BrainPy-Models

Note: We welcome your contributions for model implementations.

BrainPy-Models is a repository accompany with BrainPy, which is a framework for spiking neural network simulation. With BrainPy, we implements the most canonical and effective neuron models and synapse models, and show them in BrainPy-Models.

Here, users can directly import our models into your network, and also can learn examples of how to use BrainPy from Documentations.

We provide the following models:

Installation

Install from source code:

python setup.py install

Install BrainPy-Models using conda:

conda install bpmodels -c brainpy 

Install BrainPy-Models using pip:

pip install bpmodels

The following packages need to be installed to use BrainPy-Models:

• Python >= 3.7
• Matplotlib >= 2.0
• brainpy-simulator >= 0.3.0

Quick Start

The use of bpmodels is very convenient, let's take an example of the implementation of the E-I balanced network.

We start by importing the brainpy and bpmodels packages and set profile.

import brainpy as bp
import bpmodels
import numpy as np
import matplotlib.pyplot as plt

# set profile
bp.profile.set(jit=True, device='cpu',
               numerical_method='exponential')

The E-I balanced network is based on leaky Integrate-and-Fire (LIF) neurons connecting with single exponential decay synapses. As showed in the table above, bpmodels provides pre-defined LIF neuron model and exponential synapse model, so we can use bpmodels.neurons.get_LIF and bpmodels.synapses.get_exponential to get the pre-defined models.

V_rest = -52.
V_reset = -60.
V_th = -50.

neu = bpmodels.neurons.get_LIF(V_rest=V_rest, V_reset = V_reset, V_th=V_th, noise=0., mode='scalar')

syn = bpmodels.synapses.get_exponential(tau_decay = 2., mode='scalar')

# build network
num_exc = 500
num_inh = 500
prob = 0.1

JE = 1 / np.sqrt(prob * num_exc)
JI = 1 / np.sqrt(prob * num_inh)

group = bp.NeuGroup(neu, geometry=num_exc + num_inh, monitors=['spike'])

group.ST['V'] = np.random.random(num_exc + num_inh) * (V_th - V_rest) + V_rest

exc_conn = bp.SynConn(syn,
                      pre_group=group[:num_exc],
                      post_group=group,
                      conn=bp.connect.FixedProb(prob=prob))
exc_conn.ST['w'] = JE

inh_conn = bp.SynConn(syn,
                      pre_group=group[num_exc:],
                      post_group=group,
                      conn=bp.connect.FixedProb(prob=prob))
exc_conn.ST['w'] = -JI

net = bp.Network(group, exc_conn, inh_conn)
net.run(duration=500., inputs=(group, 'ST.input', 3.))

# visualization
fig, gs = bp.visualize.get_figure(4, 1, 2, 10)

fig.add_subplot(gs[:3, 0])
bp.visualize.raster_plot(net.ts, group.mon.spike, xlim=(50, 450))

fig.add_subplot(gs[3, 0])
rates = bp.measure.firing_rate(group.mon.spike, 5.)
plt.plot(net.ts, rates)
plt.xlim(50, 450)
plt.show()

Then you would expect to see the following output: