Installation of Firedrake

1. Installation of Firedrake#

See the official document Install

Using Docker with the image provided by the Firedrake project is recommended.

To install Firedrake, the computer should have access to the internet (including github.com and pypi).

1.1. Some Notes on Installation#

1.1.1. Install petsc with external packages pre-downloaded into a directory#

Sometimes, some of the packages that petsc depends on cannot be downloaded automatically. We can add the option

--with-packages-download-dir=<path/to/petsc/packages>

to command configure of petsc to obtain the list of required packages. Then download these packages manually and put them into the path. Afterwards, configure it again with the above option.

The python script download_petsc_pkgs.py can be used to download the packages. Save the output of petsc configure to a file, for example pkgs_info.txt. Then run the following command to download packages:

python3 download_petsc_pkgs.py -d <path/to/petsc/packages> pkgs_info.txt

1.1.2. Using Firedrake in JupyterLab (installed outside the Firedrake env)#

  1. Install jupyterlab

    python3 -m pip install jupyterlab

    You may need to add $HOME/.local/bin to the environment variable PATH:

    export PATH=$PATH:$HOME/.local/bin

  2. Configure jupyterlab

    Generate config file for customizing jupyter lab:

    jupyter lab --generate-config

    Set password for jupyter lab:

    jupyter lab password  #

  3. Configure Browser

    In wsl-ubuntu, configure the browser like this:

    export BROWSER="/path/to/chrome/or/firefox"

    An example of chrome:

    export BROWSER='/mnt/c/Program Files/Google/Chrome/Application/chrome.exe'

    Now, you can type jupyter-lab to start jupyter. You will see jupyter in browser.

  4. Configure kernels

    1. Activate env:

      $ source /your/env/path/activate

    2. Add kernels:

      (your-venv)$ ipython kernel install --name "local-venv" --user

      The output should look like:

      Installed kernelspec local-venv in /path/to/kernels/local-venv

      where the path /path/to/kernels/local-venv will have a different name in your environment.

      Also, we can find the kernelspec path by running jupyter kernelspec list, which will output all available kernels with their paths.

      Now we need to check the python path in kernel.json, which is located in the above path. Make sure it is the python in your env. Otherwise, correct it.

    3. Add environment variables to kernel.json:[1]

      An example of kernel.json:

      {
 "argv": [
  "/home/yzz/firedrake/real-int32-debug/bin/python",
  "-m",
  "ipykernel_launcher",
  "-f",
  "{connection_file}"
 ],
 "env": {
  "OMP_NUM_THREADS": "1",
  "PATH": "/home/yzz/firedrake/real-int32-debug/bin:${PATH}"
 },
 "display_name": "firedrake-real-int32",
 "language": "python",
 "metadata": {
  "debugger": true
 }
}