An exhaustive search approach for chemical kinetics experimental data fitting, rate constants optimization and confidence interval estimation

Abstract

Finding reaction rate constants in a complex chemical mechanism is a complicated problem. These complications can be attributed mainly to the nonlinearity of such systems, where the least squares optimization procedure fail when experimental data are being fitted. Here we analyze two cases: analytical and numerical. From analytical case possible issues of optimization are found and methods to cope with them are derived. The complex case shows how an exhaustive search fitting approach works. Our results indicate that certain properties of systems can be found only by running grid search on entire rate constant set and minima found have practical value only if the properties of entire fitting score surface are known.