PCA Bats

Bat Algorithm with Principal Component Analysis for multi-dimensional optimisation problems

Requirements

Install the requirements with pip:

pip install -r requirements.txt

Run

Define a run of the bat algorithm in the main method of the main file, you need to define a X-dimensional fitness function e.g.

run(
    True,               # Whether to use PCA analysis for principal individuals
    functions.rosen,    # Function to benchmark
    -100,               # Lower bound of problem
    100,                # Upper bound
    200                 # Generations
    )

Reproduce results

To reproduce the results of the paper, run the benchmark.py script twice. Once with levy=True and once with levy=False, indicating the use of levy flight over global search, in the run method of the main.py script. Then execute

python3 benchmark.py

for each option for levy. For reproducibility, all results were produced with numpy's default_rng() initialised with seed 0.