LIF and Izh neurons implemented

Manifest.toml

@@ -0,0 +1,45 @@
+# This file is machine-generated - editing it directly is not advised
+
+[[Base64]]
+uuid = "2a0f44e3-6c83-55bd-87e4-b1978d98bd5f"
+
+[[Distributed]]
+deps = ["Random", "Serialization", "Sockets"]
+uuid = "8ba89e20-285c-5b6f-9357-94700520ee1b"
+
+[[InteractiveUtils]]
+deps = ["Markdown"]
+uuid = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
+
+[[Logging]]
+uuid = "56ddb016-857b-54e1-b83d-db4d58db5568"
+
+[[Markdown]]
+deps = ["Base64"]
+uuid = "d6f4376e-aef5-505a-96c1-9c027394607a"
+
+[[OrderedCollections]]
+deps = ["Random", "Serialization", "Test"]
+git-tree-sha1 = "c4c13474d23c60d20a67b217f1d7f22a40edf8f1"
+uuid = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
+version = "1.1.0"
+
+[[Parameters]]
+deps = ["OrderedCollections"]
+git-tree-sha1 = "b62b2558efb1eef1fa44e4be5ff58a515c287e38"
+uuid = "d96e819e-fc66-5662-9728-84c9c7592b0a"
+version = "0.12.0"
+
+[[Random]]
+deps = ["Serialization"]
+uuid = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
+
+[[Serialization]]
+uuid = "9e88b42a-f829-5b0c-bbe9-9e923198166b"
+
+[[Sockets]]
+uuid = "6462fe0b-24de-5631-8697-dd941f90decc"
+
+[[Test]]
+deps = ["Distributed", "InteractiveUtils", "Logging", "Random"]
+uuid = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

Project.toml

@@ -2,3 +2,6 @@ name = "nnsim"
 uuid = "2a7ddd44-2043-11ea-1b23-d1c817c05b75"
 authors = ["Sam Buercklin <[email protected]>"]
 version = "0.1.0"
+
+[deps]
+Parameters = "d96e819e-fc66-5662-9728-84c9c7592b0a"

README.md

@@ -1,6 +1,2 @@
 # NNsim
-<<<<<<< HEAD
-
 `NNsim` is a Julia package meant for fixed-time-step simulations of neural networks, both artificial and spiking neural networks (ANNs and SNNs respectively). 
-=======
->>>>>>> 3736522c4565f06620c6c4d8c2280346f7a3a0e5

src/neurons.jl

@@ -0,0 +1,72 @@
+# LIF Neuron
+@with_kw mutable struct LIF{F}<:AbstractNeuron
+    τ::F = 5.         # Time Constant (ms)
+    R::F = 10.E3      # "Resistor" (kOhms)
+    θ::F = 30.      # Threshold voltage (mV)
+    v0::F = -55.     # Reset voltage (mV)
+    I::F = 40.      # Background current injection (mA)
+    v::F = -55.     # Membrane potential (mV)
+end
+
+function update!(neuron::LIF, input_update, dt, t) 
+    retval = 0
+    # If an impulse came in, add it
+    neuron.v += input_update
+
+    # Euler method updates to potential in 2 steps for numerical stability
+    println("pre:", neuron.v)
+    neuron.v += (dt/2 * 1/neuron.τ) * (-neuron.v + neuron.R*neuron.I)
+    neuron.v += (dt/2 * 1/neuron.τ) * (-neuron.v + neuron.R*neuron.I)
+    println("post:", neuron.v)
+
+    # Check for thresholding
+    if neuron.v >= neuron.θ
+        neuron.v = neuron.v0
+        retval = 1 # Binary output
+    end
+
+    return retval
+end
+
+function reset!(neuron::LIF)
+    neuron.v = neuron.v0
+end
+
+# QIF Neuron
+# @with_kw struct QIF<:AbstractNeuron
+#     c
+# end
+
+# Izhikevich Neuron
+@with_kw struct Izh{F}<:AbstractNeuron
+   a::F = 0.02  
+   b::F = 0.2
+   c::F = -65.
+   d::F = 8.
+   I::F = 25.       # Background current injection (mA)
+   v0::F = -65.     # Reset voltage (mV)
+   θ::F = 30.       # Threshold potential (mV)
+
+   v::F = -65.      # Membrane potential (mV)
+   u::F = 0.        # Recovery variable
+end
+
+function update!(neuron::Izh, input_update, dt, t)
+    retval = 0
+
+    # Euler method updates to potential in 2 steps for numerical stability
+    neuron.v += (dt/2)*(0.05 * neuronv.v^2 + 5*neuron.v + 140 - neuron.u + neuron.I)
+    neuron.u += (dt/2)*(neuron.a)*(neuron.b*neuron.v-neuron.u)
+
+    neuron.v += (dt/2)*(0.05 * neuronv.v^2 + 5*neuron.v + 140 - neuron.u + neuron.I)
+    neuron.u += (dt/2)*(neuron.a)*(neuron.b*neuron.v-neuron.u)
+
+    # Check for thresholding
+    if neuron.v >= neuron.θ
+        neuron.v = neuron.v0
+        neuron.u += d
+        retval = 1
+    end
+
+    return retval
+end

src/nnsim.jl

@@ -1,5 +1,10 @@
 module nnsim
 
-greet() = print("Hello World!")
+using Parameters
+
+include("types.jl")
+include("neurons.jl")
+
+export AbstractNetwork, AbstractNeuron, Layer
 
 end # module

src/types.jl

@@ -1,6 +1,7 @@
-abstract type Neuron
+abstract type AbstractNetwork end
 
-struct Layer{T}
-    neurons::T
-    W
-end
+abstract type AbstractNeuron <: AbstractNetwork end
+
+struct Layer{T}<:AbstractNetwork
+
+end