Strategies forWalkingonPotentialEnergySurfaces

Al! algorithm for locating stationarypoints corresponding to lociu minima and transition stateson potential energysurfacesis further analyzed.This method utilizes local gradientand Hessian (Le., fust and secondenergyderivative) information to generatea seriesof "steps " that are folIowedto thedesired stationarypoint. By designingthestepsequenceto moveenergeticallydownhill in all coordinates,local minima can be found. By steppinguphill along one local eigenmodeof the Hessian wIDIe minimizing the energyalong all other modes, one)ocates transition states.Key elementsof this developmentare moceefficient parameterizationof the stepvector in terms of quantities that permit the direction (Le., uphill or downhill), and lengthofthe stepto,becarefullycontrolled, and implementationofthe ability to explore "side channels " as attractiveoptions occur.,"4