Title of program: NORMAL COORDINATE ANALYSIS threshold serial Jacobi method [31and the Householder procedure [4]. Although the latter is faster than the Jacobi Catalogue number: ACKJ method, the eigenvectors are more accurate in the case of the Program obtainable from: CPC Program Library, Queen's threshold serial Jacobi procedure. University of Belfast, N. Ireland (see application form in this issue) Restriction on the complexity of the problem Computer: CYBER 74; Installation: State University Gronin-The program is dimensioned for 20 atoms per unit cell. The number of degrees of freedom may be increased to 99, degen pending on the available storage capacity. Operating system: SCOPE 3.4.1., level 373 Program language used: FORTRAN 4.1. Typical running time For the eigenvalue problem solved by threshold serial Jacobi High speed store required: 155k (octal) method for KNiF 3 (without iteration procedure): 12.6 sec. No. of bits in a word: 60 Unusual features of the program Overlay structure: None Using the symmetry coordinates, it is possible to diagonalize No. of magnetic tapes required: None the dynamical matrix in blocks. The symmetry coordinates Other peripherals used: Card reader, line printer can be calculated in some cases by the method and the program given by Warren and Worlton [5] , using a reducible Number of cards in combined program and test deck: 1325 multiplier representation of the point group of the wavevec-Card punching code: IBM(029) EBCDIC
tor (k = 0). However, the input of the symmetry coordinates is not necessary for our program. Block-diagonalization of Keywords: Solid state, dynamical matrix, normal coordinate the dynamical matrix set up in cartesian coordinates is often analysis, infrared spectra analysis, eigenvalue problem, threshnot very useful for solving the elgenvalue problem. old Jacobi method, Householder procedure.
โข References Nature of the physical problem
[1] T. Shimanouchi, M. Tsuboi and T. Miyazawa, J. Chem. A program has been written for solving the vibrational secular Phys. 35 (1961) 1597. equation in cartesian coordinates [1] and for adjusting a set [2] T. Shimanouchi and I. Suzuki, J. Chem. Phys. 42 (1965) of f~e~ctnstantsto give a fit of calculated and observed 296. frequencies [2] which are measured by means of infrared [3] J .H. Wilkinson, The Algebraic Eigenvalue Problem (Clarentransmission spectroscopy. don Press, Oxford, 1965) p. 277. [4] A.S. Householder,J. Ass. Comp. Mach. 5 (1958) 205-243, Method of solution 335-342. Two methods for calculating the eigenvalues and eigenvectors [5] J.L. Warren and T.G. Worlton, Computer Phys. Commun. are included in the program deck. These methods are: the 8 (1974) 71-84.