A minimalistic environment for spatial reasoning.
Step 1:
git clone --recursive https://github.com/lil-lab/knotgym.gitcd knotgym/
uv venv
uv pip install -e .Step 2: Install pyknotid dependency with cython extension
cd external/pyknotid
uv pip install -e .
uv run python setup.py build_ext --inplace
cd ../..To verify
uv run python -c "from pyknotid.spacecurves import chelpers" # no error/warningWe rely on these chelpers to compute Gauss code. The speedup is quite significant, if you plan on training with them.
Step 3: Open MuJoCo visualizer
# Locating mjpython could be tricky on MacOS
cd knotgym
MUJOCO_GL=glfw mjpython script/mjc_interact.pySee baselines/README.md
@misc{chen2025knotsimpleminimalisticenvironment,
title={Knot So Simple: A Minimalistic Environment for Spatial Reasoning},
author={Zizhao Chen and Yoav Artzi},
year={2025},
eprint={2505.18028},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2505.18028},
}