Hi Blechta,
Thanks for your reply and I had resolved my boundary problem. But now, I
meet another problem. There is a item (y, z) in my weak form, where y and z are
2 order tensor. When code runs,
error “shape mismatch” appears. I gauss this error caused by define function
space Vy in a wrong way. I look through all demos and don’t find answer to
this error. I want to know how to define function space for tensor variable.
Could you tell me how to do it or give me some hints? Thanks.
# Define function spaces
Vy = VectorFunctionSpace(mesh, "CG", 1)
Vr = FunctionSpace(mesh, "CG", 2)
Vp = FunctionSpace(mesh, "CG", 1)
Vu = FunctionSpace(mesh, "CG", 2)
Vqp = VectorFunctionSpace(mesh, "CG", 2)
W = MixedFunctionSpace([Vqp, Vu, Vp, Vr, Vy])
# Define boundary conditions
u0 = Constant((0.0,0.0))
u1 = Constant(0.0)
bc0 = DirichletBC(W.sub(0), u0, "on_boundary")
bc1 = DirichletBC(W.sub(1), u1, "on_boundary")
# Collect boundary conditions
bcs = [bc0, bc1]
# Define variational problem
(p1,u, p, r, y) = TrialFunctions(W)
(eta, v, q, s, z) = TestFunctions(W)
f = Constant(0.0)
a = (inner(y, z) - inner(grad(p1),z) - inner(grad(s), grad(u))
+inner(grad(s), p1)+ q*rot(p1)-inner(grad(r), grad(v))
-inner(y, grad(eta))+inner(grad(r), eta)+ p*rot(eta))*dx
L = f*v*dx
> On Nov 9, 2015, at 8:50 PM, Jan Blechta <[email protected]> wrote:
>
> On Sat, 7 Nov 2015 15:02:06 +0800
> zheng li <[email protected]> wrote:
>
>> Dear developer,
>>
>> I am using FEniCS to do mixed finite element case. There are
>> five element spaces in my case and the boundary conditions for space
>> Vqp and Vu are homogeneous Dirichlet condition. I define the BCs in
>> my code like following way:
>>
>> # Define function spaces
>> Vqp = VectorFunctionSpace(mesh, "CG", 2)
>> Vu = FunctionSpace(mesh, "CG", 2)
>> Vp = VectorFunctionSpace(mesh, "CG", 1)
>> Vr = FunctionSpace(mesh, "CG", 2)
>> Vy = VectorFunctionSpace(mesh, "CG", 1)
>>
>> W = Vqp * Vu * Vp * Vr * Vy
>
> Here's the problem. In Python FunctionSpace.__mul__ operator cannot be
> flattened so W is not what you expect but
>
> W = (((( Vqp * Vu ) * Vp ) * Vr ) * Vy )
>
> You should do rather
>
> W = MixedFunctionSpace([Vqp, Vu, Vp, Vr, Vy])
>
> This will work with DOLFIN <= 1.6. But MixedFunctionSpace is deprecated
> from future release of DOLFIN 1.7.0, see
> http://fenicsproject.org/documentation/dolfin/dev/python/demo/documented/stokes-iterative/python/documentation.html
> for a correct solution.
>
> Jan
>
>>
>> # Define boundary conditions
>> u0 = Constant((0.0,0.0))
>> u1 = Constant(0.0)
>> bc0 = DirichletBC(W.sub(0), u0, "on_boundary")
>> bc1 = DirichletBC(W.sub(1), u1, "on_boundary")
>>
>> # Collect boundary conditions
>> bcs = [bc0, bc1]
>>
>> However, error occurred when my code is running. The detail error
>> infos are
>>
>> Traceback (most recent call last):
>> File "demo_stokes-taylorhood.py", line 49, in <module>
>> bc0 = DirichletBC(W.sub(0), u0, "on_boundary")
>> File
>> "/Applications/FEniCS.app/Contents/Resources/lib/python2.7/site-packages/dolfin/fem/bcs.py",
>> line 128, in __init__ cpp.DirichletBC.__init__(self, *args) File
>> "/Applications/FEniCS.app/Contents/Resources/lib/python2.7/site-packages/dolfin/cpp/fem.py",
>> line 1140, in __init__
>> _fem.DirichletBC_swiginit(self,_fem.new_DirichletBC(*args))
>> RuntimeError:
>>
>> ***
>> -------------------------------------------------------------------------
>> *** DOLFIN encountered an error. If you are not able to resolve this
>> issue *** using the information listed below, you can ask for help at
>> *** *** [email protected]
>> ***
>> *** Remember to include the error message listed below and, if
>> possible, *** include a *minimal* running example to reproduce the
>> error. ***
>> ***
>> -------------------------------------------------------------------------
>> *** Error: Unable to create Dirichlet boundary condition. ***
>> Reason: Illegal value dimension (2), expecting (6). *** Where:
>> This error was encountered inside DirichletBC.cpp. *** Process:
>> unknown
>>
>> I don’t know what caused this. Is there any wrong with my definition
>> of BCs? Looking forward to your reply. Thanks a lot.
>>
>>
>> Best wishes,
>>
>>
>> —Zheng
>>
>>
>>
>
_______________________________________________
fenics mailing list
[email protected]
http://fenicsproject.org/mailman/listinfo/fenics
