Some improvements of the ART method for finding transition pathways on potential energy surfaces

The Activation-Relaxation Technique nouveau (ARTn) is an eigenvector following method for systematic search of saddle points and transition pathways on a given potential energy surface. We propose a variation of this method aiming at improving the efficiency of the local convergence close to the saddle point. We prove the convergence and robustness of this new algorithm. The efficiency of the method is tested in the case of point defects in body centered cubic iron.