An initial implementation of "open loop in natura economic planning". Note that this is not your standard "data - train - validate - test" machine learning code, the topic of the paper is very different, and the purpose of the experiments is to showcase feasibility, rather then benchmark vs a previous method (which does not exist).

All the generated data for the plots is in ./plot_data. All data required for the village experiment is in ./data/. The random generated matrices are missing, but (a) they can be generated using ./generate_matrices.py (b) they are 22GB.

./selvaria.py executes the experiments and generates the experiment for optimising the economy of a small alien village.

./online_time_complexity.py optimises for one tick for various random economies of different sizes. All the generated files are around 22GB, so they are not included (but you can generate them).

./coeffs.py has the coefficients for the production units of the village economy. The corresponding consumption data is in ./data. If you just run the file it will create the coefficient plots for those three production units (i.e. factories).

./utils.py include various helper functions.

./generate_matrices.py includes the code to generate random sparse matrices.

./plot_selvaria.py includes the code for investment plots, as presented in the paper.

./plot_time.py plots the (empirical) time complexity of solving for matrices of certain size.

./metrics.py includes the code for humanity and other metrics.