dolfin team mailing list archive

[DOLFIN <dolfin@xxxxxxxxxx>]

One or more new changesets pushed to the primary DOLFIN repository.
A short summary of the last three changesets is included below.

changeset:   1858:1c1dee8860f0b7528f3c176712fff630db7fb8d7
tag:         tip
parent:      1857:086f9ccc608960d2987290807f18db8b490067e2
parent:      1856:aea74fea41ea6daec697aa4965e8eaa854b06e5d
user:        "Garth N. Wells <g.n.wells@xxxxxxxxxx>"
date:        Wed Apr 12 13:34:40 2006 +0200
files:       
description:
merge


changeset:   1857:086f9ccc608960d2987290807f18db8b490067e2
parent:      1855:d48a6c9ed162d2a85f9fa468889258258fa0cb51
user:        "Garth N. Wells <g.n.wells@xxxxxxxxxx>"
date:        Wed Apr 12 13:33:26 2006 +0200
files:       src/kernel/la/DenseMatrix.cpp
description:
Add #include <boost/numeric/ublas/operation.hpp> to DenseMatrix.cpp


changeset:   1856:aea74fea41ea6daec697aa4965e8eaa854b06e5d
user:        "Anders Logg <logg@xxxxxxxxx>"
date:        Wed Apr 12 07:07:07 2006 -0500
files:       src/demo/ode/reaction/bench.log src/demo/ode/reaction/domain.log src/kernel/la/KrylovSolver.cpp src/kernel/ode/MultiAdaptiveNewtonSolver.cpp
description:
Initial fixes for ODE benchmark:

Results are now similar (close enough) to the results with
DOLFIN 0.5.12. A few default parameters were changed in the
Krylov solvers in addition to moving to PETSc 2.3.1 which
affected the multi-adaptive Newton solver. Scaling the system
with the size of the residual before solving and rescaling
after in each Newton step fixes the problem. (Scaling only
done locally in ODE solvers, not in Krylov solver.)

Next step is to clean up benchmarks and generate results.


-------------------------------------------------------
For more details, visit http://www.fenics.org/hg/dolfin