Search

The lower‐bounds method presented by Löwdin in the preceding paper has been applied to oscillators perturbed by third‐ and fourth‐power terms in the potential‐energy expression. For favorable cases agreement between upper and lower bounds is easily carried to many more figures than are likely to be physically significant. In many cases the lower bounds agreed more closely to the true eigenvalue than did the corresponding upper bounds. For a given basis set, this method gives closer bounds than that of Bazley and Fox, except for energy levels too high to be satisfactorily treated in the given basis. The only disadvantage found was that for close bounds double precision proved necessary, indicating more than ordinary loss of computational accuracy.

The bracketing theorem in the partitioning technique for solving the Schrödinger equation may be used in principle to determine upper and lower bounds to energy eigenvalues. Practical lower bounds of any accuracy desired may be evaluated by utilizing the properties of ``inner projections'' on finite manifolds in the Hilbert space. The method is here applied to the ground state and excited states of a Hamiltonian H=H(sub 0)+V having a positive definite perturbation V. Even if inspiration is derived from the method of intermediate Hamiltonians, the final results are of bracketing type and independent of this approach. The method is numerically illustrated in some accompanying papers.

This paper, following a brief introduction, is divided into five parts. Part I outlines the theory of the molecular orbital method for the ground, ionized and excited states of molecules. Part II gives a brief summary of the interaction integrals and their tabulation. Part III outlines an automatic program designed for the computation of various states of molecules. Part IV gives examples of the study of ground, ionized and excited states of CO, BH and N2 where the program of automatic computation and molecular integrals have been utilized. Part V enlists some special problems of Molecular Quantum Mechanics are being tackled at New York University.

The methods developed in a previous paper for extending the neutron scattering formalism of Zemach and Glauber to any type of molecular rotator have been employed to derive generalized forms generalized forms of the differential cross sections for rotator scattering. A mass-ratio expansion for the treatment of the high-energy limit is illustrated on the classical cross section and then employed in the treatment of the more general quantum-mechanical expression for the differential cross section. The results apply to an arbitrarily asymmetric rotator. The very low energy approximation is carried out for the symmetric rotator, and the procedure is compared with the explicit summing of the partial cross sections for individual rotational transitions. The inelastic correction to the static approximation for interference scattering is calculated to an accuracy of first order in the mass ratios for the case of the symmetric rotator.

This method is investigated for semi-infinite multiple-slab configurations of arbitrary width, composition, and source distribution. Isotropic scattering in the laboratory system is assumed. Isotropic scattering implies that the fraction of neutrons scattered in the i(sup th) volume element or subregion that will make their next collision in the j(sup th) volume element or subregion is the same for all collisions. These so-called "transfer probabilities" between subregions are calculated and used to obtain successive-collision densities from which the flux and transmission probabilities directly follow. For a thick slab with little or no absorption, a successive-collisions technique proves impractical because an unreasonably large number of collisions must be followed in order to obtain the flux. Here the appropriate integral equation is converted into a set of linear simultaneous algebraic equations that are solved for the average total flux in each subregion. When ordinary diffusion theory applies with satisfactory precision in a portion of the multiple-slab configuration, the problem is solved by ordinary diffusion theory, but the flux is plotted only in the region of validity. The angular distribution of neutrons entering the remaining portion is determined from the known diffusion flux and the remaining region is solved by higher order theory. Several procedures for applying the numerical method are presented and discussed. To illustrate the calculational procedure, a symmetrical slab ia vacuum is worked by the numerical, Monte Carlo, and P(sub 3) spherical harmonics methods. In addition, an unsymmetrical double-slab problem is solved by the numerical and Monte Carlo methods. The numerical approach proved faster and more accurate in these examples. Adaptation of the method to anisotropic scattering in slabs is indicated, although no example is included in this paper.