Poisson Algorithm

poisson(signal: ndarray, interval_length: int, seed: int = 0) → ndarray[source]

Perform Poisson encoding on the input signal.

This function generates a spike train using a Poisson distribution, where the probability of emitting a spike in a given interval is determined by the normalized rate of the signal.

See the Poisson Encoding description for a detailed explanation of the Poisson encoding algorithm.

Note

  • If interval_length == 1, each input value is treated directly as the firing rate (after normalization/clipping), resulting in a standard bin-by-bin approximation of Poisson encoding (at most one spike per time step).

  • If interval_length > 1, the average value over each interval block determines the constant firing rate for that block. Spikes are placed within the block using exponentially distributed inter-spike intervals, allowing multiple spikes per block (exact Poisson generation for the block’s constant rate).

Code Example:

import numpy as np
from spikify.encoders.rate import poisson
signal = np.array([0.2, 0.5, 0.8, 1.0])
np.random.seed(0)
interval_length = 2
encoded_signal = poisson(signal, interval_length)
Parameters:

  • signal (numpy.ndarray) – Input signal to encode (1D or 2D: timestamps × features).

  • interval_length (int) – Length of each time block for rate computation and spike placement. Must evenly divide the signal length. interval_length=1 uses the instantaneous value as rate (bin-by-bin); larger values use the block mean as constant rate (allows multiple spikes per block).

  • seed (int) – Random seed for reproducibility. Default is 0.

Returns:

A numpy array representing the encoded spike train.

Return type:

numpy.ndarray

Raises:

  • ValueError – If the input signal is empty or if signal length is not divisible for the interval length

  • TypeError – If the signal is not a numpy.ndarray