[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: matrix inversion

**From**: |
David Bateman |

**Subject**: |
Re: matrix inversion |

**Date**: |
Wed, 22 Nov 2006 12:45:00 +0100 |

**User-agent**: |
Thunderbird 1.5.0.7 (X11/20060921) |

Gorazd Brumen wrote:
>* Hello,*
>
>* I have a square matrix that is composed of N^2 block matrices, where*
>* every block matrix is relatively large t x t diagonal matrix (both*
>* N and t large), i.e.*
>
>* A = [ A_{11}, A_{12} , ... A_{1N}; A_{21} ... A_{2N} ...],*
>* where every A_{ij} is a diagonal matrix of size txt.*
>
>* My question is the following: How many operations would*
>* spinv need to do this inversion?*
>
>
>* Thanks and regards,*
>* Gorazd*
>
>* *
What does dmperm on this matrix return? I suspect from the above that it
a block-triangular factorization might be appropriate in this case as a
means of accelerating the inversion. If this is the case, then blocks
returned by dmperm can be used to sub-section the matrix, and these
sub-blocks might be inverted separately and used to reconstructed the
full inverted matrix.
D.
--
David Bateman address@hidden
Motorola Labs - Paris +33 1 69 35 48 04 (Ph)
Parc Les Algorithmes, Commune de St Aubin +33 6 72 01 06 33 (Mob)
91193 Gif-Sur-Yvette FRANCE +33 1 69 35 77 01 (Fax)
The information contained in this communication has been classified as:
[x] General Business Information
[ ] Motorola Internal Use Only
[ ] Motorola Confidential Proprietary