Solver method for LUSolver for Epetra backends fails

[Johan Hake]

The following modified test script fails as the Epetra LUSolver does not get properly instantiated:

from dolfin import *

parameters.linear_algebra_backend = "Epetra"

# Create mesh and define function space
mesh = UnitSquare(32, 32)
V = FunctionSpace(mesh, "Lagrange", 1)

# Define Dirichlet boundary (x = 0 or x = 1)
def boundary(x):
    return x[0] < DOLFIN_EPS or x[0] > 1.0 - DOLFIN_EPS

# Define boundary condition
u0 = Constant(0.0)
bc = DirichletBC(V, u0, boundary)

# Define variational problem
u = TrialFunction(V)
v = TestFunction(V)
f = Expression("10*exp(-(pow(x[0] - 0.5, 2) + pow(x[1] - 0.5, 2)) / 0.02)")
g = Expression("sin(5*x[0])")
a = inner(grad(u), grad(v))*dx
L = f*v*dx + g*v*ds

A = assemble(a)
b = assemble(L)

bc.apply(A, b)

# Both fails!
solver = LUSolver("mumps")
#solver = LUSolver("umfpack")

# Default works
#solver = LUSolver()

# Compute solution
u = Function(V)
solver.solve(A, u.vector(), b)

################
We need a unit test that iterate over all backends and possible solvers and try to solve a problem using the allowed options.

[Anders Logg (logg)]

The initialization of the solver object is now checked and an error message is displayed. Your problem is (I think) that your Epetra installation does not have UMFPACK. This bug is superseded by bug #891350.

[Anders Logg (logg)]

Fixed in 1.0.x-logg

[Johan Hake (johan-hake)]

On Wednesday November 16 2011 13:06:35 Anders Logg wrote:
> The initialization of the solver object is now checked and an error
> message is displayed. Your problem is (I think) that your Epetra
> installation does not have UMFPACK. This bug is superseded by bug
> #891350.

Sounds plausible! I am just using the default dorsal installation of trilinos.
Maybe we should add:

  -D TPL_ENABLE_UMFPACK:BOOL=ON

to trilinos.package in dorsal?

Johan

[Anders Logg (logg)]

On Thu, Nov 17, 2011 at 05:40:02AM -0000, Johan Hake wrote:
> On Wednesday November 16 2011 13:06:35 Anders Logg wrote:
> > The initialization of the solver object is now checked and an error
> > message is displayed. Your problem is (I think) that your Epetra
> > installation does not have UMFPACK. This bug is superseded by bug
> > #891350.
>
> Sounds plausible! I am just using the default dorsal installation of trilinos.
> Maybe we should add:
>
> -D TPL_ENABLE_UMFPACK:BOOL=ON
>
> to trilinos.package in dorsal?

Sounds good to me.

What does Harish say?

--
Anders

[Harish Narayanan (hnarayanan)]

On 11/17/11 10:09 AM, Anders Logg wrote:
> On Thu, Nov 17, 2011 at 05:40:02AM -0000, Johan Hake wrote:
>> On Wednesday November 16 2011 13:06:35 Anders Logg wrote:
>>> The initialization of the solver object is now checked and an error
>>> message is displayed. Your problem is (I think) that your Epetra
>>> installation does not have UMFPACK. This bug is superseded by bug
>>> #891350.
>>
>> Sounds plausible! I am just using the default dorsal installation of trilinos.
>> Maybe we should add:
>>
>> -D TPL_ENABLE_UMFPACK:BOOL=ON
>>
>> to trilinos.package in dorsal?
>
> Sounds good to me.
>
> What does Harish say?

I'll add it.

Harish