Good spot, this is the split(foo) vs foo.split() "gotcha".

split(foo) is supposed to give symbolic UFL objects that can be used in forms:

U = FunctionSpace(...); V = FunctionSpace(...)
W = U*V
urho = TrialFunction(W)
u, rho = split(urho)  # could have done u, rho = TrialFunctions(W)
a = (u*.... + div(u)*.... + rho*...)*dx

Whilst foo.split() splits a Function into subfunctions for further use (assignation, output, etc)

urho_out = Function(W)
solve(a == L, urho_out, bcs=...)
u_out, rho_out = urho_out.split()
File("temp.pvd").write(rho_out)

Furthermore, these somehow work interchangeably most of the time, but not all of the time.  Apologies.

On 20 March 2017 at 07:48, Floriane Gidel [RPG] <mmfg@leeds.ac.uk> wrote:

Sorry,
Z0.assign(Z1_out)
De : Floriane Gidel [RPG]
Envoyé : lundi 20 mars 2017 07:47:28
À : firedrake@imperial.ac.uk
Objet : RE: Mismatching function spaces error
 
Hi Anna,

I think you need to write something like

Phi_out, eta_out, mu_out, Z1_out, W1_out, I_out = w.split()
Z0.assign(Z0_out)

Best,
Floriane
De : firedrake-bounces@imperial.ac.uk <firedrake-bounces@imperial.ac.uk> de la part de Anna Kalogirou <A.Kalogirou@leeds.ac.uk>
Envoyé : dimanche 19 mars 2017 22:18:38
À : firedrake@imperial.ac.uk
Objet : [firedrake] Mismatching function spaces error
 
Dear all,

I am defining a mixed system as follows:

V = FunctionSpace(mesh, "CG", 1)
V_R = FunctionSpace(mesh, "R", 0)
W = V * V * V * V_R * V_R * V_R

and some of the functions defined are:

Z0 = Function(V_R)
w = Function(W)
phi, eta, mu, Z1, W1, I = split(w)

After defining and solving some solvers etc, I do the following assignment but run into a “Mismatching function spaces” error:
Z0.assign(Z1)

Variable Z1 is solved in a mixed system and I want to assign it to Z0, which belongs to space V_R.
Shouldn’t the assignment line above work? Why am I getting the error?

Thank you,

Anna.


On 19 Feb 2017, at 23:09, Anna Kalogirou <a.kalogirou@leeds.ac.uk> wrote:

Dear all,

I am trying to solve a set of PDEs and ODEs together as a mixed system. The ODE variables only depend on time, so I define them as functions from the real function space R.

First question: Is this a correct approach?

Second question: I am running into a “Mismatching function spaces” error; can someone please take a look at my code (found here) and let me know what I’m doing wrong? The error arises when the value of one of the mixed-system variables is assigned to another variable which belongs to the same function space. [I remember having the same issue some time ago, but I thought that it was fixed.]

Thank you.

Best, Anna.
_______________________________________________
firedrake mailing list
firedrake@imperial.ac.uk
https://mailman.ic.ac.uk/mailman/listinfo/firedrake