www.neuron.yale.edu

Hello,

I am trying to implement in neuron a very simple model composed of a network of N neurons, each being subject to 2 synaptic noises (one inhibitory, one excitatory)
those noises are implemented by 2 NetStim for each neuron

all the NetStim in my network are seeded by calling rng.discunif(0,32000) where rng is a Random object. rng 's seed is a parameter of the program

1- Is such a seeding of the NetStim good to assure independence of the NetStims ?

2- If I do 2 successive calls to run() inside my .hoc code, the statistics of the NetStim differ. I suppose this means the NetStim don't reset their generator. Is this correct ?

Thank you very much for your time !

Guillaume

guillaume wrote:Is such a seeding of the NetStim good to assure independence of the NetStims ?

You will want to use the Random class's MCellRan4 generator. Read about it in the Programmer's Reference.
http://www.neuron.yale.edu/neuron/stati ... #MCellRan4

If I do 2 successive calls to run() inside my .hoc code, the statistics of the NetStim differ. I suppose this means the NetStim don't reset their generator

NEURON's run() does not reset the random number generators. If you want to get the same results on each run (which is essential for debugging and general program development) it is necessary to seed the reseed the ACG or MLCG, or if you're using MCellRan4 the highindex must be restored to its original value. If you have multiple instances of the Random class, this must be done for each instance.

Thank you very much for your answer, but I'm unsure what I should do:

- do I need to tell NetStim to use MCellRand4 instead of using its default rng ?
- do I need to program another source of noise based on MCellRand4 ?
- do I just go:
rng = new Random()
rng.MCellRan4()
for i=0, netStimList.count()-1
{ netStimList.object(i).seed() = rng.uniform(0,32000) }

See if this helps: http://www.neuron.yale.edu/neuron/node/59