Causal Entropic Forces [Wissner-Gross & Freer, 2013a] is a 2013 paper by Alexander Wissner-Gross and Cameron Freer. The paper describes an agent that acts to maximize causal entropy: a measure of the diversity of futures in an agent-environment system. The authors argue that such behavior mathematically formalizes the word "intelligence"—and justify their argument with computer simulations showing that intelligent behaviors of tool use and multi-agent cooperation emerge from maximizing causal entropy.
This repository reimplements the simplest experiment (the particle in a box) in Causal Entropic Forces [2013a, Fig.2a, 2013b, pp. 2-3, 10–11]. Under causal entropic forcing, a particle in a box begins to move towards the center of the plot (I have not ran the simulation for sufficiently long timesteps for it to reach the center—the following plot took ~2 hours to generate on my laptop):
See https://github.com/dyth/causal-entropic-forces/blob/main/tutorial.ipynb for more details.
conda create --name entropica python=3.13.4
conda activate entropica
pip install notebook==7.2.2 ipython==8.29.0 numpy==2.1.2 matplotlib==3.9.2 scipy==1.14.1
A nice way to run the jupyter notebook from a remote server is using the command
nohup jupyter notebook --no-browser --ip 0.0.0.0 &