Efektivní algoritmy pro simulaci DE systémů

Abstract

This master thesis focuses on DE systems simulation, the changes of these systems occur at discrere points of time. The simulation is based on the Monte Carlo method and its effectivity is ensured by using variance reduction techniques and advanced methods of implementation. The variance reduction methods include antithetic random variables, control variables, conditional Monte Carlo and importance sampling, the emphasis is on rare event probabilities estimation. The thesis also deals with sensitivity analysis of DE systems using central differences and the score function method. All of the methods were implemented in Matlab and GPU parallelized using CUDA technology. The simulation methods were also applied to problems of system reliability.