TemperatureFromXDMF work in a conda environment

User Support
1

Hi all,

It seems that when trying to use the TemperatureFromXDMF class in combination with using chemical potential such as in the following script, seems to fail using festim 1.0.0 in a conda environment of fenics, but passes within a docker environment. I can make an accompanying issue on the repo too.

import fenics as f
import festim as F

mesh = f.UnitSquareMesh(10, 10)
V = f.FunctionSpace(mesh, "CG", 1)
expr = f.Expression("1 + x[0] + 2*x[1]", degree=2)
T = f.interpolate(expr, V)
# write function to temporary file
T_file = "T.xdmf"
f.XDMFFile(T_file).write_checkpoint(
    T, "T", 0, f.XDMFFile.Encoding.HDF5, append=False
)


my_mesh = F.Mesh()
my_mesh.mesh = mesh

my_model = F.Simulation()

my_model.mesh = my_mesh

my_model.materials = F.Materials(
    [
        F.Material(
            id=1,
            D_0=1.0,
            E_D=0.1,
            S_0=1.0,
            E_S=0.1,
)])

my_model.T = F.TemperatureFromXDMF(filename=T_file, label="T")

my_model.settings = F.Settings(
    transient=False,
    chemical_pot=True,
    absolute_tolerance=1e01,
    relative_tolerance=1e-06,
)

my_model.initialise()
2

Can you provide more information?( Error message, and how you created the conda environment.)

What is the festim version in each environment (conda, docker) with:

pip show festim
3

This PR should fix the bug. Once merged, we’ll release a patch version.

4

Yes of course,

The error message being raised:

Traceback (most recent call last):
  File "/home/jdark/repos/FESTIM-review/gallery/breeding_blanket/mwe.py", line 41, in <module>
    my_model.initialise()
  File "/home/jdark/anaconda3/envs/festim-env/lib/python3.11/site-packages/festim/generic_simulation.py", line 255, in initialise
    if self.T.is_steady_state():
       ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jdark/anaconda3/envs/festim-env/lib/python3.11/site-packages/festim/temperature/temperature.py", line 54, in is_steady_state
    return "t" not in sp.printing.ccode(self.value)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jdark/anaconda3/envs/festim-env/lib/python3.11/site-packages/sympy/printing/codeprinter.py", line 741, in ccode
    return c_code_printers[standard.lower()](settings).doprint(expr, assign_to)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jdark/anaconda3/envs/festim-env/lib/python3.11/site-packages/sympy/printing/codeprinter.py", line 155, in doprint
    expr = self._handle_UnevaluatedExpr(expr)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jdark/anaconda3/envs/festim-env/lib/python3.11/site-packages/sympy/printing/codeprinter.py", line 115, in _handle_UnevaluatedExpr
    return expr.replace(re, lambda arg: arg if isinstance(
           ^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'replace'

The version of festim in both cases was 1.0.0.

It seems from investigation this error is a result of the version of sympy within the docker container (1.1.1) vs that which is in the conda env (1.11.0). Specifically, changes to how the function sp.printing.ccode() operates.

5

FESTIM 1.1.1 was just released, can you try updating with

pip install festim==1.1.1

And confirm that the bug is fixed?

6

Can confirm it now works, cheers!