Quickstart¶

Installation¶

As btorch is not yet on PyPI, install from source:

git clone https://github.com/Criticality-Cognitive-Computation-Lab/btorch.git
cd btorch
pip install -e . --config-settings editable_mode=strict

Basic Neuron Usage¶

Create and run a simple LIF neuron:

import torch
from btorch.models.neurons import LIF

# Create a single LIF neuron
neuron = LIF(n_neuron=100, tau=20.0, v_threshold=1.0)

# Run for 100 timesteps
spikes = []
for t in range(100):
    # Random input current
    x = torch.randn(100) * 0.5
    spike = neuron(x)
    spikes.append(spike)

# Stack spikes into a tensor [time, neurons]
spike_train = torch.stack(spikes)

Using Heterogeneous Parameters¶

Neurons support per-neuron parameters:

import torch
from btorch.models.neurons import LIF

# Each neuron has its own time constant
taus = torch.rand(100) * 30 + 10  # 10-40 ms range

neuron = LIF(n_neuron=100, tau=taus, v_threshold=1.0)

Basic Analysis Usage¶

Analyze spike trains:

import numpy as np
from btorch.analysis import isi_cv, fano, firing_rate

# Generate sample spike data [time, batch, neurons]
spike_data = np.random.rand(1000, 10, 50) > 0.95

# Coefficient of variation of ISIs
cv, isi_total, isi_stats = isi_cv(spike_data, dt_ms=1.0)

# Fano factor (variance/mean of spike counts)
fano_values, fano_stats = fano(spike_data, window_ms=100, dt_ms=1.0)

# Firing rate via convolution
rates = firing_rate(spike_data, dt_ms=1.0, smooth_ms=50)

Shape Conventions¶

Input shape: (*batch, n_neuron) where *batch can be any number of dimensions
n_neuron is stored as a tuple; use .size for total neuron count
Use init_net_state(..., batch_size=(...)) for multi-dimensional batch setups