Import functions from the.txt/.csv file

[RokyC]

~~~~~ I have one file contains node coordinates and displacement field/temperature field data and would like to import function in Dolfinx as boundary conditions. I wonder how this works in the Dolfinx version?
~~~~~ I tried with reference to Impose Velocity Profile on Inlet Face using Array - #6 by dokken , but some of the code under dolfinx may not apply.

# read mesh and meshtags
with dolfinx.io.XDMFFile(MPI.COMM_WORLD, 'mesh.xdmf', 'r') as xdmf:
    mesh = xdmf.read_mesh(name="Grid")  
    subdomains = xdmf.read_meshtags(mesh, name="Grid")  
mesh.topology.create_connectivity(mesh.topology.dim, mesh.topology.dim-1)  
with dolfinx.io.XDMFFile(MPI.COMM_WORLD, 'mt.xdmf', 'r') as xdmf:
    boundaries = xdmf.read_meshtags(mesh, name="Grid")

# Define Function
V = dolfinx.fem.FunctionSpace(mesh, ("CG", 1))
u_p = dolfinx.fem.Function(V)  
dof_coords = V.tabulate_dof_coordinates()  

# from .txt import data
dofs = np.array([[x0,y0,z0],...,[xn,yn,zn]])
u_x = np.array([u_x])
u_y = np.array([])
u_z = np.array([])

# Function degrees of freedom are assigned through /.txt X/Y/Z value data
u_p.vector.setValueLocal(dof_index, values[u_x]/values[u_y]/values[u_z])

# Apply the dirichlet boundary to S0 surface of meshtag
bcl = DirichletBC(V, u_p, S0)

[dokken]

I would suggest to make a minimal example that either produces an error, or produces a wrong result, as shown in: Impose Velocity Profile on Inlet Face using Array - #2 by dokken

[RokyC]

I think I might have a better description of the problem.I want to import the pressure distribution (.txt) known on the surface of the solid structure as boundary conditions. I found examples that define boundary conditions:Elasto-plastic analysis of a 2D von Mises material — Numerical tours of continuum mechanics using FEniCS master documentation (comet-fenics.readthedocs.io)

But here it has a constant pressure “q”, and if the pressure is a spatial distribution q(x,y,z), how do I define it in variational form? I now have to import pressure to dolfinx.fem.functionspace definition of function, but the test-function “v” is dolfinx.fem.VectorFunctionSpace defined.

[dokken]

Is the pressure distribution an explicit function of the coordinates? i.e. p=p(x,y,z) or is it only defined at certain nodes?

[RokyC]

Pressure distribution is the form data from [x, y, z, pvalue] stored in .csv interpolation to the function defined in dolfinx.fem.functionspace.

[belmont]

Perhaps you may find the discussion in this topic usefull

Fenicsx: use an array of data for boundary conditions

Hi. I have the following problem: Consider the following matrix in a 3d Poisson equation problem: (x_i, y_i, z_i, bc_i). The (x_i, y_i, z_i) are the bounds of a 3d mesh and bc_i is the relevant boundary conditions. So, consider now making the 3d mesh and perform a numerical calculation to find the bc_i values of each boundary point (x_i, y_i, z_i). Using those boundary conditions, I would like to use fenicsx to solve the Poisson equation in the whole mesh. I have solved this problem some year…

I also wanted to feed my own values to the boundary conditions and the discussion in this topic helped me.

[RokyC]

~~~~~ Yes, I also got a lot of help from this topic before. However, the difference between my current problem and this is that the distribution I imported is not as DirichletBC, it appears in the linear term as Neumann boundary.
~~~~~ At this time, the p is defined in dolfinx.fem.functionspace and the test-function “v” is defined in dolfinx.fem.VectorFunctionSpace. p[pressure] is just one column of degrees of freedom, and V is three columns of degrees[v_x,v_y,v_z] of freedom, I don’t know how do they multiply. Profile - RokyC - FEniCS Project, In the above example, the problem is avoided because p is a constant in space.

[dokken]

The pressure p cannot be a scalar value, as there is a dot product between p and v, meaning that they are of the same dimension.
I guess you are assuming that the pressure p(x,y,z) is working in the normal direction?

[RokyC]

~~~ Yes, the pressure is air pressure, it is working in the normal direction of solid surface. In this case, how do I implement it?

[dokken]

As stated above, you create a scalar space, as in:

Fenicsx: use an array of data for boundary conditions

You should remove this code, as the whole point of was to replace the interpolate command

and then your variational form is
inner(p*n,v)*ds were n is the facetnormal