Dear Hans, thank you for your reply and apologies for not giving more detail on my simulation. You are absolutely correct, i create a feed forward network of three serially connected hypercolumns, each HC contains 50 minicolumns of 30 neurons each - so 4,500 neurons in total. I send a signal into this network, and the number i present is a measure of the signal exiting the network - although its not a spike rate it is very similar (i send the signal into the network with a sloped distribution, and measure the slope of the signal exiting the network).
I am intrigued by your comment and they are very helpful as i am not educated on the internal architecture of Nest. My understanding that running "srun -n 10 python3 run.py" on the PDC would create 10 independent executions, but I understand from what you are saying that Nest is still able to communicate across them? While this seems to be a very useful feature, for the purpose of this particular simulation is there a possibility to avoid this?
many thanks and regards Johan