Next: 33.2.10 Conical Intersection optimization Up: 33.2 Geometry optimization Previous: 33.2.8 Approximating hessian matrix

33.2.9 Transition state (saddle point) optimization (ROOT)

ROOT,root

Specifies the eigenvector of the hessian to be followed.

root=1
specifies a minimization (default).
root=2
specifies a transition state (saddle point) optimization.

In the present implementation a saddle point search is possible with the rational funtion method (METHOD,RF), the geometry DIIS method (METHOD,DIIS) and the quadratic steepest descent method of Sun and Ruedenberg (METHOD,SRTRANS).

Note that convergence is usually much more difficult to achieve than for minimizations. In particular, a good starting geometry and a good approximation to the hessian is needed. The latter is achieved by evaluating the hessian numerically (see NUMHES section 29.2.18) or using a precomputed cartesian hessian with the HSTART command (see section 29.2.19).



P.J. Knowles and H.-J. Werner
molpro-support@tc.bham.ac.uk
Mar 8, 2000