Re: consistent crash using gammrnd

[David Bateman]

Daniel Oberhoff wrote:
> Hehe, I was hoping one could mod some of the octave test suite for this.
>
> I suppose we would have to set up a scan of all the functions, i.e.
> input sets for each parameter, run that in octave and matlab, record
> the result into mat files and then measure. shouldn't actually be that
> hard. We need a list of functions and sensible parameter ranges. once
> that is all there this can be automated with shell scripts. the
> m-files would be tha same for both sides (well, taking care to be
> protable).
>
> Daniel

The list of functions to test is fairly easy to get..

cd scripts/statistics/distributions
export files=`ls *{inv,cdf,pdf}.m`

you might also treat the *rnd.m files, though the tests will be harder
as the output depends on the base random generator and as these have
different sequences in Octave and Matlab, the *rnd.m will not give the
same output..

As for the valid ranges of the input parameters, that will take a bit
more intelligence,,,

D.



-- 
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary