DOLFIN

  1. Bug #927539
  2. Comment #3

Comment 3 for bug 927539

Revision history for this message
Anders Logg (logg) wrote : Re: [Bug 927539] Re: Missing diagonals after subdomain assembly

On Mon, Feb 06, 2012 at 12:46:05PM -0000, Joachim Haga wrote:
> Your example doesn't make (syntactical) sense, but to be clear:

It should be

c = Expression("0.0")
A = assemble(c*dot(grad(u), grad(v))*dx)

> There is no problem with 0.0 on the diagonal. The problem is missing
> diagonal entries, ie. diagonals that are not present in the sparsity
> pattern. And it should be possible to ensure that this doesn't
> happen, no?

ok, now I see your point. Yes, it seems reasonable to ensure that we
always have an entry on the diagonal in the sparsity pattern. Possibly
in combination with a global option to turn it off.

--
Anders