Simulated-annealing-based genetic algorithm for modeling the optical constants of solids

Area of Science:

• Computational physics
• Materials science
• Optimization algorithms

Background:

• Model parameter estimation is crucial for understanding material properties.
• Accurate estimation requires robust optimization techniques to avoid local minima.
• Existing methods like genetic algorithms can struggle with complex, multi-modal landscapes.

Purpose of the Study:

• To develop and evaluate a hybrid algorithm combining simulated annealing and genetic algorithms for parameter estimation.
• To compare the performance of the proposed algorithm against traditional genetic algorithms.
• To assess the algorithm's efficacy in finding global minima and improving parameter accuracy.

Main Methods:

• A simulated-annealing-based genetic algorithm (SA-GA) was developed.
• SA-GA was tested on synthetic data mimicking metal optical constants.
• Performance was benchmarked against standard genetic algorithms (GA).
• Key metrics included global minimum convergence and parameter estimation accuracy.

Main Results:

• The SA-GA demonstrated superior performance compared to the plain GA.
• The hybrid algorithm showed enhanced ability to locate the global minimum.
• More accurate model parameter values were obtained using the SA-GA.
• The algorithm successfully fitted the dielectric function model to platinum and aluminum data.

Conclusions:

• The simulated-annealing-based genetic algorithm is an effective approach for complex parameter estimation problems.
• This hybrid method offers improved accuracy and global convergence over standard genetic algorithms.
• The algorithm shows promise for applications in materials science, particularly in optical property analysis.