First order simultaneous optimization of molecular geometry and electronic wave function

Abstract

We present a very efficient simultaneous optimization method for both molecular geometry and electronic wave function. We introduce a simultaneous least squares scheme to minimize errors in a full local quadratic space of both geometric and wave function vectors. Several self-consistent-field iterations are taken until a uniform search direction is achieved in every geometry optimization step. Not only does this new simultaneous optimization method significantly reduce the number of self-consistent-field iterations to reach convergence, it is able to overcome shallow potential wells to find a better stationary point. This new method exhibits up to approximately 60% savings in computational cost compared to conventional geometry optimization methods.