Repository logo
  • English
  • 中文
Log In
Have you forgotten your password?
  1. Home
  2. College of Science / 理學院
  3. Mathematics / 數學系
  4. Efficient Scheme for Computing the k-th Eigenvalue
 
  • Details

Efficient Scheme for Computing the k-th Eigenvalue

Date Issued
2014
Date
2014
Author(s)
Xia, You-Huai
URI
http://ntur.lib.ntu.edu.tw//handle/246246/264031
Abstract
In some applications of physics problems, we may want to know certain eigenvalues we''re interested in. Here our target is an arbitrarily chosen k-th eigenvalue of a generalized eigenvalue problems. According to the Sylvester''s law of inertia in Linear algebra, we may get the eigenvalue counts before an arbitrarily given number via $LDL^T$ decomposition. If we view the eigenvalue counts before a given number as a step function f, then our problem of finding the k-th eigenvalue can be transformed into a root-finding problem ( solving $f(x)=k$ for x ). As far as our problem concerned, our goals is to find a real value $s$ such that $f(s)=k$ or $f(s)=k-1$ so that we can find the exact k-th eigenvalue by an shift-invert eigenvalue solver. In this paper, we do not emphasize the choice of a specific eigenvalue solver, but focus on developing the root-finding algorithm and introduce an approximating method which can reduce time consuming relative to $LDL^T$ decomposition. In the literature, the Bisection method is an existing stable root-finding algorithm which can be applied to step functions. Here we will first extend the Bisection method to the Multiple-section method which can be naturally parallelized. We will also developed a sequential method based on piecewise Monotone cubic interpolation and then combine it with Multiple-section method to get a parallel algorithm. With the same number of MPI processes, it can reduce the total number of function evaluations for one time root-solving in most cases relative to the pure Multiple-section method. On the other hand, besides computing the exact function value via $LDL^T$ decomposition, we also took a stochastic scheme for estimating the eigenvalue counts as an alternative to achieve the aim for reducing the time consuming for one-time function evaluation. The result of numerical experiment shows that in the case of large sparse matrix, we can save function evaluation time by sacrificing some exactness of function value.
Subjects
廣義特徵值問題
特徵值估計
特徵值個數估計
分段單調三次插值
西爾維斯特慣性定理
Type
thesis
File(s)
Loading...
Thumbnail Image
Name

ntu-103-R00221024-1.pdf

Size

23.54 KB

Format

Adobe PDF

Checksum

(MD5):d7e7a81ee1a08676aa4a6cef22c3d05b

臺大位居世界頂尖大學之列,為永久珍藏及向國際展現本校豐碩的研究成果及學術能量,圖書館整合機構典藏(NTUR)與學術庫(AH)不同功能平台,成為臺大學術典藏NTU scholars。期能整合研究能量、促進交流合作、保存學術產出、推廣研究成果。

To permanently archive and promote researcher profiles and scholarly works, Library integrates the services of “NTU Repository” with “Academic Hub” to form NTU Scholars.

總館學科館員 (Main Library)
醫學圖書館學科館員 (Medical Library)
社會科學院辜振甫紀念圖書館學科館員 (Social Sciences Library)

開放取用是從使用者角度提升資訊取用性的社會運動,應用在學術研究上是透過將研究著作公開供使用者自由取閱,以促進學術傳播及因應期刊訂購費用逐年攀升。同時可加速研究發展、提升研究影響力,NTU Scholars即為本校的開放取用典藏(OA Archive)平台。(點選深入了解OA)

  • 請確認所上傳的全文是原創的內容,若該文件包含部分內容的版權非匯入者所有,或由第三方贊助與合作完成,請確認該版權所有者及第三方同意提供此授權。
    Please represent that the submission is your original work, and that you have the right to grant the rights to upload.
  • 若欲上傳已出版的全文電子檔,可使用Open policy finder網站查詢,以確認出版單位之版權政策。
    Please use Open policy finder to find a summary of permissions that are normally given as part of each publisher's copyright transfer agreement.
  • 網站簡介 (Quickstart Guide)
  • 使用手冊 (Instruction Manual)
  • 線上預約服務 (Booking Service)
  • 方案一:臺灣大學計算機中心帳號登入
    (With C&INC Email Account)
  • 方案二:ORCID帳號登入 (With ORCID)
  • 方案一:定期更新ORCID者,以ID匯入 (Search for identifier (ORCID))
  • 方案二:自行建檔 (Default mode Submission)
  • 方案三:學科館員協助匯入 (Email worklist to subject librarians)

Built with DSpace-CRIS software - Extension maintained and optimized by 4Science