[CIG-ALL] PyLith v1.6.3 bugfix release

[Brad Aagaard]

Greetings,

I am pleased to announce the release of PyLith 1.6.3, a finite-element
code designed to solve dynamic elastic problems and quasi-static
viscoelastic problems in tectonic deformation.

This release fixes several bugs in PyLith v1.6.2. We strongly
recommend all users of previous PyLith releases switch to this latest
release, especially anyone running in parallel or using fault friction.

You can download the source code and binaries from

     http://geodynamics.org/cig/software/packages/short/pylith

Detailed installation instructions for the binary packages are in the
User Manual with detailed building instructions for a few platforms
in the INSTALL file bundled with the PyLith Installer utility.


RELEASE NOTES

* Bug fixes

   - Improved error messages for problems encountered during processing
     of parameters. A backtrace of the object hierarchy is now included
     to pinpoint in which object the error occurred.

   - Added a line search to the inner friction solve in quasi-static
     simulations to increase the robustness of the nonlinear
     solve. Simulations using rate and state friction now converge
     under a much wider range of circumstances.

   - Fixed bug in updating slip state variable in slip-weakening
     friction. This caused slight errors in the cumulative slip. We
     also added a parameter that forces healing to occur in a single
     time step. This is used to confine slip to a single time step in
     quasi-static simulations. See examples/3d/hex8/step13.cfg for an
     example.

   - Tuned parameters in the slip-weakening friction and rate and state
     friction examples (step13.cfg and step14.cfg, respectively) in
     examples/3d/hex8 to give stick-slip behavior.

   - Fixed communication issue associated with writing boundary
     condition information output in parallel.

   - Changed info in Xdmf file for fields that are not scalars,
     vectors, or tensors so that the each component is extracted,
     facilitating visualization in ParaView. The corresponding HDF5
     file remains the same.

   - Added the ability to specify non-derived units (e.g., degree and
     radian). This is useful in specifying parameters for the
     Drucker-Prager elastoplastic rheologies. If no units are
     specified, radians are assumed.

* Internal changes

   - Rate and state friction with ageing law

     The implementation of rate and state friction with ageing law was
     modified to work better with the iterative solver. We switched to
     the conventional, unregularized formulation but added a minimum
     cutoff for the slip rate. Below this cutoff friction has a linear
     rather than logarithmic dependence on slip rate. As long as this
     cutoff is close to the SNES solver tolerance, the difference in
     behavior is negligible while improving the ability of the solver
     to converge for very small deformations.

KNOWN ISSUES

   The rate and state friction with ageing law has not been tested for
   dynamic rupture simulations. We plan to run the SCEC Dynamic Rupture
   benchmarks for rate and state friction as soon as we add a
   spatial-temporal specification of initial fault tractions, which are
   required for the benchmark problems.

   Running simulations with more than a million cells and large faults
   in parallel can result in severe memory imbalances among
   processors. Some processors around the fault may use 10x more memory
   than processors away from the fault. We expect this problem to
   disappear in v1.7 when we switch to new, more efficient Sieve
   implementation.