Modelling of catalytic activity of transition metals in hydrogenolysis of ethane by the pattern recognition approach

Abstract

For modelling the catalytic activity in hydrogenolysis of ethane, thirty transition metals were characterized by a set of fifteen fundamental physical variables for each of them. The dimensionality reduction of this physical model up to the dimensionality seven was carried out by the determination of seven linearly almost independent variables. This set was used for the definition of the class of catalytically active metals on the basis of their similarity using an asymmetric (one-class) pattern recognition approach. The values of similarity measure were approximated by the polynominal function of the second degree which is accounted for as a mathematical representation of the structure of the active class.