Application of evolutionary strategies to automated parametric optimization studies in physics research
Rüdiger Berlich · University Library BOCHUM (Ruhr University Bochum) · 2003
This thesis presents the application of Evolutionary Strategies, an efficient optimization algorithm, to particle physics problems taken from the BaBar experiment (SLAC / USA). An additional usage pattern involving a Dalitz plot analysis is demonstrated using data from the CB/ELSA experiment (Bonn / Germany). The "EVA" library, a new implementation of distributed Evolutionary Algorithms that was developed as part of this thesis combines as one of the first implementations the ability to perform the optimization in parallel, on devices ranging from single processor- over SMP-machines all the way to clusters and the Grid, with an intuitive interface to a standard tool in particle physics analysis - the ROOT framework.