Excitatory/Inhibitory LIF Neuron Model (Float Precision)

[monkin77]

User story

As a user, I want to model E/I LIF neurons to represent more complex behaviors containing positive and negative inputs, each with their own decay time constants (du_exc and du_inh).

On top of that, dv, du_exc and du_inh can vary on a neuron-to-neuron basis.

Conditions of satisfaction

• Create the Processes and ProcessModels that implement the neuron E/I neuron model.
• Include a Refractory version of this model

Acceptance tests

• Validate the dynamics of the neuron model only relying on biases.
• Validate the dynamics of the model when feeding the same input to a layer of neurons containing different du_exc and du_inh per neuron.
• Validate the dynamics of the refractory version of the model.

Note

This only contains the floating-point precision version of the model. I believe it is possible to implement a fixed-precision version of it, although it requires further investigation.