PlanetMath (more info)
 Math for the people, by the people. Sponsor PlanetMath
Encyclopedia | Requests | Forums | Docs | Wiki | Random | RSS  
Login
create new user
name:
pass:
forget your password?
Main Menu
Owner confidence rating: Very low Entry average rating: No information on entry rating
[parent] matrix inversion lemma (Result)

These frequently used formulae allow to quickly calculate the inverse of a slight modification of an operator (matrix) $x$ , given that $x^{-1}$ is already known.

The matrix inversion lemma states that $$\left( x+ s \sigma z^* \right)^{-1} = x^{-1} - x^{-1} s \left( \sigma^{-1} + z^* x^{-1} s \right)^{-1} z^* x^{-1},$$ where $x$ , $s$ , $z^*$ and $\sigma$ are operators (matrices) of appropriate size. The formula especially is convenient if the rank of the regular $\sigma$ is 1, or small in comparison to $x$ 's rank.

This identity, involving the inverse of Schur's complement $d- z^* x^{-1} s$ (hopefully this may be easily computed) holds as well: $$\begin{bmatrix} x & s \\ z^* & d \end{bmatrix}^{-1} = \begin{bmatrix} x^{-1}+ x^{-1} s (d- z^* x^{-1} s)^{-1} z^* x^{-1} & -x^{-1} s (d-z^*x^{-1}s)^{-1} \\ -(d-z^*x^{-1}s)^{-1} z^* x^{-1} & (d-z^*x^{-1}s)^{-1} \end{bmatrix}.$$




"matrix inversion lemma" is owned by kronos.
(view preamble | get metadata)

View style:

See Also: Schur complement

Other names:  Sherman-Morrison formula, Woodbury matrix identity, Woodbury formula, rank-k correction

This object's parent.
Log in to rate this entry.
(view current ratings)

Cross-references: Schur's complement, identity, regular, rank, formula, size, matrix, operator, modification, inverse, calculate

This is version 3 of matrix inversion lemma, born on 2006-01-31, modified 2006-01-31.
Object id is 7577, canonical name is MatrixInversionLemma.
Accessed 13395 times total.

Classification:
AMS MSC47S99 (Operator theory :: Other types of operator theory :: Miscellaneous)

Pending Errata and Addenda
None.
Discussion
Style: Expand: Order:
forum policy
Another name for this by ddnly on 2006-04-01 11:15:23
Isn't this Strassen's block-wise inversion formula too?
[ reply | up ]

Interact
post | correct | update request | add example | add (any)