Models, types, and functions for Julia SpikingNeuralNetworks.jl

SNNModels

The package is the Base package of SpikingNeuralNetworks.jl. It contains model types and core functionalities for the SpikingNeuralNetworks.jl ecosystem.

Documentation

The package defines models and parameters for Population, Connection, and Stimulus:

• Population <: AbstractPopulation
• Connection <: AbstractConnection
• Stimulus <: AbstractStimulus
• PopulationParameter <: AbstractPopulationParameter
• ConnectionParameter <: AbstractConnectionParameter
• StimulusParameter <: AbstractStimulusParameter
• SpikeTimes = Vector{Vector{Float32}}.

Populations, connections, and stimuli are defined under the respective folders in src

Under src/utils, the package defines macros and functions that support the functionalities of the SpikingNeuralNetwork.jl ecosystem:

• struct.jl defines the abstract model types.
• main.jl defines the sim! and train! functions that run the network simulations.
• io.jl defines functions to save and load models using .jld2 format.
• record.jl implements the recording ofthe model's variables during simulation time.
• macros.jl implements useful macros to define model types and update parameter structs.
• spatial.jl defines functions to create spatial network arrangements.
• unit.jl defines convenient shortcut for cgm unit system.
• util.jl add functions to manipulate sparse matrix representations.

Functioning

The library leverages Julia multidispatching to run models of types <: AbstractPopulation, <: AbstractConnection, and AbstractStimulus.

function sim!(p::Vector{AbstractPopulation}, c::Vector{AbstractConnection}, duration<:Real) end
function train!(p::Vector{AbstractConnection}, c:Vector{AbstractConnection}, duration<:Real) end

The functions support simulation with and without neural plasticity; the model is defined within the arguments passed to the functions. Models are composed of 'AbstractPopulation' and 'AbstractConnection' arrays.

Any elements of AbstractPopulation must implement the methods:

function integrate!(p, p.param, dt) end
function plasticity!(p, p.param, dt, T) end

AbstractConnection must implement the methods:

function forward!(p, p.param) end
function plasticity!(c, c.param, dt) end

AbstractStimulus must implement the methods:

function stimulate!(p, p.param) end