[SciPy-user] solve for symetric matrix ?

[fred]

Ronan Perrussel a écrit :
> fred a écrit :
>> But in fact, he was confused. They are negative definite (kriging 
>> matrices).
>> Bad news :-(
> Then you can solve with -A instead of A?
I tried it, of course ;-)

But cho_factor() rejects -A too.
One of the definition of the positive definite (if I understand right) 
is that the eigenvalues must be positive.

So in my case, as A has negative egv, -A has positive ones.
Although -A has positive ones, cho_factor() raise error.

-- 
http://scipy.org/FredericPetit