On May 14, 2014, at 9:39 AM, Ernesto Carrella <[log in to unmask]> wrote:

> 1) Do you think it's a good idea as long as we are still using an older version of MASON to pulse the MersenneTwisterFast ourselves? Should we do so by calling getNewInt() 1249 times?

It's a very minor thing: the chance that it'll effect your simulation results is extremely small.  But for a library I want to be extra conservative.  I wouldn't worry about it.


> 2) I don't know how common it is for other modellers but I often give each agent its own mersenne twister fast randomizer to decouple their random behavior from the activation time (and potentially from the rest of the model as well). These randomizers wouldn't be touched by state.start().

This is unwise.  You'd have to be *extremely* careful, making sure that each of these random number generators have their own seed.  Generally speaking you should *never* use more than one random number generator, ever, unless you have certain multithreading issues.

I don't know what you mean by "decoupling" from "activation time" -- a random number generator is supposed to produce *random* numbers.  Its sequence shouldn't be coupled to anything.

Sean