Hi,
I’m trying to assemble a variational form only over the vertices. I think that the integration measure dP could do the job, however, I recieve an error message when using it. Here is a minimal working example with the output of the code and the output I would like to have. Is there a way of doing that? or maybe a workaround?. I was also wondering if its possible to modify the Petsc Matrix assembled and insert this last line “manually”.
from fenics import *
N = 4
mesh = UnitIntervalMesh(N)
V = FiniteElement("CG", mesh.ufl_cell(), 1)
R = FiniteElement('R', mesh.ufl_cell(), 0)
W = FunctionSpace(mesh, MixedElement([V, R]))
u, c = TrialFunction(W)
v, d = TestFunctions(W)
w = Function(W)
u_w, c_w = split(w)
for i in range(N+2):
w.vector()[i] = i+1
#a = d*u_w*u*dP + d*c_w*c*dP Doesn't work!
a = d*u_w*u*dx + d*c_w*c*dx
A = assemble(a)
print(A.array())
Output:
[[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0.16666667 0.5 0.75 1. 0.58333333 6. ]]
What a would like:
[[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[0. 0. 0. 0. 0. 0. ]
[1. 2. 3. 4. 5. 6. ]]
What do you mean by assemble only over the vertices? Do you mean that the assembly should be a point evaluation at each vertex?
I do not think this is supported with Real function spaces, as the error message states (Please attach that next time)
Traceback (most recent call last):
File "/root/shared/mwe123.py", line 23, in <module>
A = assemble(a)
File "/usr/lib/python3/dist-packages/dolfin/fem/assembling.py", line 217, in assemble
assembler.assemble(tensor, dolfin_form)
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
***
*** fenics-support@googlegroups.com
***
*** Remember to include the error message listed below and, if possible,
*** include a *minimal* running example to reproduce the error.
***
*** -------------------------------------------------------------------------
*** Error: Unable to assemble form over vertices.
*** Reason: Expecting test and trial spaces to only have dofs on vertices for point integrals.
*** Where: This error was encountered inside Assembler.cpp.
*** Process: 0
***
*** DOLFIN version: 2019.2.0.dev0
*** Git changeset: bd54183ed40f3597fe1187499c79c54eb4759f6d
*** -------------------------------------------------------------------------
You can see that it works with Lagrange 1 with:
from fenics import *
N = 4
mesh = UnitIntervalMesh(N)
V = FiniteElement("CG", mesh.ufl_cell(), 1)
W = FunctionSpace(mesh, V)
u = TrialFunction(W)
v = TestFunction(W)
w = Function(W)
for i in range(N+1):
w.vector()[i] = i+1
dxP = dx(metadata={"quadrature_scheme":"vertex"})
a = w*u*v*dP
A = assemble(a)
print(A.array())
yielding
[[1. 0. 0. 0. 0.]
[0. 2. 0. 0. 0.]
[0. 0. 3. 0. 0.]
[0. 0. 0. 4. 0.]
[0. 0. 0. 0. 5.]]
while
from fenics import *
N = 4
mesh = UnitIntervalMesh(N)
V = FiniteElement("Real", mesh.ufl_cell(), 0)
W = FunctionSpace(mesh, V)
u = TrialFunction(W)
v = TestFunction(W)
w = Function(W)
for i in range(1):
w.vector()[i] = i+1
dxP = dx(metadata={"quadrature_scheme":"vertex"})
a = w*u*v*dP
A = assemble(a)
print(A.array())
yields the error above.
Hi, @dokken. Thank you for your response!
The idea is to use the real space to solve an algebraic equation coupled to a differential equation. This algebraic equation is only a function of the vertices values. In this sense, I want the assemble to be a point evaluation at each vertex, as you mentioned.
As a workaround, I was trying to modify the Petsc Matrix. Since I’m not very familiar with it, I was not very successful.