dmolina/shadeils

Source code of the SHADE with Iterative Local Search, an algorithm specially designed for for real-parameter optimization with high dimensionalidad (Large-Scale Global Optimization)

This source code implement the winner of the Large-Scale Global Optimization Competition organized in IEEE Congress of Evolutionary Computation 2018, http://www.tflsgo.org/special_sessions/cec2018.html

The implementation is done in Python 3, using numpy.

This source code is freely available under the General Public License (GPLv3). However, if you use it in a research paper, you should refer to the original work:

"Molina, D., LaTorre, A. Herrera, F. SHADE with Iterative Local Search for Large-Scale Global Optimization. Proceeding of the 2018, IEEE Congress on Evolutionary Computation, Rio de Janeiro, Brasil, 8-13 July, 2018, pp 1252-1259"

It was presented in the WCCI 2018, in particular in the IEEE Congress on Evolutionary Computation. The slides are available.

Install

It is recommended to use

source install.sh

That command will create a virtual environment (virtualenv) in the directory venv with all required dependencies.

Run

The source code is prepared for doing the experiments using the Large-Scale Global Optimization CEC'2013 benchmark.

Parameters:

python shadeils -f -s [-r ] ...

• function is the number of function (between 1-15).
• run is the number of run for evaluations.
• seed is a seed value (integer value between 1 and 5).

There are other optional parameters, you can run

python shadeils.py -h```

to get the descriptions of the different optional parameters.