Systematic determination of MCSCF equilibrium and transition structures and reaction paths

Publisher Website
Google Scholar
Abstract
The restricted step optimization algorithm is applied to potential energy surfaces calculated from multiconfiguration self‐consistent‐field wave functions. Equilibrium and transition‐state geometries are determined by iteratively solving a set of level‐shifted Newton–Raphson equations. At each geometry the molecular gradient and Hessian are calculated analytically, and a first‐order prediction of the wave function at the next geometry is obtained by combining the geometrical derivatives of the wave function with the geometrical step vector. The usefulness of this prediction is discussed and illustrated by test calculations. The numerical accuracy which is required in the wave function and its geometrical derivatives in order to maintain quadratic convergence in the optimization of the molecular geometry is analyzed. It is demonstrated that the Newton–Raphson step vector and the wave function prediction may be determined without calculating the molecular Hessian explicitly. Sample calculations are carried out for the potential energy surfaces of diazene (N2H2) and the diazenyl radical (N2H). Equilibrium geometries are determined in less than five iterations and the optimization of transition states requires typically ten iterations.