Re: Precision

[Diego Ruiz]

Thakns for your answer.

I'm facing the following problem:

A is an M-dimensional matrix (M^S x M^S x M). Just consider the 
case of M=3 and S=5, we have a matrix with 177147 cells, with 
double precision this means we need around 3MB of free memory or 
storage space. If we let M=4 that number becomes overwhelming.
With single precision this requirement is importantly reduced.

Actually lots of cells in A are zero. That was the reason for 
which I wanted to work with SPARSE matrices. But when trying to 
install Octave Forge in my Fedora Core 2 OS I faced several 
problems (you can check the discussion held some days ago on the 
topic).

I've not expertise in Unix/Linux operative systems and solving 
those problems turns out to be quite challenging. That's the 
reason for which considered the idea of using single precision. 
Do you have a better idea?


Thank you,
Diego Ruiz

******************************************************
                 Diego Ruiz Hernández
               www.econ.upf.es/~diruiz

Universitat Pompeu Fabra,  Dept. d'Economia i Empresa
          c/ Ramon Trias Fargas 25-27, 08025,
                   Barcelona, ESPAÑA

Fall Term:
         Edinburgh University, School Of Management
        William Robertson Building, 50 George Square
                 Edinburgh EH8 9JY, UK
******************************************************

---------- Original Message -----------
From: "John W. Eaton" <address@hidden>
To: "Diego Ruiz" <address@hidden>
Sent: Mon, 15 Nov 2004 16:19:38 -0500
Subject: Precision

> On 15-Nov-2004, Diego Ruiz <address@hidden> wrote:
> 
> | After several failed attempts... I've abandoned my 
> idea of | working with Sparse matrices... but I keep 
> loyal to Octave.
> 
> I think that sparse matrices will be a part of the core Octave
> distribution relatively soon.
> 
> | Now my question is, how can I change the precision of 
> a Matrix? | I mean, if function "funfun" returns matrix 
> A, is there any | comand equivalent to Matlab's 
> "single(A)" for converting matrix | A to single 
> precision within the function and then, when | necessary 
> expand it again with "double(A)" for performing | operations?
> 
> Not yet, no.
> 
> Why do you need single precision?
> 
> jwe
------- End of Original Message -------



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------