Comparison of influence of various methods of (pseudo)random sequences on the efficiency of algorithms from DE and CMA-ES family

Porównanie wpływu różnych metod generacji liczb (pseudo)losowych na efektywność algorytmów z rodziny CMA-ES i DE

Requirements

C++ Dependencies

For building the C++ components, you need:

• C++11 compiler (g++ or clang++)
• CMake or autotools (autoconf, automake, libtool)
• Eigen3 - for matrix operations
• Google glog - for logging (optional but recommended)
• Google gflags - for command line parsing (optional)

On Debian/Ubuntu systems:

sudo apt-get install autoconf automake libtool libgoogle-glog-dev libgflags-dev libeigen3-dev

Python Dependencies

• Python 3.8+
• All Python dependencies are listed in requirements.txt

Install Python dependencies:

pip install -r requirements.txt

Key Python packages include:

• numpy - numerical computing
• scipy - scientific computing
• matplotlib, seaborn - visualization
• pandas - data analysis
• jupyter, notebook - interactive notebooks

Building the Project

Quick Start

To build all C++ components at once from the root directory:

make

This will build:

• Random number generators (linear congruential and Mersenne Twister)
• CEC-2013 and CEC-2021 benchmark functions
• L-SHADE algorithm
• CMA-ES algorithm (including libcmaes if not already installed)

Build Individual Components

You can also build specific components:

make generators   # Build only random number generators
make benchmarks   # Build CEC-2013 and CEC-2021
make algorithms   # Build L-SHADE and CMA-ES

Or build from individual directories:

cd L-SHADE && make
cd CMA-ES && make
cd CEC-2013 && make
cd CEC-2021 && make

Project Structure

Random Number Generators

• linear_congruential.cpp - Linear congruential generator implementation
• mersenne_twister.cpp - Mersenne Twister generator implementation
• true_random.py - True random number generator using RANDOM.ORG API
• Random-Buffer/ - Random number buffer class implementation
• Random-Reader/ - Random number reader class implementation
• generators.ipynb - Jupyter notebook for generator analysis

Optimization Algorithms

• /CMA-ES - CMA-ES experiments implementation
  • results/ - Experimental results and data
  • notebooks/ - Jupyter notebooks for analysis
• /L-SHADE - L-SHADE algorithm implementation (downloaded from https://github.com/P-N-Suganthan/2021-SO-BCO/),
  and experiments implementation
  • results/ - Experimental results and data
  • notebooks/ - Jupyter notebooks for analysis

Benchmark Functions

• /CEC-2021 - CEC 2021 benchmark functions (downloaded from https://github.com/P-N-Suganthan/2021-SO-BCO/)
• /CEC-2013 - CEC 2013 benchmark functions (downloaded from https://github.com/dmolina/cec2013single)

External Dependencies

• /libcmaes - CMA-ES library (cloned from https://github.com/CMA-ES/libcmaes)
• /Random-Reader/rapidcsv.h - CSV parser library (cloned from https://github.com/d99kris/rapidcsv)

Data and Results

• /data - Generated random number data (not included in repository)
• /imgs - Plots and graphs

Authors

Michał Sadlej [michal.sadlej@gmail.com]

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

We gratefully acknowledge Polish high-performance computing infrastructure PLGrid (HPC Center: CI TASK) for providing computer facilities and support within computational grant no. PLG/2025/018266.