Peter Pulay
Two topics will be discussed:
(1) Geometry optimization of large molecules, in particular large biomolecules in redundant internal coordinates. This usually requires substantially fewer steps than optimization in Cartesians. However, it is competitive with the latter for large molecules only if the cubic scaling of the transformation of the forces, and the back-transformation of the geometry to Cartesians is eliminated. Alternative methods for this will be compared.
(2) A novel, exact formulation of Newtonian molecular dynamics in internal coordinates is described. The method needs only the first and second order B matrices, and treats the generalized centrifugal and Coriolis forces in a uniform way. It is very simple compared to previous formulations and can use arbitrary internal coordinates, not only tree-like (Z-matrix) ones. It can be used not only to freeze the rigid degrees of freedom but slow them down, so that their average value can adapt to the conformational variables.