[SciPy-user] solve for symetric matrix ?

Nils Wagner nwagner@iam.uni-stuttgart...
Thu Apr 26 04:43:54 CDT 2007


fred wrote:
> Nils Wagner a écrit :
>   
>> fred wrote:
>>   
>>     
>>> Nils Wagner a écrit :
>>>   
>>>     
>>>       
>>>> If your matrix is also positive definite you may use
>>>> linalg.cho_factor
>>>> linalg.cho_solve
>>>>       
>>>>         
> By construction, they are positive definite.
>
> But just for curiosity, I did not find any method to test directly it.
>
> Ok, I could compute eigvals but other idea ?
>
> Cheers,
>
>   
Hi Fred,

If the Cholesky decomposition doesn't exist, the matrix is not positive
definite.
For further information, see

S.~M. Rump,
Verification of positive definiteness, BIT Numerical Mathematics,
46:433--452, 2006.
http://www.ti3.tu-harburg.de/cgi-bin/smrbibsearch/publications/ti3.html?author=rump

Cheers,
Nils


More information about the SciPy-user mailing list