Hybrid evolutionary optimization of two-stage stochastic integer programming problems: an empirical investigation.

Area of Science:

• Operations Research
• Optimization
• Stochastic Programming

Background:

• Decision-making under uncertainty presents significant challenges.
• Moving horizon frameworks allow for adaptive recourse actions.
• Stochastic integer programming models complex, multi-stage decisions.

Purpose of the Study:

• To develop an efficient algorithm for moving horizon decision problems with parameter uncertainties.
• To address limitations of existing methods in handling large-scale stochastic integer programs.
• To enable effective recourse actions in dynamic decision environments.

Main Methods:

• Formulation of a mixed-integer recourse model as a two-stage stochastic integer program.
• Development of a hybrid evolutionary algorithm integrating evolutionary computation and mathematical programming.
• Stage decomposition for optimizing here-and-now and recourse decisions.

Main Results:

• The proposed hybrid algorithm effectively solves two-stage stochastic integer programs.
• Empirical analysis demonstrates superior performance in terms of speed for problems with many scenarios.
• The algorithm generates high-quality feasible solutions efficiently.

Conclusions:

• Hybrid evolutionary algorithms offer a powerful approach for dynamic decision problems under uncertainty.
• This method provides a scalable and efficient solution for complex stochastic optimization.
• The findings have implications for real-world applications requiring robust decision-making with recourse.