Here is a minimal script reproducing the bug Johan
On Saturday August 6 2011 00:14:40 Johan Hake wrote: > It looks like there is something going on with the assignment to an empty > EpetraMatrix. When the Epetra_FECrsMatrix::operator= was called on an > uninitialized object an assertion kicked in. Bug or feature on the Epetra > side? > > If it is possible we should probably initialize the Epetra_FECrsMatrix when > an empty EpetraMatrix is created hopefull avoiding this. > > For now I use the copy constructor of Epetra_FECrsMatrix that initialize > the structures correctly before it uses the Epetra_FECrsMatrix::operator=. > Then I just assign the newly copied Epetra_FECrsMatrix to the > EpetraMatrix. > > Johan > > On Friday August 5 2011 21:24:52 Johan Hake wrote: > > On Friday August 5 2011 13:26:52 Garth N. Wells wrote: > > > On Aug 5 2011, Johan Hake wrote: > > > >You are right. I will look at it this weekend. > > > > > > > > However, it "fixed" my problem with Epetra backend and Nonlinear > > > > problem where I used the assignment operator when generating the > > > > Jacobian. > > > > > > Can you post some simple code to test against? > > > > Attached is a slightly modified version of the Cahn-hillard demo. > > > > Johan > > > > > Garth > > > > > > >Johan > > > > > > > >On Aug 5, 2011, at 7:24, "Garth N. Wells" <gn...@cam.ac.uk> wrote: > > > >> This doesn't look right. Looks like a shallow copy. > > > >> > > > >> Garth > > > >> > > > >> On 04/08/11 00:02, nore...@launchpad.net wrote: > > > >>> ------------------------------------------------------------ > > > >>> revno: 6069 > > > >>> committer: Johan Hake <hake....@gmail.com> > > > >>> branch nick: work > > > >>> timestamp: Wed 2011-08-03 20:58:24 -0700 > > > >>> > > > >>> message: > > > >>> Fix a bug in the assignment operator of EpetraMatrix > > > >>> > > > >>> modified: > > > >>> dolfin/la/EpetraMatrix.cpp > > > >>> > > > >>> -- > > > >>> lp:dolfin > > > >>> https://code.launchpad.net/~dolfin-core/dolfin/main > > > >>> > > > >>> Your team DOLFIN Core Team is subscribed to branch lp:dolfin. To > > > >>> unsubscribe from this branch go to > > > >>> https://code.launchpad.net/~dolfin-core/dolfin/main/+edit-subscript > > > >>> io n > > > >> > > > >> _______________________________________________ > > > >> Mailing list: https://launchpad.net/~dolfin > > > >> Post to : dolfin@lists.launchpad.net > > > >> Unsubscribe : https://launchpad.net/~dolfin > > > >> More help : https://help.launchpad.net/ListHelp > > _______________________________________________ > Mailing list: https://launchpad.net/~dolfin > Post to : dolfin@lists.launchpad.net > Unsubscribe : https://launchpad.net/~dolfin > More help : https://help.launchpad.net/ListHelp
from dolfin import * parameters["linear_algebra_backend"] = "Epetra" A = Matrix() mesh = UnitSquare(2,2) V = FunctionSpace(mesh, "CG", 1) u, v = TrialFunction(V), TestFunction(V) A1 = assemble(u*v*dx) A.assign(A1)
_______________________________________________ Mailing list: https://launchpad.net/~dolfin Post to : dolfin@lists.launchpad.net Unsubscribe : https://launchpad.net/~dolfin More help : https://help.launchpad.net/ListHelp
