ffc team mailing list archive

[Marie Rognes <meg@xxxxxxxxxxx>]

On Wed, Nov 07, 2007 at 07:23:37PM +0100, FFC wrote:

> > One or more new changesets pushed to the primary ffc repository.
> > A short summary of the last three changesets is included below.
> >
> > changeset:   1003:be7ba517e07b4c6a1a54fd324195cd18937d48f4
> > tag:         tip
> > user:        "Marie E. Rognes <meg@xxxxxxxxxxx>"
> > date:        Wed Nov 07 19:23:22 2007 +0100
> > files:       src/demo/VectorLaplaceGradCurl.form src/ffc/compiler/codegeneration/common/evaluatebasis.py src/ffc/compiler/codegeneration/common/evaluatebasisderivatives.py src/ffc/compiler/codegeneration/common/finiteelement.py src/ffc/compiler/language/algebra.py src/ffc/fem/finiteelement.py src/ffc/fem/mapping.py src/test/regression/reference/MixedPoisson.h src/test/regression/reference/VectorLaplaceGradCurl.h
> > description:
> > Added covariant mapping for Nedelec elements.
> >
> > Think Nedelec should work now, but relies on a fix in FIAT. Added test
> > in VectorLaplaceGradCurl to /src/demo.
> >
> > evaluatebasis, evaluatebasisderivatives and interpolate_vertex_values
> > updated but rather poorly tested.
> >     

Very poorly tested turns out to be more precise.

evaluatebasis etc needs to additionally declare the inverse of the 
Jacobian in the H(curl) (mapping == Mapping.COVARIANT_PIOLA) cases.

Not sure of the prettiest way of modifying the code here. Will look at 
it with fresh eyes tomorrow unless anyone has fixed it by then ;)

--
Marie E. Rognes
Ph.D Fellow, 
Centre of Mathematics for Applications, 
University of Oslo
http://folk.uio.no/meg