Slow performance on linux - my error

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

Slow performance on linux - my error

Evan Thomas-2
I have posted twice to this list with a problem where Octave under Linux
is unusably slow solving ODEs. (Once with an erroneous solution!) John
Eaton has solved this problem for me and I include part of his e-mail
below, for the list's interest.

Thanks to the people who helped me with this.


John W. Eaton wrote:

> [cut]
> I tried your code on a Linux system and got similarly strange results.
> I noticed that the number of time that the function sEPSP was called
> varied from run to run, which seemed very odd.  Then I tried it on an
> Alpha system and got a floating point exception (happening inside
> LSODE).  In tracking that down, I found that your state vector s has 4
> elements, but you only use 3, and you only compute 3 derivatives.
> When I changed the line
>   s = [0, 0, 0, 0];
> to be
>   s = [0, 0, 0];
> I got consistent results.
> Obviously, Octave should check to make sure that the user-supplied
> function computes as many derivatives as there are states, so I've
> added some code to do that.
> [cut]
Evan Thomas
Department of Anatomy & Cell Biology
University of Melbourne
Parkville, 3052
ph: 9344-5849  fax: 9347-5219