Merge pull request #3541 from vyudu/drop_functionexpr

fix: don't call eval'd XFunctions in the tests

Author: ChrisRackauckas

test/discrete_system.jl

@@ -34,21 +34,16 @@ eqs = [S ~ S(k - 1) - infection * h,
 syss = structural_simplify(sys)
 @test syss == syss
 
-for df in [
-    DiscreteFunction(syss),
-    eval(DiscreteFunctionExpr(syss))
-]
-
-    # iip
-    du = zeros(3)
-    u = collect(1:3)
-    p = MTKParameters(syss, [c, nsteps, δt, β, γ] .=> collect(1:5))
-    df.f(du, u, p, 0)
-    @test du ≈ [0.01831563888873422, 0.9816849729159067, 4.999999388195359]
-
-    # oop
-    @test df.f(u, p, 0) ≈ [0.01831563888873422, 0.9816849729159067, 4.999999388195359]
-end
+df = DiscreteFunction(syss)
+# iip
+du = zeros(3)
+u = collect(1:3)
+p = MTKParameters(syss, [c, nsteps, δt, β, γ] .=> collect(1:5))
+df.f(du, u, p, 0)
+@test du ≈ [0.01831563888873422, 0.9816849729159067, 4.999999388195359]
+
+# oop
+@test df.f(u, p, 0) ≈ [0.01831563888873422, 0.9816849729159067, 4.999999388195359]
 
 # Problem
 u0 = [S(k - 1) => 990.0, I(k - 1) => 10.0, R(k - 1) => 0.0]

test/nonlinearsystem.jl

@@ -30,7 +30,7 @@ eqs = [0 ~ σ * (y - x) * h,
 @test eval(toexpr(ns)) == ns
 test_nlsys_inference("standard", ns, (x, y, z), (σ, ρ, β))
 @test begin
-    f = eval(generate_function(ns, [x, y, z], [σ, ρ, β])[2])
+    f = generate_function(ns, [x, y, z], [σ, ρ, β], expression = Val{false})[2]
     du = [0.0, 0.0, 0.0]
     f(du, [1, 2, 3], [1, 2, 3])
     du ≈ [1, -3, -7]

test/odesystem.jl

@@ -54,42 +54,38 @@ jac = calculate_jacobian(de)
 jacfun = eval(jac_expr[2])
 
 de = complete(de)
-for f in [
-    ODEFunction(de, [x, y, z], [σ, ρ, β], tgrad = true, jac = true),
-    eval(ODEFunctionExpr(de, [x, y, z], [σ, ρ, β], tgrad = true, jac = true))
-]
-    # system
-    @test f.sys === de
-
-    # iip
-    du = zeros(3)
-    u = collect(1:3)
-    p = ModelingToolkit.MTKParameters(de, [σ, ρ, β] .=> 4.0:6.0)
-    f.f(du, u, p, 0.1)
-    @test du == [4, 0, -16]
-
-    # oop
-    du = @SArray zeros(3)
-    u = SVector(1:3...)
-    p = ModelingToolkit.MTKParameters(de, SVector{3}([σ, ρ, β] .=> 4.0:6.0))
-    @test f.f(u, p, 0.1) === @SArray [4.0, 0.0, -16.0]
-
-    # iip vs oop
-    du = zeros(3)
-    g = similar(du)
-    J = zeros(3, 3)
-    u = collect(1:3)
-    p = ModelingToolkit.MTKParameters(de, [σ, ρ, β] .=> 4.0:6.0)
-    f.f(du, u, p, 0.1)
-    @test du == f(u, p, 0.1)
-    f.tgrad(g, u, p, t)
-    @test g == f.tgrad(u, p, t)
-    f.jac(J, u, p, t)
-    @test J == f.jac(u, p, t)
-end
+f = ODEFunction(de, [x, y, z], [σ, ρ, β], tgrad = true, jac = true)
+# system
+@test f.sys === de
+
+# iip
+du = zeros(3)
+u = collect(1:3)
+p = ModelingToolkit.MTKParameters(de, [σ, ρ, β] .=> 4.0:6.0)
+f.f(du, u, p, 0.1)
+@test du == [4, 0, -16]
+
+# oop
+du = @SArray zeros(3)
+u = SVector(1:3...)
+p = ModelingToolkit.MTKParameters(de, SVector{3}([σ, ρ, β] .=> 4.0:6.0))
+@test f.f(u, p, 0.1) === @SArray [4.0, 0.0, -16.0]
+
+# iip vs oop
+du = zeros(3)
+g = similar(du)
+J = zeros(3, 3)
+u = collect(1:3)
+p = ModelingToolkit.MTKParameters(de, [σ, ρ, β] .=> 4.0:6.0)
+f.f(du, u, p, 0.1)
+@test du == f(u, p, 0.1)
+f.tgrad(g, u, p, t)
+@test g == f.tgrad(u, p, t)
+f.jac(J, u, p, t)
+@test J == f.jac(u, p, t)
 
 #check iip_config
-f = eval(ODEFunctionExpr(de, [x, y, z], [σ, ρ, β], iip_config = (false, true)))
+f = ODEFunction(de, [x, y, z], [σ, ρ, β], iip_config = (false, true))
 du = zeros(3)
 u = collect(1:3)
 p = ModelingToolkit.MTKParameters(de, [σ, ρ, β] .=> 4.0:6.0)
@@ -146,7 +142,7 @@ eqs = [D(x) ~ σ′ * (y - x),
 @named de = ODESystem(eqs, t)
 test_diffeq_inference("global iv-varying", de, t, (x, y, z), (σ′, ρ, β))
 
-f = eval(generate_function(de, [x, y, z], [σ′, ρ, β])[2])
+f = generate_function(de, [x, y, z], [σ′, ρ, β], expression = Val{false})[2]
 du = [0.0, 0.0, 0.0]
 f(du, [1.0, 2.0, 3.0], [x -> x + 7, 2, 3], 5.0)
 @test du ≈ [11, -3, -7]
@@ -157,15 +153,15 @@ eqs = [D(x) ~ σ(t - 1) * (y - x),
     D(z) ~ x * y - β * z * κ]
 @named de = ODESystem(eqs, t)
 test_diffeq_inference("single internal iv-varying", de, t, (x, y, z), (σ, ρ, β))
-f = eval(generate_function(de, [x, y, z], [σ, ρ, β])[2])
+f = generate_function(de, [x, y, z], [σ, ρ, β], expression = Val{false})[2]
 du = [0.0, 0.0, 0.0]
 f(du, [1.0, 2.0, 3.0], [x -> x + 7, 2, 3], 5.0)
 @test du ≈ [11, -3, -7]
 
 eqs = [D(x) ~ x + 10σ(t - 1) + 100σ(t - 2) + 1000σ(t^2)]
 @named de = ODESystem(eqs, t)
 test_diffeq_inference("many internal iv-varying", de, t, (x,), (σ,))
-f = eval(generate_function(de, [x], [σ])[2])
+f = generate_function(de, [x], [σ], expression = Val{false})[2]
 du = [0.0]
 f(du, [1.0], [t -> t + 2], 5.0)
 @test du ≈ [27561]
@@ -218,11 +214,11 @@ eqs = [D(x) ~ -A * x,
 @named de = ODESystem(eqs, t)
 @test begin
     local f, du
-    f = eval(generate_function(de, [x, y], [A, B, C])[2])
+    f = generate_function(de, [x, y], [A, B, C], expression = Val{false})[2]
     du = [0.0, 0.0]
     f(du, [1.0, 2.0], [1, 2, 3], 0.0)
     du ≈ [-1, -1 / 3]
-    f = eval(generate_function(de, [x, y], [A, B, C])[1])
+    f = generate_function(de, [x, y], [A, B, C], expression = Val{false})[1]
     du ≈ f([1.0, 2.0], [1, 2, 3], 0.0)
 end
 

test/steadystatesystems.jl

@@ -17,7 +17,7 @@ for factor in [1e-1, 1e0, 1e10],
     ss_prob = SteadyStateProblem(de, u0, p)
     sol = solve(ss_prob, SSRootfind()).u[1]
     @test abs(sol^2 - factor * u0_p[2]) < 1e-8
-    ss_prob = SteadyStateProblemExpr(de, u0, p)
-    sol_expr = solve(eval(ss_prob), SSRootfind()).u[1]
+    ss_prob = SteadyStateProblem(de, u0, p)
+    sol_expr = solve(ss_prob, SSRootfind()).u[1]
     @test all(x -> x == 0, sol - sol_expr)
 end