karinemiras / revolve-oee

Revolve2-Multi

Welcome to the Revolve2-Multi Tutorial, it goes along with Dimitri Kachler's thesis, "Predator-Prey Dynamics in an Open-Ended Environment". This readme will guide you along the steps to be able to recreate the experiments found in the paper.

0. The main folder where our code for the experiment is found at:

/revolve2-multi/examples/sanity3P/

The main files to look at are:

optimize.py

optimizer.py

genotype.py

1. First you want to create a virtual environment with Python version 3.8 as the primary Python distribution.

Then you want to create a virtual environment to serve as a capsule for all the libraries:

python3.8 -m virtualenv .venv

Then, activate the virtual environment:

source .venv/bin/activate

2. Then you want to clone this repository to a directory, we will refer to the directory as "/yourdir/"

3. In the virtual environment, we install the necessary libraries:

pip install /yourdir/revolve2-multi/core

pip install /yourdir/revolve2-multi/standard_resources/

pip install /yourdir/revolve2-multi/genotypes/cppnwin

pip install /yourdir/revolve2-multi/runners/mujoco

4. We change our directory to the experiment folder, this is also where all the data will be:

cd /yourdir/revolve2-multi/examples/sanity3P/

5. The experiment runs by activating optimize.py. You should receive preliminary warnings, at which point the experiment is running:

python3 optimize.py

6. After 6500 seconds (instead of 6000 just to be safe), the experiment will automatically stop and the data will be stored in two files: "countries.csv" and "deathBorn.csv".

7. Copy and rename both files according to the experiment run, i.e. "countries4.csv" and upload both files to the Deepnote project to visualize the data:

https://deepnote.com/workspace/oasis-4a75-9d2a5e51-2ae1-4a9f-9e63-b0cb3d7f90e0/project/VisualizeMetrics-6e3b7bcb-ad31-4c62-8372-a7e756152eb6/notebook/visualizeWalls-592bf1878e7447f59ea271b009d4575f

8. After all 10 runs have been completed and uploaded to Deepnote, the visualization may begin

Documentation For Revolve2

ci-group.github.io/revolve2