Benchmark 4 of the simulator review (CUBA-PS)

---

Run this example as a Jupyter notebook:

JupyterHub service

See our guide for more information and troubleshooting.

---

This script creates a sparsely coupled network of excitatory and inhibitory neurons which exhibits self-sustained activity after an initial stimulus. Connections within and across both populations are created at random. Both neuron populations receive Poissonian background input. The spike output of 500 neurons from each population are recorded. Neurons are modeled as leaky integrate-and-fire neurons with voltage jump synapses. Spike times are not constrained to the discrete time grid. The model is based on the Vogels & Abbott network model [1].

This is Benchmark 4 of the FACETS simulator review (Brette et al., 2007) [2]: - Neuron model: integrate-and-fire (iaf_psc_delta) - Synapse model: current-based (CUBA) - Synapse time course: delta (voltage jump) - Spike times: off-grid (precise spiking)

References

[1]

Vogels TP, Abbott LF. 2005. Signal propagation and logic gating in networks of integrate-and-fire neurons. Journal of Neuroscience. 25(46):10786-10795. https://doi.org/10.1523/JNEUROSCI.3508-05.2005

[2]

Brette R, Rudolph M, Carnevale T, Hines M, Beeman D, Bower JM, et al. 2007. Simulation of networks of spiking neurons: a review of tools and strategies. Journal of Computational Neuroscience. 23(3):349-398. https://doi.org/10.1007/s10827-007-0038-6

import nest
from brette_et_al_2007_benchmark import run_simulation

Set benchmark parameters

params = {
    "model": "iaf_psc_delta",  # neuron model (delta synapses for voltage jump)
    "model_params": {
        "E_L": -49.0,  # resting membrane potential [mV]
        # see Brette et al, J Comput Neurosci 23:349 (2007), p 393
        "V_m": -49.0,  # initial membrane potential [mV]
        "V_th": -50.0,  # threshold [mV]
        "V_reset": -60.0,  # reset potential [mV]
        "C_m": 200.0,  # capacity of the membrane [pF]
        "tau_m": 20.0,  # membrane time constant [ms]
        "t_ref": 5.0,  # duration of refractory period [ms]
    },
    "delay": 0.125,  # synaptic delay [ms]
    "E_synapse_params": {
        "weight": 0.25,  # excitatory amplitude [mV]
    },
    "I_synapse_params": {
        "weight": -2.25,  # inhibitory amplitude [pA]
    },
    "stimulus": "poisson_generator",
    "stimulus_params": {
        "rate": 300.0,  # rate of initial Poisson stimulus [spikes/s]
        "start": 1.0,  # start of Poisson generator [ms]
        "stop": 51.0,  # stop of Poisson generator [ms]
        "origin": 0.0,  # origin of time [ms]
    },
    "recorder": "spike_recorder",
    "recorder_params": {
        "record_to": "ascii",
        "label": "cuba_ps",
    },
    "Nrec": 500,  # number of neurons per population to record from
    "Nstim": 50,  # number of neurons to stimulate
    "simtime": 10000.0,  # simulated time [ms]
    "dt": 0.125,  # simulation step [ms]
    "NE": 3200,  # number of excitatory neurons
    "NI": 800,  # number of inhibitory neurons
    "epsilon": 0.02,  # connection probability
    "virtual_processes": 1,  # number of virtual processes to use
}

Run the simulation

if __name__ == "__main__":
    nest.set_verbosity("M_WARNING")
    # Note: iaf_psc_delta uses precise spiking by default (off-grid)
    results = run_simulation(params)