www.neuron.yale.edu

I'm interested in the single channel simulations that are possible using NEURON. However, I have concerns about the stochastic nature of the simulation.

For instance: I can load the NEURON 5.8 demo, select "Stochastic single channel:HH", and select " 1 channel". When I "Init and Run" I see a voltage transient that changes when a channel opens. Every run seems to reflect random channel openings which are dependent upon the rate constants.

However, if I load more than one instance of the NEURON demo and set each up as described about, I see the same voltage response for corresponding runs (e.g. Instance1/Run1 = Instance2/Run1, and Instance1/Run2 - Instance2/Run2, etc.). This holds true for each run I have tested (200 runs of 200 msec each).

How does NEURON achieve randomness in the stochastic simulator? Should I randomly sample sweeps from a population of simulation runs to achieve true stochastic behavior?

Thanks.

Use a different random number seed for each instance and/or simulation run.
See
http://www.neuron.yale.edu/neuron/stati ... andom.html

A reasonable strategy is to create a set of seeds in advance. For the sake of
reproducibility, you may also want to record a log file that contains important info
for each run, such as parameter values including the value of any random seed(s)
that were used.