Scientific Computing Tools for Finite Element Methods

This package contains a collection of tools for scientific computing with a focus on finite element methods. The tools are written in Python and are intended to be used in conjunction with the dolfinx.

Many users that are transitioning from legacy FEniCS to FEniCSx may find the transition difficult due to the lack of some functionalities in FEniCSx. This package aims to provide some of the functionalities that are missing in FEniCSx.

See the documentation for the API reference and runnable examples of the features below.

Backwards compatibility

We aim to support the last two stable releases of DOLFINx. Sometimes features are supported for longer if need be.

Some features might depend on functionality added to DOLFINx, which means that they won't be backwards compatible.

Features

Meshes, tags and submeshes

• Create MeshTags from a list of tags and corresponding locator functions.
• Extract a submesh of entities of any co-dimension.
• Locate the interface between two subdomains, and the exterior facets of a subdomain.

Periodic meshes

• Build a periodic mesh (MPI supported), by merging the vertices on opposite sides of the domain. The periodicity lives in the topology, so a continuous function space on the result is periodic with no constraint matrix, as opposed to DOLFINx_MPC.
• Take the vertex pairs from the $Periodic section of a gmsh model rather than from the coordinates.
• Move mesh tags onto a periodic mesh, and move a solution back onto the mesh it was built from. Required for post-processing with Pyvista or Paraview.
• Diagnose the two ways an input mesh can be unsuitable, a missing ghost layer and cells that collapse onto each other across a seam.

Function spaces and degrees of freedom

• Spaces of functions that are constant on each subdomain.
• Maps between degrees of freedom and vertices, in both directions.

Assembly, sources and boundary conditions

• Assemble a scalar or a norm over all processes in a single call.
• Point sources for usage in DOLFINx (>=v0.8.0).
  • Point sources in vector spaces are only supported on v0.9.0, post DOLFINx PR 3429. For older versions, apply one point source in each sub space.
• Interpolate an expression onto the degrees of freedom of a set of facets.
• Helpers for the PETSc vector operations that surround a solve: zeroing, ghost updates, lifting and boundary conditions.

Evaluation, interpolation and geometry

• Evaluate a function at arbitrary points, in parallel.
• Find the extrema of a UFL expression within each cell, or over a whole domain.
• Project points onto the closest point of a mesh.
• Build interpolation matrices from any ufl.core.expr.Expr into a compatible space.

Output

• Save quadrature functions as point clouds.
• Save any function that can tabulate dof coordinates as point clouds.

Biomedical

• Read MRI data onto a mesh, as a function or as cell tags. Requires the biomed extra.

Installation

The package is partly written in C++ and relies on dolfinx. Users are encouraged to install scifem with pip in an environment where dolfinx is already installed or with conda.

pip

To install the package with pip run

python3 -m pip install scifem --no-build-isolation

To install the development version you can run

python3 -m pip install --no-build-isolation git+https://github.com/scientificcomputing/scifem.git

Note that you should pass the flag --no-build-isolation to pip to avoid issues with the build environment, such as incompatible versions of nanobind.

spack

The spack package manager is the recommended way to install scifem, and especially on HPC systems. For information about the package see: spack-package: py-scifem First, clone the spack repository and enable spack

git clone --depth=2 https://github.com/spack/spack.git
# For bash/zsh/sh
. spack/share/spack/setup-env.sh

# For tcsh/csh
source spack/share/spack/setup-env.csh

# For fish
. spack/share/spack/setup-env.fish

Next create an environment:

spack env create scifem_env
spack env activate scifem_env

Find the compilers on the system

spack compiler find

and install the relevant packages

spack add py-scifem+petsc+hdf5+biomed+adios2 ^mpich ^petsc+mumps+hypre ^py-fenics-dolfinx+petsc4py
spack concretize
spack install

Finally, note that spack needs some packages already installed on your system. On a clean ubuntu container for example one needs to install the following packages before running spack

apt update && apt install gcc unzip git python3-dev g++ gfortran xz-utils -y

conda

To install the package with conda run

conda install -c conda-forge scifem

Having issues or want to contribute?

If you are having issues, feature request or would like to contribute, please let us know. You can do so by opening an issue on the issue tracker.