[SciPy-User] linear algebra: quadratic forms without linalg.inv

Souheil Inati souheil.inati@nyu....
Mon Nov 2 08:50:36 CST 2009



On Nov 2, 2009, at 9:46 AM, Bruce Southey wrote:

> On 11/01/2009 09:03 PM, Sturla Molden wrote:
>> josef.pktd@gmail.com skrev:
>>
>>> In econometrics (including statsmodels) we have a lot of quadratic
>>> forms that are usually calculate with a matrix inverse.
>>>
>> That is a sign of numerical incompetence.
>>
> By whom? :-)
> Sure there are cases that just require solving of a linear system when
> inverses perhaps should not be used. But there are a lot of other  
> cases
> especially statistical that you require an inverse such as getting
> standard errors and using solving algorithms that are way faster than
> than those that do not use inverses.
>
> Although, statistical problems are usually bad numerically, some of  
> the
> issues like speed and precision get further away with modern 64-bit
> cpus. Really it is getting the right tool for the job in hand.
>
> Bruce
>
Sorry, this statement is misleading.  Precision has nothing to do with  
64-bit cpus - that's just of storing bigger matrices.

-Souheil


---------------------------------

Souheil Inati, PhD
Research Associate Professor
Center for Neural Science and Department of Psychology
Chief Physicist, NYU Center for Brain Imaging
New York University
4 Washington Place, Room 809
New York, N.Y., 10003-6621
Office: (212) 998-3741
Fax:     (212) 995-4011
Email: souheil.inati@nyu.edu





More information about the SciPy-User mailing list