libMesh 0.6.4 is now available for download from
https://sourceforge.net/projects/libmesh/
For those of you using 0.6.4-rc1, the biggest changes are probably the
bugfixes to a UniformErrorEstimator regression and the new support for
SCALAR variables in parallel.
The biggest change that didn't make it into 0.6.4: although the
multiple-QoI support for adjoint solves is passing our first tests, I
didn't want to put code that is still in such a high state of flux
into a "stable" release. Anyone who wants to experiment with the new
adjoint support right away is advised to monitor libmesh-devel and use
the SVN head. I hope to have a libmesh 0.6.5 with a more expanded and
more stable adjoints API released within a few months.
Changes from 0.6.3-rc1 to 0.6.4 include:
API changes:
FEMSystem uses new FEMContext class for assemblies
More flexible ErrorEstimator::error_norm interface
Macro changes; e.g. error() -> libmesh_error()
PatchRecoveryErrorEstimator now defaults to H1 seminorm
Bugfixes in:
Hanging node handling in some parallel AMR cases
Pyramid element quadrature
Exodus I/O, VTK I/O
Parallel SparseMatrix::print()
BoundaryInfo
MeshData
".ele" extension handling
Load-imbalanced matrix creation
NewtonSolver convergence reporting
Efficiency improvements:
"Ghosted" vector support with PETSc
PetscVector speed improvements
Additional parallel Mesh code
Better parallel communication handling
New features:
Support for per-subdomain variables
ShellMatrix class and usage example
PETSc 3 support
Initial support for automatic adjoint-based parameter sensitivities,
goal-oriented adaptive refinement
More error estimator options
Support for extracting, performing computations on boundary meshes.
SLEPc 3, SLEPc deflation space support
Initial ALE support in FEMSystem
Support for globally supported SCALAR variable type
New Preconditioner class
Flag to allow reusing preconditioners with PETSc
More quadrature rules for monomials, gaussian types
Performance logging improvements
New Parallel:: and Threads:: utility methods
NumericVector::swap(), ::abs()
DenseVector::dot(), DenseMatrix::vector_mult(), ::get_transpose()
Beginning support for BLAS DenseMatrix operations
TypeTensor::tr()
SparseMatrix::zero_rows()
Exodus IO sideset, nodeset support
New Postscript output class for 2D
GMV pyramid output support
All-pyramid build_cube() option
More flexible partitioner interface
New BoundingBox class
Post-initialization creation of new System-supported vectors
System::zero_variable()
Per-variable discrete vector norms
More boundary id modification methods
More flexible string_to_enum inputs
New Triangle meshing code version
New Sphere constructor
libmesh_assert_cast: opt-mode static casts with debug-mode checking
libmesh_file_error: assert valid files even in opt mode
libmesh_convergence_failure: separate exception vs. logic errors
Error detection for bzip2/bunzip2 failures
Fortran compiler autodetection for user apps
------------------------------------------------------------------------------
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
