Adding Flux capability

src/helper/helperFlux.jl

+
+
+import Flux
+import Statistics
+
+using Plots;
+pyplot();
+using LaTeXStrings
+
+
+function FluxTrain(nepocs::Int32, X::Array{Float64,1}, Y::Array{Float64,1})
+
+    W = [500.0]
+    b = [10000.0]
+
+    model = Flux.Dense(W, b)
+    parameters = Flux.params(model)
+    println("Initial parameter: ", parameters)
+
+    # set up data
+    Xd = reduce(hcat, X)
+    Yd = reduce(hcat, Y)
+    data = [(Xd, Yd)]
+    println("Data", data)
+
+    # optimizer = Flux.Descent(1)
+    # optimizer = Flux.ADAM()
+    optimizer = Flux.ADAM(1, (0.99, 0.999))
+
+    loss(x, y) = Statistics.mean((model(x) .- y) .^ 2)
+
+    Yd_0 = model(Xd)
+    println("Initial solution: ", Yd_0)
+
+    plot(X, Y, st = :scatter, label = L"y")
+    Yd_nE = model(Xd)
+    for iter = 1:nepocs
+        Flux.train!(loss, parameters, data, optimizer)
+    end
+
+    println("Module: ", model)
+    println("Parameters_nE: ", parameters)
+    println("Final solution: ", Yd_nE)
+    Yd_nE = model(Xd)
+
+
+    plot!(Xd', Yd_0', lc = :green, label = L"y_0")
+    plot!(Xd', Yd_nE', lc = :red, label = L"y_{n_\mathrm{E}}")
+    plot!(xlabel = L"x", ylabel = L"y", title = "Data for linear model")
+    gui()
+
+end
+
+
+function test()
+
+    x::Array{Float64,1} = [256, 4096, 512, 512, 8192]
+    y::Array{Float64,1} = [435867, 2959963, 1475489, 1485569, 4518030]
+
+    nepocs::Int32 = 100000
+
+    FluxTrain(nepocs, x, y)
+
+end
+
+
+test()