Responsive image
博碩士論文 etd-0619118-103715 詳細資訊
Title page for etd-0619118-103715
論文名稱
Title
矩陣QR分解揭示低秩與高秩演算法的探討
A study on low-rank and high-rank revealing QR algorithms
系所名稱
Department
畢業學年期
Year, semester
語文別
Language
學位類別
Degree
頁數
Number of pages
43
研究生
Author
指導教授
Advisor
召集委員
Convenor
口試委員
Advisory Committee
口試日期
Date of Exam
2018-06-28
繳交日期
Date of Submission
2018-08-14
關鍵字
Keywords
QR 分解、奇異值、數值秩
QR factorization, singular value, numerical rank
統計
Statistics
本論文已被瀏覽 5693 次,被下載 34
The thesis/dissertation has been browsed 5693 times, has been downloaded 34 times.
中文摘要
低秩或高秩矩陣的秩揭示問題在科學計算中廣泛存在,例如找低秩近似矩陣、計算主成分分析的主要子空間、非滿秩最小平方和問題或者檢測同倫法中的曲線結構等等,雖然可以應用 Golub-Reinsch SVD 算法計算所有的奇異值,從而揭示秩,但在這些情況下相對昂貴。由 Tony Chan 和 P. C. Hansen 提出的 RRQR 和 L-RRQR 算法,當 rank gap 足夠大且矩陣是高秩或低秩時,在揭示秩方面是相對有效和可靠的。而這兩個演算法的準確性和對應於最大或最小奇異值的奇異向量估計有很大的相關。在這篇論文中我們研究奇異向量的估計並更正了一個證明。
Abstract
The rank revealing problem for low rank or low nullity matrices
arises widely in scientific computing, such as finding a low rank
approximation to a matrix, computing the dominate subspace in principal
component analysis, solving rank deficient least squares problems,
and detecting the curve structure in homotopy
continuation method. Although the Golub-Reinsch SVD algorithm can be applied
to calculate all singular values and thus reveal the rank, it becomes
relatively expensive in these situations.

RRQR and L-RRQR algorithm, proposed by Tony Chan and P. C. Hansen,
are efficient and reliable methods for rank revealing when the rank gap is
well-defined and the matrix is of low rank or low nullity. The accuracy of
the methods heavily depends on the estimation of singular vector
corresponding to the largest or smallest singular value. In this thesis we
study the singular vector estimation and correct a proof in a deflation
theorem.
目次 Table of Contents
[論文審定書, i]
[摘要, ii]
[Abstract, iii]
[1 Introduction ,1]
[2 Preliminary, 5]
[3 Rank Revealing QR Factorization for High Rank Matrix, 8]
[3.1 High-rank Revealing QR algorithm, 8]
[3.2 Compute the smallest singular vector ,10]
[3.3 Deflation process in RRQR , 11]
[4 Rank Revealing QR Factorization for Low Rank Matrix , 13]
[4.1 Low-rank Revealing QR algorithm , 13]
[4.2 Compute the biggest singular value , 15]
[4.3 Deflation process in L-RRQR , 17]
[5 Numerical Experiments, 18]
[5.1 Behavior of power iteration and inverse iteration, 18]
[5.2 The distribution of diagonal elements in QR, 19]
[5.3 Comparison of QR, OCP, and RRQR ,24]
[5.4 Comparison of QR, OCP, and L-RRQR , 26]
[5.5 Conclusion , 28]
[6 Applications , 29]
[6.1 Finding low rank approximation, 29]
[6.2 Solving the rank deficient least squares problem, 29]
[6.3 Subset selection problems , 30]
[7 Appendix 32]
[7.1 Pseudoinverse of a triangular matrix , 32]
[7.2 A posteriori bounds on singular values , 34]
參考文獻 References
[1] C. H. Bischof and G. Quintana-Orti, Algorithm 782: Codes for rank-revealing QR factorizations
of dense matrices, ACM Trans. Math. Software, 24, pp. 254–257, 1998.
[2] Bj¨orck, Numerical Methods for Least Squares Problems, SIAM, Philadelphia, 1996.
[3] Tony F. Chan, Rank Revealing QR Factorizations, Lin. Alg. Appl., 88/89, pp. 67–82, 1987.
[4] T. R. Chan and P. C. Hansen, Low-rank revealing QR factorizations, Numer. Lin. Alg.
Appl., 1, pp. 33–44, 1994.
[5] B. H. Dayton and Z. Zeng, Computing the multiplicity structure in solving polynomial systems,
Proceedings of ISSAC ‘05, ACM Press, pp. 116–123, 2005.
[6] Z. Drmaˇc and Z. Bujanovi´c, On the Failure of Rank-Revealing QR Factorization Software
- A Case Study, ACM Trans. Math. Software, 35, Article 12, 2008.
[7] G. H. Golub and C. F. Van Loan, Matrix Computations, 3rd ed., Johns Hopkins Univ. Press,
Baltimore, MD, 1996.
[8] R. A. Horn and C. R. Johnson, Matrix Analysis, Cambridge University Press, 1985.
[9] R. Hanson and C. Lawson, Solving Least Squares Problems, Prentice-Hall, 1974.
[10] T. M. Hwang, W. W. Lin and E. K. Yang, Rank-revealing LU Factorization, Lin. Alg.
Appl., 175, pp. 115–141, 1992.
[11] C. T. Pan, On the existence and computation of rank-revealing LU factorizations, Lin. Alg.
Appl., 316, pp. 199–222, 2000.
31
電子全文 Fulltext
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。
論文使用權限 Thesis access permission:自定論文開放時間 user define
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus: 已公開 available


紙本論文 Printed copies
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。
開放時間 available 已公開 available

QR Code