Realization the model of network in Brain 2

Science, Projects, and Showcases
21

For a different thread on Izhikevich neurons I worked through an example script with adds units to the (sort-of) dimensionless way the original equations are specified.

In doing so, I added some useful features to the script that may be worth taking a look at, including a dictionary of neuron types:

# a,b,c,d define the type of neuron
# --------------------------------------     a      b       c     d
abcd_neurons = {'regular spiking':       [0.02/ms, 0.2,  -65*mV, 8*mV],
                'intrinsically bursting':[0.02/ms, 0.2,  -55*mV, 4*mV],
                'chattering':            [0.02/ms, 0.2,  -50*mV, 2*mV],
                
                'fast spiking':          [0.10/ms, 0.2,  -65*mV, 2*mV],
                'low-threshold spiking': [0.02/ms, 0.25, -65*mV, 2*mV],
                
                'thalamo-cortical':      [0.02/ms, 0.25, -65*mV, 0.05*mV],
                'resonator':             [0.10/ms, 0.25, -65*mV, 8*mV],
                }
a,b,c,d = abcd_neurons[neuron_type]

of particular interest to @serge25, I was able to get intrinsic bursting:
izhi_intrinsically bursting
as well as thalamo-cortical post-inhibitory rebound (bursting)
izhi_thalamo-cortical_burst

see the script here
@mstimberg is this worth turning into an example for the documentation?

22

Would be great to have this in the example section! I wonder whether it would make sense to have the constants a etc. instead as constant parameters in the equations (i.e. a : 1/second (constant) etc.). This way, one could do a single simulation for all classes and plot them all at once. On the other hand, this would make the code a bit more complex to read, I guess. Having to choose one parametrization in the script would be perfectly fine as well, after all, e.g. Example: Brette_Gerstner_2005 — Brian 2 2.5.4 documentation does this as well.

1 Like