Error when using NMDA synapses with PoissonGroup

[Eduarda Susin]

Hello,

I successfully implement a network in brian2 containing AMPA, GABA and NMDA Synapses. I copy some of the code bellow:

#####################

#Models

#####################

eqs = """

dv/dt = (gL*(EL - v) + gL*DeltaT*exp((v - VT)/DeltaT) + ge*(Ee-v)+ gi*(Ei-v)+ G_nmda*(Enmda-v)*B*R - w + I)/C : volt (unless refractory)

IsynE=ge*(Ee-v) : amp

IsynI=gi*(Ei-v) : amp

dge/dt = -ge/tau_e : siemens

dgi/dt = -gi/tau_i : siemens

dw/dt = (a*(v - EL) - w)/tauw : amp

B=1./(1+exp(-0.062*v/mV)*(Mg/3.57)) : 1

R : 1

taum= C/gL : second

I : amp

a : siemens

b : amp

DeltaT: volt

Vcut: volt

C: farad

gL: siemens

"""

NMDA_equation='''

dx_nmda/dt = -x_nmda/tau_nmda_r : 1 (clock-driven)

ds_nmda/dt = -s_nmda/tau_nmda_d + alpha*x_nmda*(1-s_nmda) : 1 (clock-driven)

R_post= s_nmda : 1 (summed)

 

'''

#############

#Neuron groups

#############

#FS

neuronsI = NeuronGroup(NI, eqs, threshold='v>Vcut',reset="v=V_reset; w+=b", refractory=t_ref)

#RS

neuronsE = NeuronGroup(NE, eqs, threshold='v>Vcut',reset="v=V_reset; w+=b", refractory=t_ref)

####################

#Synapses (only E-E shown)

####################

con_ee = Synapses(neuronsE, neuronsE, model=NMDA_equation, on_pre='ge_post += Ge; x_nmda += 1', delay=Delay)

con_ee.connect(p=prob_p)

In this code, my external noise is given by PoissonGroup which excites the network by means of an increase of the excitatory conductances in the post-synaptic neurons. Like in the line bellow:

ExternalStimulus= PoissonGroup(N_ext, rates=5*Hz, dt=defaultclock.dt)

con_ExtE=Synapses(ExternalStimulus, neuronsE, on_pre='ge_post += Ge')

con_ExtE.connect(p=prob_p)

I also wanted to include NMDA synapses in the connection with the PoissonGroup, like bellow 

con_ExtStN_E=Synapses(ExternalStimulus, neuronsE, model=NMDA_equation, on_pre='ge_post += Ge; x_nmda += 1')

con_ExtStN_E.connect(i=Pre_exte,j=Pos_exte)

 but I get following error:

NotImplementedError: Multiple "summed variables" target the variable "R" in group "neurongroup_1". Use multiple variables in the target group instead.

Is there a solution for my problem? What am I doing wrong?

Thank you very much for your help,

Eduarda

[Marcel Stimberg]

Hi Eduarda,

we have moved all questions/discussions around Brian to the new support
forum at https://brian.discourse.group , please use it instead of this
mailing list. The forum is still quite fresh and not many questions have
been asked/answered, but you are lucky because this very question
already came up :)
https://brian.discourse.group/t/multiple-summed-variables/117

Best,

  Marcel

[Eduarda Susin]

Hello Marcel,

Thank you for your answer. I didn't know about the forum. I will use it next time!

But unfortunately, I couldn't understand the solution in the link you gave me.

That's true that the error message is the same, but I still don't know how to solve it in my case.

The error only appears if I try to connect the PoissonGroup "ExternalStimulus"

using on_pre='ge_post += Ge; x_nmda += 1' . If I connect two simple neuron groups (like neuronsE with neuronsE) and

use on_pre='ge_post += Ge; x_nmda += 1' no error comes out. So I think the issue is with the PoissonGroup "ExternalStimulus".

How do I solve this problem in my particular case?

Should I post this question on the forum?

Thank you for your help,

Eduarda

[Marcel Stimberg]

Hi Eduarda,

The error only appears if I try to connect the PoissonGroup "ExternalStimulus"

using on_pre='ge_post += Ge; x_nmda += 1' . If I connect two simple neuron groups (like neuronsE with neuronsE) and

use on_pre='ge_post += Ge; x_nmda += 1' no error comes out. So I think the issue is with the PoissonGroup "ExternalStimulus".

How do I solve this problem in my particular case?

The issue is not about the on_pre statements or the PoissonGroup, but about the "R_post = s_nmda : 1 (summed)" in the synaptic equations. You get the error because you have two Synapses objects targetting the same variable R in a target group. Instead you need two variables in the target group (e.g. R_1 and R_2), and target them separately ("R_1_post = ..." and "R_2_post = ...")

Should I post this question on the forum?

If you are still having issues, yes please!

Best,

  Marcel