ffc team mailing list archive

[ffc@xxxxxxxxxx (FFC)]

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

changeset:   1446:054927ef17e140fa32fbaa2a91efdcfa0043ad79
tag:         tip
user:        Anders Logg <logg@xxxxxxxxx>
date:        Thu Mar 19 23:27:57 2009 +0100
files:       ffc/compiler/representation/tensor/monomialtransformation.py
description:
Bug fix for handling float_value in monomial representation. Now down to
just two nonworking demos for tensor representation:

  MixedPoisson
  VectorLaplaceGradCurl


changeset:   1445:61a664524bf3543bf3b5719d00a5c1f7e20d1fe9
user:        Anders Logg <logg@xxxxxxxxx>
date:        Thu Mar 19 21:14:53 2009 +0100
files:       ffc/compiler/codegeneration/tensor/ufltensorgenerator.py ffc/compiler/format/ufcformat.py ffc/compiler/representation/tensor/ufltensorrepresentation.py
description:
Bug fix for tensor representation. All forms now compile, both with
FFC and GCC. The following integrals are currently giving incorrect
results:

Quadrature:

EnergyNormFunctional_cell_integral_0:                         *** (diff = 0.0004636)
EquationLinearForm_cell_integral_0:                           *** (diff = 0.00296852)
FunctionOperatorsBilinearForm_cell_integral_0:                *** (diff = 0.010091)
HeatLinearForm_cell_integral_0:                               *** (diff = 0.0033219)
MassBilinearForm_cell_integral_0:                             *** (diff = 1e-09)
NeumannProblemLinearForm_cell_integral_0:                     *** (diff = 0.0029685)
OptimizationLinearForm_cell_integral_0:                       *** (diff = 0.00170424)
PoissonDGLinearForm_cell_integral_0:                          *** (diff = 0.0029685)
PoissonSystemLinearForm_cell_integral_0:                      *** (diff = 0.0029685)
PoissonLinearForm_cell_integral_0:                            *** (diff = 0.0029685)
QuadratureElementBilinearForm_cell_integral_0:                *** (diff = 0.249892)
QuadratureElementLinearForm_cell_integral_0:                  *** (diff = 0.10106)
StokesLinearForm_cell_integral_0:                             *** (diff = 0.00140453)
VectorLaplaceGradCurlLinearForm_cell_integral_0:              *** (diff = 0.00206928)

Tensor:

ElasticityBilinearForm_cell_integral_0:                   *** (diff = 0.482675)
EquationBilinearForm_cell_integral_0:                     *** (diff = 0.757096)
EquationLinearForm_cell_integral_0:                       *** (diff = 0.332041)
MixedPoissonBilinearForm_cell_integral_0:                 *** (diff = 1.23)
StokesBilinearForm_cell_integral_0:                       *** (diff = 0.286666)
SubDomainsBilinearForm_cell_integral_1:                   *** (diff = 0.024654)
VectorLaplaceGradCurlBilinearForm_cell_integral_0:        *** (diff = 3.44101)
VectorLaplaceGradCurlLinearForm_cell_integral_0:          *** (diff = 0.00226017)


changeset:   1444:a9d8b074951c1f6539240e147cc0db0c9b3b3830
user:        Anders Logg <logg@xxxxxxxxx>
date:        Thu Mar 19 20:38:47 2009 +0100
files:       ffc/compiler/representation/tensor/uflmonomialintegration.py test/simple_verify_tensors/test.py
description:
Add list of forms for which tensor representation is not expected to
work for new UFL-based tensor representation:

  FunctionOperators
  QuadratureElement
  TensorWeightedPoisson

All forms that should now compile with tensor representation but
some generated code still won't compile with GCC or give correct
results.

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