grassgp

Code for dimension-reduction project.

Installation

Recommended method (nix flake):

(Note this has been tested on Ubuntu 22.04 and NixOS only but should work as long as you have nix installed)

Steps:

1. Ensure nix is installed following the instructions here. (Note: If you are using NixOS you can skip this step.)
2. Enable nix flakes following the instructions here.
3. Clone the repo
4. cd into the repo (cd grassgp)
5. Activate the nix flake by running nix develop (this might take a while).
6. Use poetry to install the Python source code and the Python source dependencies: poetry install
7. (Optional): Install the Julia source code to generate the data for the localised active subspace examples:
   1. Activate a Julia REPL: julia
   2. Open the Julia pkg REPL by pressing ] from the Julia REPL.
   3. Activate the environment contained in the repo root: activate .
   4. Instantiate the enviroment via: instantiate (this might take a while).
8. You can now spawn a Python shell with all the dependencies using poetry shell.
9. (Alternatively/recommended): you can launch a Jupyterlab environment containing the Python and Julia dependencies using: poetry run jupyter lab

Alternative method (Generic Linux - without nix):

Steps:

1. Clone the repo
2. cd into the repo (cd grassgp)
3. Create a virtual environment
4. Install poetry -- suggested method:
   1. Install pipx
   2. Install poetry with pipx install poetry
5. (Optional/recommended): upgrade pip with pip install --upgrade pip
6. Activate virtual environment.
7. Use poetry to install: poetry install
8. (Optional): Install the Julia source code to generate the data for the localised active subspace examples:
   1. Install Julia stable following the instructions here.
   2. Activate a Julia REPL: julia
   3. Open the Julia pkg REPL by pressing ] from the Julia REPL.
   4. Activate the environment contained in the repo root: activate .
   5. Instantiate the enviroment via: instantiate (this might take a while).
9. You can now spawn a Python shell with all the dependencies using poetry shell.
10. (Alternatively/recommended): you can launch a Jupyterlab environment containing the Python and Julia dependencies using: poetry run jupyter lab