This repository contains code accompanying the paper “Any-dimensional equivariant neural networks”. It includes simple implementations of the computational recipe defined in the paper and scripts to run all the numerical experiments in the paper.
Warning: the instructions assume that your current working directory is the base of this repository.
For the following instructions, you will need Python 3.9 and pip installed. Consider creating a virtual environment for this project. For example, by running:
$ python -m venv .venv
$ source .venv/bin/activate
Then, install the requirements:
$ pip install -e .[EXPTS]
All the scripts to run the numerical examples are in the experiments
folders. Here is a table with all the scripts.
The results will be saved to a new folder within the results
directory.
Warning: running any of this experiments might take a while.
Script |
free_trace.py |
free_diagonal_extraction.py |
free_symmetric_projection.py |
free_singular_vector.py |
free_O_invariant.py |
After running any of the scripts above, you will have a new folder within the directory results/<name_of_experiment>
. Modify the last few lines of the script experiments/generate_figures.py
to include said folder. Run
$ python generate_figures.py
this will generate images like the ones in the paper. If you don’t modify the path in this script, it will simply generate the figures in the paper.