Adaptive Latents project codebase

Quickstart

# download the repo
git clone https://github.com/draeloslab/AdaptiveLatents
cd AdaptiveLatents

# install dependencies
conda env create --file=environment.yml

# install the repo locally
conda activate adaptive_latents
pip install -e .

# run the following line if you want to use the GPU version of jax
pip install -U "jax[cuda12]"
# note: if using apple silicon, you may need to build jax from source

# create an adaptive_latents_config.yaml file if you wish
# the default copy is found in adaptive_latents/adaptive_latents_config.yaml
# just make sure your copy is in the current working directory when you import adaptive latents

# test if everything is working
python workspace/main.py # this should produce a gif in the current working directory
pytest .

Requirements

Our algorithm is implemented in python with some extra packages including: numpy, jax (for GPU), and matplotlib (for plotting).

We used python version 3.9 and conda to install the libraries listed in the environment file. We provide an environment file for use with conda to create a new environment with these requirements, though we note that jax requires additional setup for GPU integration (https://github.com/google/jax).

Development

# check if most tests pass
pytest . -n 5 # the `-n 5` uses 5 processes; this is nice for a quick check

# check test code coverage
coverage run -m pytest --nbval-lax --longrun
coverage html
open tests/reports/coverage-html/index.html 2>/dev/null

# run the yapf formatter
yapf --in-place --recursive .

Reference

The core Bubblewrap algorithm was initially described here: 'Bubblewrap: Online tiling and real-time flow prediction on neural manifolds'.