dolfin team mailing list archive
-
dolfin team
-
Mailing list archive
-
Message #02829
[HG] Initial work to prepare for porting of ODE solvers to uBlas.
One or more new changesets pushed to the primary DOLFIN repository.
A short summary of the last three changesets is included below.
changeset: 2034:a5328ce357611ce6c8df269c208fd4827844d22d
tag: tip
user: "Anders Logg <logg@xxxxxxxxx>"
date: Tue Jul 04 19:51:27 2006 +0200
files: src/bench/ode/reaction/bench.log src/kernel/ode/MonoAdaptiveFixedPointSolver.cpp src/kernel/ode/MonoAdaptiveNewtonSolver.cpp src/kernel/ode/MonoAdaptiveTimeSlab.cpp src/kernel/ode/MultiAdaptiveTimeSlab.cpp src/kernel/ode/TimeSlab.cpp src/kernel/ode/dolfin/MonoAdaptiveTimeSlab.h src/kernel/ode/dolfin/MultiAdaptiveTimeSlab.h src/kernel/ode/dolfin/TimeSlab.h
description:
Initial work to prepare for porting of ODE solvers to uBlas.
The solvers now rely less on working with offsets on large
arrays. Data is now copied before and after the call to f().
The extra cost seems to be neglibible.
changeset: 2033:f803317cded1e2f71fb98ddc7701070a8d87232e
user: "Anders Logg <logg@xxxxxxxxx>"
date: Tue Jul 04 14:59:30 2006 +0200
files: src/bench/ode/reaction/bench.log src/demo/ode/homotopy/simple/main.cpp src/demo/ode/stiff/run-tests.sh src/kernel/ode/Homotopy.cpp
description:
Fig bug in homotopy solver: old names used for parameters
changeset: 2032:02ee0dc9785831a77ae0e9203ecbc8286365bd3e
user: "Anders Logg <logg@xxxxxxxxx>"
date: Tue Jul 04 11:13:55 2006 +0200
files: src/kernel/la/DenseVector.cpp src/kernel/la/Makefile.am src/kernel/la/Makefile.in src/kernel/la/dolfin/DenseVector.h src/kernel/la/dolfin/Makefile.am src/kernel/la/dolfin/Makefile.in src/kernel/la/dolfin/dolfin_la.h src/kernel/la/dolfin/uBlasDummyPreconditioner.h src/kernel/la/dolfin/uBlasILUPreconditioner.h src/kernel/la/dolfin/uBlasKrylovSolver.h src/kernel/la/dolfin/uBlasPreconditioner.h src/kernel/la/uBlasDummyPreconditioner.cpp src/kernel/la/uBlasILUPreconditioner.cpp src/kernel/la/uBlasKrylovSolver.cpp src/test/main.cpp
description:
Move uBlas ILU preconditioner to separate class and add
general interface for uBlas preconditioners. Seems to be just
as fast as before (perhaps marginally slower, see separe post).
-------------------------------------------------------
For more details, visit http://www.fenics.org/hg/dolfin