[CIG-ALL] PyLith v1.4.2

[Brad Aagaard]

Greetings,

I am pleased to announce the release of PyLith 1.4.2, 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 v1.4.0 and v1.4.1. The
changes include (1) parameters in the power-law rheology are now
consistent with laboratory experiments, (2) errors in the
implementation of point forces have been fixed, (3) output of slip was
corrected to remove integration in time, (4) the stable time step for
implicit time-stepping with linear viscoelastic models has been
increased from 0.1 times the Maxwell time to 0.2 times the Maxwell
time based on accuracy tests, (5) the preconditioner is only
recomputed when necessary, (6) some unnecessary allocation and
deallocations were removed through the use of persistent data
structures, and (7) several memory leaks associated with output and
running in parallel have been eliminated.

We strongly recommend all users of previous PyLith releases to switch
to this latest release. Switching from v1.4.0 or v1.4.1 to v1.4.2 only
requires changes to parameter files when using the power-law rheology or
adaptive time stepping with a non-default stability factor (see the
manual and release notes below for details). See the README file for
changes required to switch to the v1.4.x release series from previous
versions.

You can download the source code and binaries from

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

Detailed installation instructions are in the User Manual with example
installation procedures for a few platforms in the bundled doc/install
directory.


RELEASE NOTES

  * Bug fixes

   - Parameters for the power-law rheology are now consistent with the
     parameters often reported in laboratory experiments. This should
     make it easier to use values from laboratory experiments. See the
     user manual for details.

   - Numerous errors in the Python and SWIG point source implementation
     were fixed. Point forces did not work in v1.4.0 and v1.4.1.

   - Output of slip on faults was actually the integral of slip in time
     in v1.4.0 and v1.4.1. This erroneous integration was removed.

   - Based on accuracy tests the stable time step for implicit
     time-stepping for the Maxwell viscoelastic models has been
     increased to 0.2 times the Maxwell time. The default stability
     factor in the adaptive time-stepping has been corresponding
     adjusted to provide a time step of 0.1 times the Maxwell time.

   - The preconditioner for the linear solver is only recomputed when
     the sparse matrix has been changed. This reduces runtime and
     unnecessary creation and destruction of the sparse matrix.

   - Several memory leaks associated with output and running in
     parallel have been fixed.