Post Reply 
Matrix Inversion ANA
06-16-2016, 10:57 AM
Post: #10
RE: Matrix Inversion ANA
(06-16-2016 10:27 AM)Werner Wrote:  Even without a temporary vector of length n (I suppose you're not referring to the permutation vector)

The extra vector of length n is the permutation vector contain indices that define the partial pivots.

I believe, but am not certain, that these values are stored in the matrix descriptor itself rather than occupying other registers. Each permutation index fits in a nibble and there is plenty of space in the matrix description. Look at Synthetic Matrices. This means that the matrix can be quickly converted from the LU decomposition back into its original form. E.g. solving a system of equations doesn't destroy the matrix but there isn't space to store a second copy containing the decomposition -- I suspect it is converted in-place, the equations are then solved and it is converted back (possibly later).


Pauli
Find all posts by this user
Quote this message in a reply
Post Reply 


Messages In This Thread
Matrix Inversion ANA - churichuro - 06-12-2016, 06:29 PM
RE: Matrix Inversion ANA - Joe Horn - 06-12-2016, 09:30 PM
RE: Matrix Inversion ANA - churichuro - 06-13-2016, 03:27 PM
RE: Matrix Inversion ANA - rprosperi - 06-13-2016, 06:30 PM
RE: Matrix Inversion ANA - churichuro - 06-14-2016, 10:55 PM
RE: Matrix Inversion ANA - rprosperi - 06-15-2016, 01:32 AM
RE: Matrix Inversion ANA - Paul Dale - 06-16-2016, 08:12 AM
RE: Matrix Inversion ANA - churichuro - 06-16-2016, 12:23 AM
RE: Matrix Inversion ANA - Werner - 06-16-2016, 10:27 AM
RE: Matrix Inversion ANA - Paul Dale - 06-16-2016 10:57 AM
RE: Matrix Inversion ANA - Werner - 06-16-2016, 11:06 AM
RE: Matrix Inversion ANA - Paul Dale - 06-16-2016, 11:20 AM
RE: Matrix Inversion ANA - Werner - 06-16-2016, 11:21 AM
RE: Matrix Inversion ANA - Paul Dale - 06-16-2016, 09:45 PM



User(s) browsing this thread: 1 Guest(s)