Re: [Help-gsl] Usage of GSL random in parallel code [OpenMP]

[Peter Johansson]

On 05/06/2014 02:17 AM, Matthias Sitte wrote:
> 3) You have to properly initialize the GSL RNGs. There are different 
> ways to do that. I like using microseconds obtained from time 
> functions. Beware that if you initialize using only the thread number, 
> then subsequent calls of the same program will always generate the 
> same result. If you'd use some time function, there would be some 
> "random" element to it, so you're (much) less likely to generate the 
> same result running the program twice. 
Hi Mattias,

This is the desired behaviour in many research applications when the 
reproducibility is crucial.

Cheers,

--
Peter Johansson