engine.neuron.analog

Analog neurons emit real numbers, usually onto static synapses.

Analog neurons inherit all properties of their parent class Neuron, extending it by implementing a generic forward method that simply adds incoming input to their numeric state tensor voltage.

Analog neurons may perform normalization or provide otherwise transformed input to downstream layers.

class engine.neuron.analog.AnalogNeuron(**kwargs)

Generic analog neuron endowed with a trivial forward() method.

Defines instance attributes shared by all derived analog neuron classes.

Parameters:

spiked (Tensor) – A binary representation of spike events, registered as a PyTorch buffer.

forward(data: Tensor) → dict

Updates the numeric state stored in the instance attribute tensor voltage to data.

These default analog neurons integrate the total input impinging on them on every simulation step.

Parameters:

data (Tensor) – External input current to be added to this unit’s numeric state tensor voltage.

Raises:

RuntimeError – If data tensor is not on the same hardware device as voltage tensor.

Returns:

Dictionary containing the numeric state tensor voltage.

Return type:

dict

Modules

engine.neuron.analog.oscillator

Oscillators are waveform generators.