Re: Calculation issue with octave (data unexpectedly go to zero at some point)

[Michael Goffioul]

On Sat, Feb 22, 2014 at 11:59 AM, rchretien <address@hidden> wrote:

Hi everybody,

I contact you because of my poor knowledge of octave. For my master thesis,
I have to use Octave installed on a remote cluster whilst I am running
matlab at home. For the same code, matlab and octave do not produce the same
results. Or rather, should I precise, they produce the same results during a
certain amount of time, until octave produces an unexpected collapse of the
data to zero.

Here is an example of what happens :

http://hpics.li/3e16497

Under matlab (I repeat that I have used the same version of the code), the
output is basically the same, except for abscissae greater than ~975 (i.e.
the abscissa from where octave collapses) where no crash is observed
(contrarily to what can be seen here above). Roughly speaking, the code
behaves as expected when run under matlab and not when run under octave.

My question is, since this is manifestedly not a coding problem (in which
case, the output from matlab would also collapse), is this problem linked to
an option, or some kind of precision, or to a version problem, in octave ?
Has this problem already been observed ? If so, how can I fix it ?

I’m looking forward to reading your opinions and suggestions,

Just a general suggestion.

If you want useful answers, you should post sample code that illustrates the problem, so that others can reproduce it, and more importantly understand it. Otherwise, you question sounds literally like: I have a problem, how can I fix it? (without explaining what the problem is).

Michael.