On 12/05/2009, at 3:55 PM, Christopher DiCaprio wrote:
>>>> Hi,
>>>>
>>>> It's not clear to me what is supported for the nonuniform, 
>>>> automatic  time stepping. The release notes for 1.3.0 say "For the 
>>>> current bulk  constitutive models, the automatically determined time 
>>>> step is  independent of the deformation rate, so it is uniform."
>>>>
>>>> The release notes for 1.3.1 don't mention time stepping, but the  
>>>> manual implies that there is support for non-uniform time steps.
>>>>
>>>> One other thing: when using nonuniform, automatic time stepping, 
>>>> the  *_info.vtk files are overwritten with what should go in the  
>>>> *_t0000.vtk files. This doesn't happen when using user-specified 
>>>> time  stepping.
>>>>
>>>> Thanks,
>>>> Chris
>>>>
>>>>
>>>> _______________________________________________
>>>> CIG-SHORT mailing list
>>>> CIG-SHORT at geodynamics.org <mailto:CIG-SHORT at geodynamics.org>
>>>> http://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 <mailto:C.Williams at gns.cri.nz>
>>> NOTE NEW E-MAIL ADDRESS
>>> ------------------------------------------------------------------------
>>> _______________________________________________
>>> CIG-SHORT mailing list
>>> CIG-SHORT at geodynamics.org
>>> http://geodynamics.org/cgi-bin/mailman/listinfo/cig-short
>>
> 
>