This article presents an efficient and numerically stable algorithm, along with a complete listing of the associated computer program, developed for the accurate computation of specified roots and associated vectors of the eigenvalue problem Aq = λBq with band symmetric A and B, B being also positive definite. The desired roots are first isolated by the Sturm sequence procedure; then a special variant of the inverse iteration technique is applied for the individual determination of each root along with its vector. The algorithm fully exploits the banded form of relevant matrices, and the associated program written in FORTRAN V for the JPL UNIVAC 1108 computer proves to be most significantly economical in comparison to similar existing procedures.
The program may be conveniently utilized for the efficient solution of practical engineering problems including free vibration and buckling analysis of structures. Results of such analyses are presented for representative structures.
[1]
J. H. Wilkinson.
The algebraic eigenvalue problem
,
1966
.
[2]
James Hardy Wilkinson,et al.
Eigenvalues of Ax = λBx with band symmetric A and B
,
1969,
Comput. J..
[3]
K. Gupta.
Free Vibrations of Single-branch Structural Systems
,
1969
.
[4]
K. K. Gupta.
Vibration of frames and other structures with banded stiffness matrix
,
1970
.
[5]
G. Peters,et al.
$Ax = \lambda Bx$ and the Generalized Eigenproblem
,
1970
.
[6]
K. K. Gupta.
Solution of eigenvalue problems by Sturm sequence method.
,
1972
.
[7]
K. K. Gupta.
Free vibration analysis of spinning structural systems
,
1973
.
[8]
K. K. Gupta.
Numerical solution of quadratic matrix equations for free vibration analysis of structures
,
1973
.