dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

kent-and@xxxxxxxxx wrote:
> > On Thu, Nov 20, 2008 at 06:46:23PM +0100, kent-and@xxxxxxxxx wrote:
> > > > On Thu, Nov 20, 2008 at 05:24:33PM +0000, Garth N. Wells wrote:
> > > > > DOLFIN wrote:
> > > > > > One or more new changesets pushed to the primary dolfin repository.
> > > > > > A short summary of the last three changesets is included below.
> > > > > >
> > > > > > changeset:   5192:cae1753335d21eddc205e16ed3ac3cef16462454
> > > > > > tag:         tip
> > > > > > user:        Anders Logg <logg@xxxxxxxxx>
> > > > > > date:        Thu Nov 20 18:22:47 2008 +0100
> > > > > > files:       site-packages/dolfin/pde.py
> > > > > > description:
> > > > > > Add note about not using assemble_system until we have worked out a
> > > > > bug
> > > > > > with integration over exterior facets.
> > > > > Is this also a C++ issue?
> > > > >
> > > > > Garth
> > > > I haven't tried but I suspect it is.
> > > >
> > > > The error can be tested easily with the following script which is a
> > > > version of a script Harish just sent me:
> > > >
> > > >
> > > > from dolfin import *
> > > >
> > > > mesh = UnitSquare(2, 2)
> > > > element = FiniteElement("Lagrange", "triangle", 1)
> > > >
> > > > v = TestFunction(element)
> > > > u = TrialFunction(element)
> > > > n = FacetNormal("triangle", mesh)
> > > >
> > > > a = v*u*dx
> > > > L = v*dot(n, n)*ds
> > > >
> > > > # Does not work (b will be 0)
> > > > (A, b) = assemble_system(a, L, [], mesh)
> > > > b.disp()
> > > >
> > > > # Works (b will be nonzero)
> > > > b = assemble(L, mesh)
> > > > b.disp()
> > > b is not zero for me (dolfin-0.8.1).
> > >
> > > Kent
> > Are both bs nonzero? Or just the last?
> >
> > --
> > Anders
>
> Sorry, just the last. I see DG is broken as well. I probably need to
> have a look at this function and break it up.

I think that the development version works for DG. I spent some time 
fixing it.

I was tempted to break up the function, but since it's faster than the 
standard assembly we should code a broken up version next to the 
monolithic function so that we can benchmark it.

Garth

> Kent
>
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev