[aspect-devel] Compositional fields with different densities and viscosities

[Wolfgang Bangerth]

> I did run into a problem with a model considering 8 compositional fields; it
> got stopped because it was exceeding the allowed virtual and physical memory
> usage.
> I restarted the same run, but with a CLF number of 1.0e-6. Hence, the number
> of elements, degrees of freedom, global mass etc is constant. However, the
> memory usage keeps increasing (see snapshot).
> A similar increase is seen for a high resolution, uniform and constant grid
> Rayleigh-Taylor instability using 1 compositional field and material model
> simple.cc, only then the usage doesn't exceed the limit.
> Do you have an idea about the cause of this memory runaway?

Interesting. This looks like a memory leak somewhere. This is typically 
difficult to find :-( How many processors did you run this with and with how 
many degrees of freedom (see the statistics file)?

Can you come up with a minimal setup that shows the increase (but likely at a 
level that doesn't trigger the termination)? E.g., an input file for the RT 
instability. If you could send this to us, we can take a look but I'd love to 
have an as small as possible testcase first.

Thanks
  Wolfgang

-- 
------------------------------------------------------------------------
Wolfgang Bangerth               email:            bangerth at math.tamu.edu
                                 www: http://www.math.tamu.edu/~bangerth/