Section connection problem in multi-compartmental model

Support
1

We are working on creating a multicompartmental model on Brian2, but we are facing with problem with my output. We realise that problem may occur because the sections of the neuron are not connecting to each other. The membrane potential should decrease far away from the soma but we are seeing huge increase in other sections. I am not sure about the problem is in my code or in simulation.

#biophysics prameters 
gl = 3e-4*siemens/cm**2
El = -76.5*mV  # from code, text says: -69.85*mV
E_Na = 50*mV
E_K = -100*mV

eqs = Equations('''
Im = gl*(El-v) - I_Na - I_K: amp/meter**2
I_inj : amp (point current)

# HH-type currents for spike initiation
g_Na : siemens/meter**2
g_K: siemens/meter**2
''')
CA1_morpho = Soma(diameter = 30*um)
CA1_morpho.axon = Section (n=10, diameter =[1.5,1.3,1,1,1,1,1,1,1,1,2]*um, length=[5,40,100,2,100,2,100,2,100,10]*um, start_diameter=1*um, type='axon')
CA1_morpho.apical = Section(n=3,diameter=[4,3,2,2]*um,length=[40,40,40]*um, start_diameter=2*um, type='dend')

CA1 = SpatialNeuron(morphology=CA1_morpho, model=eqs, Cm=Cm, Ri=Ri,clock=defaultclock,method='exponential_euler')
2

To add, we realize that in source code may be causing the error, in source code there is part says:

space_constant = (2/pi)**(1.0/3.0) * (area/(1/r_length_1 + 1/r_length_2))**(1.0/6.0) /
                         (2*(Ri*gtot__private)**(1.0/2.0)) : meter

this may cause propagation in model, even we utilize Na channels to ‘0’, it still generate an action potential and overstimulate the model

Thanks for your replies and advices!

3

Hi @cags. Could you please provide some minimal running code that shows the problem? You say that “We realise that problem may occur because the sections of the neuron are not connecting to each other” but I don’t quite see how that would explain a “huge increase in other sections”.

I also don’t quite understand how the definition of the space constant would “cause propagation” in the model. This constant (as well as the time_constant) is only given for informational purposes and shows the spatial scale for signal decay based on the given morphology and resistance, it does not influence the simulation itself. Or do you mean that the values for this constant are wrong/infinite/NaN or something like that?