TensorFunctionSpace to FunctionSpace

General
1

Hello FEniCSx community!

I have been using TensorFunctionSpace to calculate stress values. But according to this post, I realized that it is discontinued from the dolfinx.fem module.

The way I used the TensorFunctionSpace is:

    Function_space_for_sigma = TensorFunctionSpace(mesh, ("DG", 1))

    expr= Expression(sigma(u_n,p_n),Function_space_for_sigma.element.interpolation_points())
    sigma_values = Function(Function_space_for_sigma)
    sigma_values.interpolate(expr)

This is very much similar to what they have done in: The von Mises stresses

My question is, if I only changed the first line to:
Function_space_for_sigma = FunctionSpace(mesh, ("DG", 1))
is the output exactly the same as it was in the case of:
Function_space_for_sigma = TensorFunctionSpace(mesh, ("DG", 1))

Thank you

2

If you read the first post you linked you will see it points you to the GitHub pull request that updates the process for both Vector and Tensor spaces. Reviewing Pull #2766 you can see that under changes to “python/test/unit/fem/test_assembler.py” the replacements for those spaces are provided. Therefore the answer is no, you need to modify the FunctionSpace appropriately as given in that example.

2 Likes
3

@Sven Thanks a lot for the proper direction on to the git page. With that help, I was able to get done!
My code will look like this:

    gdim = mesh.geometry.dim
    Function_space_for_sigma = FunctionSpace(mesh, ("DG", 1, (gdim, gdim)))
    
    expr= Expression(sigma(u_n,p_n), Function_space_for_sigma.element.interpolation_points())

    sigma_values = Function(Function_space_for_sigma)

    sigma_values.interpolate(expr)