Calibration of parameters in Dynamic Energy Budget models using Direct-Search methods

Area of Science:

• Metabolic theory
• Ecological modeling
• Quantitative biology

Background:

• Dynamic Energy Budget (DEB) theory quantifies metabolism across species.
• DEB model parameterization relies on observational and experimental data.
• Current DEB toolbox uses Nelder-Mead Simplex, facing convergence and constraint limitations.

Purpose of the Study:

• To numerically compare the Nelder-Mead Simplex method with the SID-PSM algorithm for DEB parameter calibration.
• To introduce and evaluate a hybrid Simplex Directional Direct-Search algorithm.
• To address limitations in constrained optimization for DEB model parametrization.

Main Methods:

• Numerical comparison of Nelder-Mead Simplex and SID-PSM algorithms.
• Implementation of a hybrid Simplex Directional Direct-Search algorithm.
• Evaluation of algorithm performance on constrained optimization problems in DEB theory.

Main Results:

• The SID-PSM algorithm demonstrates convergence for both unconstrained and constrained problems.
• The hybrid Simplex Directional Direct-Search algorithm proves robust and efficient.
• The hybrid method effectively solves constrained optimization problems in DEB model parametrization.

Conclusions:

• The Simplex Directional Direct-Search algorithm enhances DEB model calibration.
• This hybrid approach overcomes limitations of derivative-free direct-search methods.
• Improved DEB model parametrization leads to better quantitative understanding of metabolism.