Stochastic Diffusion Adaptive Optimization (SDAO)

Description

The Stochastic Diffusion Adaptive Optimization (SDAO) is a stochastic optimization algorithm inspired by adaptive diffusion. This algorithm is used to solve optimization problems in high-dimensional search spaces. SDAO is compared against various state-of-the-art (SoA) optimization algorithms across different benchmark scenarios.

Benchmark Functions

The SDAO model has been tested against a variety of benchmark functions, which are divided into the following categories:

1. Standard Benchmark Functions: These are classical test functions commonly used in optimization literature.
2. Stochastic Benchmark Functions: These functions include stochastic noise to simulate uncertainty in function evaluations.
3. Real-World Functions: These functions represent real-world optimization problems, such as Vehicle Routing Problem (VRP), predictive maintenance, and more.
4. CEC Functions: These are test functions from the Competitions on Evolutionary Computation (CEC), which are widely used to evaluate optimization algorithms.

Compared Algorithms

SDAO is compared against several state-of-the-art optimization algorithms, including:

• Stochastic Fractal Search (SFS)
• Algebraic Stochastic Gradient Descent (ASGD)
• SHADE with Iterative Local Search (SHADEwithILS)
• Path Relinking
• Adaptive Multi-Swarm Optimization (AMSO)
• Two-Level Particle Swarm Optimization (TLPSO)

Requirements

To run the model and tests, you will need to have the following Python packages installed:

• numpy
• scipy
• pandas
• matplotlib
• seaborn
• optuna
• tqdm
• pydash
• statsmodels

You can install these packages using poetry:

poetry install

Project Structure

The project is organized as follows:

• SDAO/model/: Contains the implementation of the SDAO model and other optimization algorithms.
  • __init__.py: Initializes the module.
  • __main__.py: Runs the SDAO algorithm.
  • functions/: Contains the benchmark functions.
  • performance_metrics/: Contains performance metrics and scripts to generate plots and tables.
  • soa/: Contains the implementation of state-of-the-art optimization algorithms.
  • solver.py: Defines the solver interface for testing and comparing models.
  • utils/: Contains utilities for parameter optimization and statistical analysis.

Running the Model

Running the SDAO Algorithm

To run the SDAO algorithm, you can use the following command:

poetry run python -m model

This command will execute the SDAO algorithm with the default parameters and generate the benchmark results.

Parameter Optimization

To optimize the SDAO parameters using Optuna, you can run the following script:

poetry run python -m model.utils.optimize_params

This script will perform a hyperparameter search using Optuna and display the best parameters found.

Performance Metrics

To generate performance metrics and plots, you can run the following script:

poetry run python -m model.performance_metrics.run -s all -d 50 -pdf

This command will generate performance metrics for all scenarios and specified dimensions, and save the plots in PDF format.

License

This project is licensed under the MIT License. You can view the license file for more details.