Merge pull request #408 from SciML/forward_tests

Forward tests

.github/workflows/CI.yml

@@ -20,6 +20,7 @@ jobs:
           - IntegroDiff
           - NNSTOPPINGTIME
           - NNRODE
+          - Forward
     steps:
       - uses: actions/checkout@v2
       - uses: julia-actions/setup-julia@v1

test/forward_tests.jl

@@ -0,0 +1,81 @@
+using Flux
+println("forward_tests")
+using DiffEqFlux
+println("Starting Soon!")
+using ModelingToolkit
+using DiffEqBase
+using Test, NeuralPDE
+println("Starting Soon!")
+using SciMLBase
+import ModelingToolkit: Interval
+
+@testset "ODE" begin
+    @parameters x
+    @variables u(..)
+
+    Dx = Differential(x)
+    eq = Dx(u(x)) ~ 0.
+    bcs = [u(0.) ~ u(0.)]
+    domains = [x ∈ Interval(0.0,1.0)]
+    chain = FastChain((x,p) -> x.^2)
+
+    chain([1],Float64[])
+    strategy_ = NeuralPDE.GridTraining(0.1)
+    discretization = NeuralPDE.PhysicsInformedNN(chain,strategy_;init_params = Float64[])
+    @named pde_system = PDESystem(eq,bcs,domains,[x],[u(x)])
+    prob = NeuralPDE.discretize(pde_system,discretization)
+
+    train_data =prob.f.f.loss_function.pde_loss_function.pde_loss_functions.contents[1].train_set
+    inner_loss =prob.f.f.loss_function.pde_loss_function.pde_loss_functions.contents[1].loss_function
+
+    dudx(x) = @. 2*x
+    @test inner_loss(train_data, Float64[]) ≈ dudx(train_data) rtol = 1e-8
+end
+
+@testset "derivatives" begin
+    chain = FastChain(FastDense(2,16,Flux.σ),FastDense(16,16,Flux.σ),FastDense(16,1))
+    initθ = Float64.(DiffEqFlux.initial_params(chain))
+
+    eltypeθ = eltype(initθ)
+    parameterless_type_θ = DiffEqBase.parameterless_type(initθ)
+    phi = NeuralPDE.get_phi(chain,parameterless_type_θ)
+    derivative = NeuralPDE.get_numeric_derivative()
+
+    u_ = (cord, θ, phi)->sum(phi(cord, θ))
+
+    phi([1,2], initθ)
+
+    phi_ = (p) -> phi(p, initθ)[1]
+    dphi = Zygote.gradient(phi_,[1.,2.])
+
+    function get_ε(dim, der_num,eltypeθ)
+        epsilon = cbrt(eps(eltypeθ))
+        ε = zeros(eltypeθ, dim)
+        ε[der_num] = epsilon
+        ε
+    end
+
+    eps_x = get_ε(2, 1,Float64)
+    eps_y = get_ε(2, 2,Float64)
+
+    dphi_x = derivative(phi,u_,[1.,2.],[eps_x],1,initθ)
+    dphi_y = derivative(phi,u_,[1.,2.],[eps_y],1,initθ)
+
+    #first order derivatives
+    @test isapprox(dphi[1][1], dphi_x, atol=1e-8)
+    @test isapprox(dphi[1][2], dphi_y, atol=1e-8)
+
+    dphi_x = derivative(phi,u_,[1.,2.],[[ 0.0049215667, 0.0]],1,initθ)
+    dphi_y = derivative(phi,u_,[1.,2.],[[0.0,  0.0049215667]],1,initθ)
+
+    hess_phi = Zygote.hessian(phi_,[1,2])
+
+    dphi_xx = derivative(phi,u_,[1.,2.],[eps_x,eps_x],2,initθ)
+    dphi_xy = derivative(phi,u_,[1.,2.],[eps_x,eps_y],2,initθ)
+    dphi_yy = derivative(phi,u_,[1.,2.],[eps_y,eps_y],2,initθ)
+
+    #second order derivatives
+    @test isapprox(hess_phi[1], dphi_xx, atol=1e-5)
+    @test isapprox(hess_phi[2], dphi_xy, atol=1e-5)
+    @test isapprox(hess_phi[4], dphi_yy, atol=1e-5)
+end

test/runtests.jl

@@ -38,6 +38,9 @@ const is_CI = haskey(ENV,"CI")
         @time @safetestset "NNRODE" begin include("NNRODE_tests.jl") end
         @time @safetestset "NNParamKolmogorov" begin include("NNParamKolmogorov_tests.jl") end
   end
+  if GROUP == "All" || GROUP == "Forward"
+        @time @safetestset "Forward" begin include("forward_tests.jl") end
+  end
 
   if !is_APPVEYOR && GROUP == "GPU"
      @safetestset "NNPDE_gpu" begin include("NNPDE_tests_gpu.jl") end