|
Main /
Geometry OptimizationA geometry optimization is automatically carried out, as soon as {$*$}s are specified in the Z-Matrix after the variable names. The default choice is a search for a minimum on the potential energy surface using analytically evaluated gradients within a Quasi-Newton scheme. The Hessian update is then done using the BFGS scheme starting with a unit matrix in the first iteration. If a Hessian matrix is available, a Newton-Raphson scheme is applied. A Powell-update of the Hessian (instead of the BGFS update) is performed when GEO_METHOD=RFA is specified. A transition-state search is requested via GEO_METHOD=TS (for a detailed description how transition states can be determined, see the corresponding example and recommendations) The convergence criterium for completion is set via GEO_CONV=N. The value is specified in Hartree/bohr and the default is N=5 leading to a threshold value of 10**-5. The maximum number of iterations is specified via GEO_MAXCYC. Default is here 50. It is in many cases advantageous to perform the geometry optimization using a pre-calculated force-constant matrix. Such a force-constant matrix can be supplied by simply copying the corresponding FCM or FCMINT file into the working directory. Another option is to compute the force-constant matrix within the same job at the same or at a lower level than the actual geometry optimization. This can be accomplished by specifying a %fcm section in the ZMAT file which contains all the relevant information for this preceeding force-constant calculation. Examples geometry optimizations based on analytic gradients geometry optimizations based on single-point energies |
CFOUR is partially supported by the U.S. National Science Foundation.