Access

You are not currently logged in.

Access your personal account or get JSTOR access through your library or other institution:

login

Log in to your personal account or through your institution.

If you need an accessible version of this item please contact JSTOR User Support

A Rank-One Reduction Formula and its Applications to Matrix Factorizations

Moody T. Chu, Robert E. Funderlic and Gene H. Golub
SIAM Review
Vol. 37, No. 4 (Dec., 1995), pp. 512-530
Stable URL: http://www.jstor.org/stable/2132903
Page Count: 19
  • Get Access
  • Cite this Item
If you need an accessible version of this item please contact JSTOR User Support
A Rank-One Reduction Formula and its Applications to Matrix Factorizations
Preview not available

Abstract

Let A ∈ Rm × n denote an arbitrary matrix. If x ∈ Rn and y ∈ Rm are vectors such that ω = yT Ax ≠ 0, then the matrix B := A - ω-1 AxyT A has rank exactly one less than the rank of A. This Wedderburn rank-one reduction formula is easy to prove, yet the idea is so powerful that perhaps all matrix factorizations can be derived from it. The formula also appears in places such as the positive definite secant updates BFGS and DFP as well as the ABS methods. By repeatedly applying the formula to reduce ranks, a biconjugation process analogous to the Gram-Schmidt process with oblique projections can be developed. This process provides a mechanism for constructing factorizations such as LDMT, QR, and SVD under a common framework of a general biconjugate decomposition VT AU = Ω that is diagonal and nonsingular. Two characterizations of biconjugation provide new insight into the Lanczos method and its breakdown. One characterization shows that the Lanczos algorithm (and the conjugate gradient method) is a special case of the rank-one process; in fact, these processes can be identified with the class of biconjugate direction methods so that history is pushed back by about twenty years.

Page Thumbnails

  • Thumbnail: Page 
512
    512
  • Thumbnail: Page 
513
    513
  • Thumbnail: Page 
514
    514
  • Thumbnail: Page 
515
    515
  • Thumbnail: Page 
516
    516
  • Thumbnail: Page 
517
    517
  • Thumbnail: Page 
518
    518
  • Thumbnail: Page 
519
    519
  • Thumbnail: Page 
520
    520
  • Thumbnail: Page 
521
    521
  • Thumbnail: Page 
522
    522
  • Thumbnail: Page 
523
    523
  • Thumbnail: Page 
524
    524
  • Thumbnail: Page 
525
    525
  • Thumbnail: Page 
526
    526
  • Thumbnail: Page 
527
    527
  • Thumbnail: Page 
528
    528
  • Thumbnail: Page 
529
    529
  • Thumbnail: Page 
530
    530