Re: [boost] uniform_real rather alarmingly non-random?

From: Manfred Doudar (manfred.doudar_at_[hidden])
Date: 2006-03-19 02:35:39

John Maddock wrote:
>>If memory serves, I've hit on this before - but I don't think you'll
>>like answer - you're going to have to push the variate_generator gen
>>(, and its dependants rnd, and ur_a) into the global namespace.
> 
> 
> Sorry but I don't understand that, how does that influence runtime 
> behaviour?
> 
> John. 
> 

Wish I could say definitively, but I would not seemingly get randomness 
otherwise - never looked for the reasons why at the time, but it worked 
- though like yourself I don't see why it would influence the runtime 
behaviour of the beast.

However, the reasons posted by others on the list seem to sound 
plausible - I on the other hand did only recant from experience upon 
artificially manufacturing wave burst profiles from wave spectra using 
an IFFT, and do recall that the documentation for the mersene twister 
did the same.

Cheers,
-- 
Manfred
19.03.2006