www.neuron.yale.edu

I have a network model with two separate population of neurons without any connection with each other. I am running the model in mpi environment. Population 1 rests at 23 C and population 2 at 30 C. I assign their temperature using the keyword 'celsius'. I observe that the spiking behivour of Population 1 is drastically affected (frequency and spike shape changes). The spiking behaviour of population 1 is restored to its control value when I make the both the population resting at same temparature or when I restrain using keyword 'celsius' in Population 2( Here I declare the temparature of Population 2 in its mechanisms mod files using a different variable (not 'celsius')). What do you think of this ? Isn't it odd that the usage of keyword 'celsius' in one population affects the behaviour of other eventhough they are not connected with each other.

As far as I'm aware, celsius is a global variable in hoc, meaning it has the same value everywhere in your NEURON session.
If you want to assign different temperatures to different populations, you could declare a RANGE variable 'local_temperature' in all your NMODL files that are affected by temperature, set the temperature value for each instance of this mechanism separately and replace the keyword celsius in the NMODL files with 'local_temperature' (most likely in various rate functions...).
Not sure if this is the best/most elegant solution, but it should accomplish what you want...

regger wrote:As far as I'm aware, celsius is a global variable in hoc, meaning it has the same value everywhere in your NEURON session.

True.

If you want to assign different temperatures to different populations, you could declare a RANGE variable 'local_temperature' in all your NMODL files that are affected by temperature, set the temperature value for each instance of this mechanism separately and replace the keyword celsius in the NMODL files with 'local_temperature' (most likely in various rate functions...).

This would work.