Optimization methods for individual-based model parameter estimation in predictive microbiology

Abstract

In the framework of microbiology, Individual-based Models are discrete models in which the main entities are microbes. Their use in simulations as ‘virtual experiments’ to predict the evolution of populations under specific conditions requires accurate setting of the parameters involved. We adapted and tested two optimization methods for Individual-based Model parameter estimation: the Nelder-Mead Threshold Accepting (NMTA) and the NEWUOA. These methods presented no convergence problems, and the best results in terms of time expenditure were derived with the latter.