This repository contains directions for regenerating the results in the publication entitled "Animorphic Ensemble Optimization: A Large-scale Island Model" by Dean Price and Majdi Radaideh.

Prerequisites:

1. Install NEORL from https://neorl.readthedocs.io/en/latest/guide/install.html
2. Install tqdm from https://pypi.org/project/tqdm/

Data generation:

Most of the results are pre-generated and placed into textfiles which are then used to generate plots. To generate these files, first make two directories in the main repository called "comp_results_p1" and "comp_results_p2". Then, run the "aeo_config_runs.py" script. This can be a very lengthy computation, the "nproc" variable included at the beginning of this script can be used to change how many processors are used. Next, run the "non_aeo_runs.py" script. Finally, the "make_netcdf.py" file should be run.

Figure 5:

This figure can be generated by running the "comparison_withinAEO_plot.py" script.

Table 1:

This table can be generated by running the "comparison_withinAEO.py" script.

Table 2:

This table can be generated by running the "comparisons_outsideAEO.py" script.

Table 3:

This table can be generated by running the "wilcox_tables.py" script.

Table 4:

This table can be generated by running the wilcox_tables.py" after removing the "#" at the beginning of line 13.

Summary Tables:

A large set of results from all optimization routines can be generated by running the "summary_tables.py" script.