VariationalProblem(J, F, bc, nonlinear=True) do not work

Anne,

I think you should remove the nonlinear keyword.

Kent

On 3 May 2011 16:23, Anne Kværnø <email address hidden>wrote:

> New question #155708 on DOLFIN:
> https://answers.launchpad.net/dolfin/+question/155708
>
> The test program nlPoisson_vp2.py do not work. I got the following error
> message
>
> Traceback (most recent call last):
> File "nlPoisson_vp2.py", line 55, in <module>
> problem = VariationalProblem(J, F, bc, nonlinear=True)
> TypeError: __init__() got an unexpected keyword argument 'nonlinear'
>
> I am using Ubuntu 10.10, python 2.6.6 and dolfin 0.9.10
>
> On a different computer, using Ubuntu 10.04, python 2.6.5 and dolfin 0.9.9,
> this test program worked perfectly fine.
>
>
> --
> You received this question notification because you are a member of
> DOLFIN Team, which is an answer contact for DOLFIN.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~dolfin
> Post to : <email address hidden>
> Unsubscribe : https://launchpad.net/~dolfin
> More help : https://help.launchpad.net/ListHelp
>

On 03/05/11 17:15, Johan Hake wrote:
> Question #155708 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/155708
>
> Johan Hake proposed the following answer:
> On Tuesday May 3 2011 07:23:20 Anne Kværnø wrote:
>> New question #155708 on DOLFIN:
>> https://answers.launchpad.net/dolfin/+question/155708
>>
>> The test program nlPoisson_vp2.py do not work. I got the following error
>> message
>>
>> Traceback (most recent call last):
>> File "nlPoisson_vp2.py", line 55, in <module>
>> problem = VariationalProblem(J, F, bc, nonlinear=True)
>> TypeError: __init__() got an unexpected keyword argument 'nonlinear'
>>
>> I am using Ubuntu 10.10, python 2.6.6 and dolfin 0.9.10
>>
>> On a different computer, using Ubuntu 10.04, python 2.6.5 and dolfin 0.9.9,
>> this test program worked perfectly fine.
>
> The interface to a nonlinear VariationalProblem changed between these two
> releases. Take a look at the docstring of VariationalProblem.
>
> In essence: Just change the order of the forms. Put the nonlinear (linear in
> the test function) form first, and the linearised Jacobian as the second form.
> Then will the VariationalProblem be interpreted as a Nonlinear problem. Also
> skip the "nonlinear" argument.
>
> If you ask me, I think this change is abit too magic and it is not clear at
> all that this is what to do ;)
>

I agree. I think that it's very unclear.

Why not just have

    LinearVariationalProblem

and

    NonlinearVariationalProblem

They could share common code behind the scenes.

Garth

> Johan
>

After reading Johan's post I was able to get past the nonlinear problem but now my program crashes. Could anyone help me with this?

Here is my program:

#!/usr/bin/python

from dolfin import *
import numpy, sys

# MESHING
mesh = Rectangle(0.0, -0.5, 1.0, 0.5, 10, 10, "right/left")

# FUNCTION SPACE
scalar = FunctionSpace(mesh, "CG", 1) #Pressure
vector = VectorFunctionSpace(mesh, "CG", 2) #Velocity
system = vector * scalar #Mixed Function Space

# BOUNDARIES
right = compile_subdomains('x[0] == 1.0')
left = compile_subdomains('x[0] == 0.0')
top = compile_subdomains('x[1] == 0.5')
bottom = compile_subdomains('x[1] == -0.5')

# INFLOW VELOCITY BC FOR TOP
bc1 = DirichletBC(system.sub(0), Constant((1.0, 0.0)), top)

# NO-SLIP BC FOR TOP-BOTTOM
noslip = Constant((0.0, 0.0))
bc0 = DirichletBC(system.sub(0), noslip, left)
bc3 = DirichletBC(system.sub(0), noslip, bottom)
bc2 = DirichletBC(system.sub(0), noslip, right)

# STORAGE OF BCs
bcs = [bc1, bc2, bc0, bc3]

# VARIATIONAL PROBLEM
ts = TestFunction(system)
ds = TrialFunction(system)
s = Function(system)

v, q = split(ts)
dv, dq = split(ds)
u , p = split(s)

f = Constant((0, 0))
nu = 0.001

#F = (nu*inner(grad(v), grad(u)) - div(v)*p + q*div(u) - inner(v,f))*dx # Stokes
F = (nu*inner(grad(v), grad(u)) - div(v)*p + q*div(u) + inner(dot(u,grad(u)),v) - inner(v,f))*dx # Navier-Stokes
J = derivative(F,s,ds)

# SET UP PDE
problem = VariationalProblem(F, J, bcs)
#problem.is_nonlinear=True

# SOLVE PDE
(U, P) = problem.solve().split()

# FILES FOR PARAVIEW
ufile_pvd = File("velocity.pvd")
ufile_pvd << U
pfile_pvd = File("pressure.pvd")
pfile_pvd << P

# PLOTS
plot(U)
plot(P)
interactive()

On 03/05/11 17:12, Kent-Andre Mardal wrote:
>
> Anne,
>
> I think you should remove the nonlinear keyword.
>

This change will make the code work, but it will produce an incorrect
result. DOLFIN will assume that the equation is linear, based on the
order of J and F (which as I've already posted I think is bad).

Garth

> Kent
>
> On 3 May 2011 16:23, Anne Kværnø <<email address hidden>
> <mailto:<email address hidden>>> wrote:
>
> New question #155708 on DOLFIN:
> https://answers.launchpad.net/dolfin/+question/155708
>
> The test program nlPoisson_vp2.py do not work. I got the following
> error message
>
> Traceback (most recent call last):
> File "nlPoisson_vp2.py", line 55, in <module>
> problem = VariationalProblem(J, F, bc, nonlinear=True)
> TypeError: __init__() got an unexpected keyword argument 'nonlinear'
>
> I am using Ubuntu 10.10, python 2.6.6 and dolfin 0.9.10
>
> On a different computer, using Ubuntu 10.04, python 2.6.5 and dolfin
> 0.9.9, this test program worked perfectly fine.
>
>
> --
> You received this question notification because you are a member of
> DOLFIN Team, which is an answer contact for DOLFIN.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~dolfin
> <https://launchpad.net/%7Edolfin>
> Post to : <email address hidden>
> <mailto:<email address hidden>>
> Unsubscribe : https://launchpad.net/~dolfin
> <https://launchpad.net/%7Edolfin>
> More help : https://help.launchpad.net/ListHelp
>
>
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~dolfin
> Post to : <email address hidden>
> Unsubscribe : https://launchpad.net/~dolfin
> More help : https://help.launchpad.net/ListHelp

Kyle,
you will get faster answers if you post
the error messages you get in addition
to the program. :)

Martin

On 3 May 2011 18:27, Kyle <email address hidden> wrote:
> Question #155708 on DOLFIN changed:
> https://answers.launchpad.net/dolfin/+question/155708
>
> Kyle requested more information:
> After reading Johan's post I was able to get past the nonlinear problem
> but now my program crashes. Could anyone help me with this?
>
> Here is my program:
>
> #!/usr/bin/python
>
> from dolfin import *
> import numpy, sys
>
> # MESHING
> mesh = Rectangle(0.0, -0.5, 1.0, 0.5, 10, 10, "right/left")
>
> # FUNCTION SPACE
> scalar = FunctionSpace(mesh, "CG", 1) #Pressure
> vector = VectorFunctionSpace(mesh, "CG", 2) #Velocity
> system = vector * scalar #Mixed Function Space
>
> # BOUNDARIES
> right = compile_subdomains('x[0] == 1.0')
> left = compile_subdomains('x[0] == 0.0')
> top = compile_subdomains('x[1] == 0.5')
> bottom = compile_subdomains('x[1] == -0.5')
>
> # INFLOW VELOCITY BC FOR TOP
> bc1 = DirichletBC(system.sub(0), Constant((1.0, 0.0)), top)
>
> # NO-SLIP BC FOR TOP-BOTTOM
> noslip = Constant((0.0, 0.0))
> bc0 = DirichletBC(system.sub(0), noslip, left)
> bc3 = DirichletBC(system.sub(0), noslip, bottom)
> bc2 = DirichletBC(system.sub(0), noslip, right)
>
> # STORAGE OF BCs
> bcs = [bc1, bc2, bc0, bc3]
>
> # VARIATIONAL PROBLEM
> ts = TestFunction(system)
> ds = TrialFunction(system)
> s = Function(system)
>
> v, q = split(ts)
> dv, dq = split(ds)
> u , p = split(s)
>
> f = Constant((0, 0))
> nu = 0.001
>
> #F = (nu*inner(grad(v), grad(u)) - div(v)*p + q*div(u) - inner(v,f))*dx # Stokes
> F = (nu*inner(grad(v), grad(u)) - div(v)*p + q*div(u) + inner(dot(u,grad(u)),v) - inner(v,f))*dx # Navier-Stokes
> J = derivative(F,s,ds)
>
> # SET UP PDE
> problem = VariationalProblem(F, J, bcs)
> #problem.is_nonlinear=True
>
> # SOLVE PDE
> (U, P) = problem.solve().split()
>
> # FILES FOR PARAVIEW
> ufile_pvd = File("velocity.pvd")
> ufile_pvd << U
> pfile_pvd = File("pressure.pvd")
> pfile_pvd << P
>
> # PLOTS
> plot(U)
> plot(P)
> interactive()
>
> --
> You received this question notification because you are a member of
> DOLFIN Team, which is an answer contact for DOLFIN.
>
> _______________________________________________
> Mailing list: https://launchpad.net/~dolfin
> Post to     : <email address hidden>
> Unsubscribe : https://launchpad.net/~dolfin
> More help   : https://help.launchpad.net/ListHelp
>

Thanks Johan Hake, that solved my question.

Thanks, this solved the problem.