Re: simple matrix problem

[Dmitri A. Sergatskov]

address@hidden wrote:
> On Nov 5, 2003 at 2:47pm, Geraint Paul Bevan wrote:
>
> g.beva >data = [ data ( : , 1 ) , detrend ( data ( : , 2 ) ) ]
>
> Not to knit-pick, but an in-place assignment is faster than
> re-allocating memory via the [...] operator:
>
> ---
> octave> version
> ans = 2.1.50
> octave> A = rand(50,50);
> octave> tic;A(:,1) = rand(50,1);toc
> ans = 0.0013770
> octave> tic;A = [ rand(50,1), A(:,2:50)];toc
> ans = 0.0026010
> ---

Though I do not disagree with the conclusion I do not think those number
make much sense (tic;toc timing is not that accurate):

address@hidden dima]$ octave
GNU Octave, version 2.1.50 (i686-pc-linux-gnu).
...
octave:1> A = rand(50,50);
octave:2> tic;A(:,1) = rand(50,1);toc
ans = 0.0082260
octave:3> tic;A = [ rand(50,1), A(:,2:50)];toc
ans = 0.00092900
octave:4>


> It's almost exactly twice as fast. Sorry Geraint, this is just my
> curiosity running amok! :-)

So [...] almost 10 times as fast! Sure...

You need to make a test case that takes few seconds, eg.:
octave:5> A = ones(5000,5000);
octave:6> tic;A(:,1) = rand(5000,1);toc
ans = 0.0099300
octave:7> tic;A = [ rand(5000,1), A(:,2:5000)];toc
ans = 28.603

> ~Tomer

Regards,

Dmitri.



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