[gate-users] Random Number files

[Mutanga Theodore]

Hello there, I did exactly like this and I am pondering about the same
question concerning the degree of correlation of the seeds. Like most
other things concerning GATE we have to find out for ourselves. Funny
though that the developpers have made batch runs and some of them are
following these posts.
I'll do a post if I find something out.
cheers

Theo

> Hello
>
> I am having the same question.  What I have in mind is this.  To
> generate different endOfRun.rndm files for different jobs, run a short
> (mock) acquisition (unrelated to the experiment we are doing), at the
> end, copy endOfRun.rndm to endOfRun_4Job1.rndm.  Run the short
> acquisition again, with
>
> /random/resetEngineFrom endOfRun.rndm
>
> at the end of this, a different endOfRun.rndm is created, copy this to
> endOfRun_4Job2.rndm.  Do this a number of times, depending on how many
> jobs I want to run in parallel.  Then in the real experiment, use
>
> /random/resetEngineFrom endOfRun_4job1.rndm  (in job 1)
> /random/resetEngineFrom endOfRun_4job2.rndm (in job 2)
>
> and so on.  Now my question is, I am not sure the different .rndm files
> generated this way will make the parallel runs truly uncorrelated, since
> they are generated sequentially from the same mock run, each time taking
> the previous as the initial random engine.
>
> About modifying endOfRun.rndm files, there is an earlier post saying
> modifying the seed itself is not enough.   So anyone has a simple way of
> creating uncorrelated rndm files?  Thanks.
>
> jy
> _______________________________________________
> gate-users mailing list
> gate-users at lphe1pet1.epfl.ch
> http://lphe1pet1.epfl.ch/mailman/listinfo/gate-users
>