Re: Help with 2.9.9

[David Bateman]

Tom Holroyd (NIH/NIMH) [E] wrote:
> help matrix_type has language problems. I can't understand this:
>       'This allows rapid for solutions of linear equations involving A to be 
> performed.'
>   
Grrr.. Ok.. I'd deal with it..

>
> Hmm, 1 2
>      2 4 has a zero eigenvalue but matrix_type says it's positive definite. 
> That's wrong.
>   
See my other mail on this. Try in particular

R = [1,2;2,4];
matrix_type(R)
x = R \ ones(2,1);
matrix_type(R)

and see what you get... The probing code for PD is fast but
approximative. If it thinks it has a PD matrix it calls the cholesky
solver, then failing it calls the LU solver, and it also fails and so it
finally calls the minimum norm solver...

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