|
Title:
|
DEVELOPMENT OF ORTHOGONALITY OF SINGULAR VECTORS COMPUTED BY I-SVD ALGORITHM |
|
Author(s):
|
Masami Takata , Taro Konda , Kinji Kimura , Yoshimasa Nakamura |
|
ISBN:
|
978-972-8924-30-0 |
|
Editors:
|
Nuno Guimarães and Pedro Isaías |
|
Year:
|
2007 |
|
Edition:
|
Single |
|
Keywords:
|
singular value decomposition, singular vector, orthogonality, twisted factorization |
|
Type:
|
Short Paper |
|
First Page:
|
437 |
|
Last Page:
|
442 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Recently an )(2mO algorithm named Integrable-Singular Value Decomposition (I-SVD) for bidiagonal singular value decomposition is developed. Here m is the dimension size. The modified discrete Lotka-Volterra (dLV) with shift is used as the fast singular value computation. Then, each singular vector related to a singular value is computed independently through a twisted factorization with the dLV variable type transformation. The orthogonality of the resulting singular vectors is not efficient in general. To avoid this problem, in this paper, we improve the singular vector computation part by introducing scaling and shift techniques. Once singular values and their distribution are changed by scaling and shift, then the orthogonality of singular vectors is sufficiently. |
|
|
|
|
|
|
