random insertion points

Submitted by jordantcox on Mon, 04/10/2017 - 11:30

Hello,

As I understand it LIGGGHTS uses a seed value to generate the random positions for insert/rate/region and other insertion methods. I would like to evaluate averages of a simulation. To do this I would like to able to re-run the same simulation, but have it produce different insertion coordinates each time the simulation is run. From what I understand LIGGGHTS is deterministic such that if the seed value remains the same, the insertion points for particles remain the same. I would just like to be able to go into the LIGGGHTS src folder and alter the random number generator so it produces its own seed each time rather than take in an exact seed so I can produce some statistical results from the simulation.

I attended the CFDEM training session and when asking about this I was told this method of defining the seed value was implemented such that the code could be tested using a series of pre-defined values. It would be nice if there was a command that could be entered that switched this feature on and off.

-Cheers,
Jordan

j-kerbl's picture

j-kerbl | Mon, 04/10/2017 - 17:45

Hi Jordan,

a quick workaround for you might be to put a selection (also random) into an outside script, e.g. a combination of bash and octave. From there you can call LIGGGHTS with the -var flag to hand-over the seeds as variables, which can be used then within the script.

Cheers,
Josef

jordantcox | Thu, 04/13/2017 - 10:34

Thanks everyone,

I am working on it and will add more as it works out.