A multidimensional optimisation is a technique used in many problems of physics and constitutes challengers for numerical methods of computer science. As an example the development of a new genetic algorithm designed for the data analysis in the Coulomb excitation experiments will be presented. Calculations performed with a multidimensional test function identified the weakness of standard operators of a real representation implemented to the genetic algorithm. The corrections to the algorithm were proposed and the method was tested on a real case data. A probabilistic theory and graph theory were applied to establish a new method of a multidimensional surface investigation close to the optimum. New features as statistical analysis of chi square surface sampling will be demonstrated to determine the quality of a fit as well as an experimental uncertainty of the result.
A successful implementation to the data analysis of Coulomb execution of 100Mo will be presented and possible further application of the method will be discussed.