DYNAMIC PROGRAMMING, REDUCTION OF DIMENSIONALITY AND EIGENVALUE PROBLEMS—II. MULTIDIMENSIONAL CASE