[ODE] Randomness in ODE

[Martin C. Martin]

I use ODE for evolving virtual creatures, and a creature isn't mutated,
then in the next generation it goes through exactly the same steps and
ends up in exactly the same location (roundoff errors and all), and
therefore gets exactly the same fitness.  Also, if I print it out to a
text file at the end of the run, the load it in at the beginning of the
next run, it again goes through exactly the same steps and gets exactly
the same result.

I'm not using all parts of ODE, but I suspect even the parts I'm not using
work this way.  But I can't say for sure...

- Martin

Kenneth Holmlund wrote:
> 
> I know ODE is deterministic when the same app is started
> in several instances and compared.
> This means that are no relevant beta decay in RAM :)
> In addition it means that there are no memory bugs
> (i.e. the app doesn't read random data from mis-reference
> memory).
> 
> However, is ODE also deterministic when the simulation
> is reset from the app, i.e. when you go back to the orignal
> state of the system without restarting the app?
> 
> My experience is that this often is a problem in physics apps,
> since  auxiliary parameters not always are reset, and because
> of rounding errors relating to ascii/float casts etc.
> 
> /Ken
> 
> Martin C. Martin wrote:
> 
> >ODE is completely deterministic, at least the parts that I'm using...
> >
> >- Martin
> >