Optimizer : Great

[Kristian Nielsen]

> Date: Thu, 30 Sep 1999 18:15:37 +0200
> From: Farhi <farhi at ill.fr>

> I now have an optimizer that increases count numbers on detector by
> about a factor 40 to 50 !

That sounds very nice indeed! Do you have some idea where the neutrons
are getting lost when the optimizer is not used (guide absorption,
missing monochromator or sample, wrong energy, ...)?

> What's long in computation is the random generator. ABSORB comes after,
> and so do not affect drastically the computation time (perhaps a factor
> 2). A better solution is to use the 'bad' random numbers and to convert
> them to some more effective values. That's what I do.

Yes, this problem is actually why I choose to integrate the optimizer in
the source component; just by generating the useless neutrons you are
already wasting a log of computation time, even more so since most
neutrons usually get lost early in the instrument. But this also sounds
like I may need to look at optimizing the random number generation
routine (in fact I already put in a better-quality random number
generator in the version you are now using).

But anyway, I am very much looking forward to seeing your code when it
is ready. Have you considered writing some documentation that would make
your optimizer suitable for inclusion in the McStas manual and thereby
in the official release?

 - Kristian.