On the Asymptotic Behavior of Scaled Singular Value and QR Decompositions (Q3677065)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 3898119
Language Label Description Also known as
default for all languages
No label defined
    English
    On the Asymptotic Behavior of Scaled Singular Value and QR Decompositions
    scientific article; zbMATH DE number 3898119

      Statements

      On the Asymptotic Behavior of Scaled Singular Value and QR Decompositions (English)
      0 references
      0 references
      1984
      0 references
      weighted least squares problem
      0 references
      asymptotic behaviour
      0 references
      singular value decomposition
      0 references
      QR-decomposition
      0 references
      The asymptotic behaviour of the singular value decomposition of a matrix X, if some of the columns are going to zero, is studied. Let \(X=(X_ 1,X_ 2)\in C_{n,p}\) and for \(t\geq 0\) \(X_ t=(X_ 1,tX_ 2)\). It is shown (Theorem 2.1): Let rank (X)\(=p\) and \(B=(X^ T_ 1X_ 1)^{- 1}X^ T_ 1X_ 2\). To each singular value \(s_ 1\) of \(X_ 1\) there is associated a unique singular value \(s_ 1^{(t)}\) of \(X_ t\) with \(s_ 1^{(t)}=S_ 1+O(t^ 2)\). If \(s_ 1\) is simple and its right singular vector is \(v_ 1\), then the corresponding right singular vector of \(X_ t\) satisfies \(v_ 1^{(t)}=\left( \begin{matrix} v_ 1+O(t)^ 2)\\ tB^ Tv_ 1+O(t^ 3)\end{matrix} \right).\) For the left singular vectors the relation \(u_ 1^{(t)}=u_ 1+O(t^ 2)\) holds. The other singular values of \(X_ t\) are of the form \(s_ 2^{(t)}=t\bar s_ 2+O(t^ 3)\), where \(\bar s_ 2\) are singular values of \(\bar X_ 2=X_ 2-X_ 1B\). Analogous results are presented for the QR-decomposition.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references