Move interface tests to OrdinaryDiffEqDifferentiation and OrdinaryDiffEqNonlinearSolve

Move tests from the global test suite to their respective subpackages so they
only run when the relevant subpackage changes, reducing CI cost.

Tests moved to OrdinaryDiffEqDifferentiation:
- differentiation_traits_tests.jl (from InterfaceI)
- autodiff_error_tests.jl (from InterfaceIV)
- nojac_tests.jl (from InterfaceIII)

Tests moved to OrdinaryDiffEqNonlinearSolve:
- linear_nonlinear_tests.jl (from InterfaceII)
- linear_solver_tests.jl (from InterfaceII)
- linear_solver_split_ode_tests.jl (from InterfaceII)
- mass_matrix_tests.jl (from InterfaceII)
- wprototype_tests.jl (from InterfaceII)
- dae_initialization_tests.jl (from InterfaceV)
- checkinit_tests.jl (from InterfaceII)

Tests kept in global suite (require special envs):
- autosparse_detection_tests.jl (SparseConnectivityTracer dep conflict)
- gpu_autodiff_interface_tests.jl (GPU env)
- nonfulldiagonal_sparse.jl (ComponentArrays dep)
- AD tests (Enzyme/Zygote/Mooncake env)
- ModelingToolkit tests (downstream dep)

Co-Authored-By: Chris Rackauckas <accounts@chrisrackauckas.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: ChrisRackauckas

lib/OrdinaryDiffEqDifferentiation/Project.toml

@@ -4,7 +4,7 @@ uuid = "4302a76b-040a-498a-8c04-15b101fed76b"
 authors = ["Chris Rackauckas <accounts@chrisrackauckas.com>", "Yingbo Ma <mayingbo5@gmail.com>"]
 
 [targets]
-test = ["DiffEqDevTools", "OrdinaryDiffEqSDIRK", "Random", "SafeTestsets", "Test", "SparseArrays", "Pkg"]
+test = ["DiffEqDevTools", "OrdinaryDiffEqBDF", "OrdinaryDiffEqRosenbrock", "OrdinaryDiffEqSDIRK", "Random", "RecursiveFactorization", "SafeTestsets", "Test", "SparseArrays", "Pkg"]
 
 [compat]
 Pkg = "1"
@@ -22,7 +22,10 @@ ConstructionBase = "1.5.8"
 LinearAlgebra = "1.10"
 SciMLBase = "2.141"
 OrdinaryDiffEqCore = "3"
+OrdinaryDiffEqBDF = "1"
+OrdinaryDiffEqRosenbrock = "1"
 OrdinaryDiffEqSDIRK = "1"
+RecursiveFactorization = "0.2, 1"
 SparseArrays = "1.10"
 ConcreteStructs = "0.2"
 Aqua = "0.8.11"
@@ -69,12 +72,21 @@ Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 DiffEqDevTools = "f3b72e0c-5b89-59e1-b016-84e28bfd966d"
 Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
 AllocCheck = "9b6a8646-10ed-4001-bbdc-1d2f46dfbb1a"
+OrdinaryDiffEqBDF = "6ad6398a-0878-4a85-9266-38940aa047c8"
+OrdinaryDiffEqRosenbrock = "43230ef6-c299-4910-a778-202eb28ce4ce"
 OrdinaryDiffEqSDIRK = "2d112036-d095-4a1e-ab9a-08536f3ecdbf"
+RecursiveFactorization = "f2c3362d-daeb-58d1-803e-2bc74f2840b4"
 SafeTestsets = "1bc83da4-3b8d-516f-aca4-4fe02f6d838f"
 
 [sources.OrdinaryDiffEqCore]
 path = "../OrdinaryDiffEqCore"
 
+[sources.OrdinaryDiffEqBDF]
+path = "../OrdinaryDiffEqBDF"
+
+[sources.OrdinaryDiffEqRosenbrock]
+path = "../OrdinaryDiffEqRosenbrock"
+
 [sources.OrdinaryDiffEqSDIRK]
 path = "../OrdinaryDiffEqSDIRK"
 

lib/OrdinaryDiffEqDifferentiation/test/autodiff_error_tests.jl

@@ -0,0 +1,82 @@
+using OrdinaryDiffEqRosenbrock, OrdinaryDiffEqSDIRK, OrdinaryDiffEqBDF, Test, ADTypes
+using OrdinaryDiffEqDifferentiation
+
+const a = Float64[1.0]
+
+function lorenz(u, p, t)
+    du1 = 10.0(u[2] - u[1])
+    a[1] = u[2]
+    du2 = u[1] * (28.0 - u[3]) - u[2]
+    du3 = u[1] * u[2] - (8 / 3) * u[3]
+    return [du1, du2, du3]
+end
+u0 = [1.0; 0.0; 0.0]
+tspan = (0.0, 1.0)
+prob = ODEProblem(lorenz, u0, tspan)
+@test_throws OrdinaryDiffEqDifferentiation.FirstAutodiffJacError solve(prob, Rosenbrock23())
+
+function lorenz(u, p, t)
+    du1 = 10.0(u[2] - u[1])
+    a[1] = t
+    du2 = u[1] * (28.0 - u[3]) - u[2]
+    du3 = u[1] * u[2] - (8 / 3) * u[3]
+    return [du1, du2, du3]
+end
+@test_throws OrdinaryDiffEqDifferentiation.FirstAutodiffTgradError solve(
+    prob, Rosenbrock23()
+)
+
+function lorenz!(du, u, p, t)
+    du[1] = 10.0(u[2] - u[1])
+    a[1] = u[2]
+    du[2] = u[1] * (28.0 - u[3]) - u[2]
+    return du[3] = u[1] * u[2] - (8 / 3) * u[3]
+end
+u0 = [1.0; 0.0; 0.0]
+tspan = (0.0, 1.0)
+prob = ODEProblem(lorenz!, u0, tspan)
+@test_throws OrdinaryDiffEqDifferentiation.FirstAutodiffJacError solve(prob, Rosenbrock23())
+
+function lorenz2!(du, u, p, t)
+    du[1] = 10.0(u[2] - u[1])
+    a[1] = t
+    du[2] = u[1] * (28.0 - u[3]) - u[2]
+    return du[3] = u[1] * u[2] - (8 / 3) * u[3]
+end
+prob = ODEProblem(lorenz2!, u0, tspan)
+@test_throws OrdinaryDiffEqDifferentiation.FirstAutodiffTgradError solve(
+    prob, Rosenbrock23()
+)
+
+## Test that nothing is using duals when autodiff=AutoFiniteDiff()
+## https://discourse.julialang.org/t/rodas4-using-dual-number-for-time-with-autodiff-false/98256
+
+for alg in [
+        Rosenbrock23(autodiff = AutoFiniteDiff()),
+        Rodas4(autodiff = AutoFiniteDiff()),
+        Rodas5(autodiff = AutoFiniteDiff()),
+        QNDF(autodiff = AutoFiniteDiff()),
+        TRBDF2(autodiff = AutoFiniteDiff()),
+        KenCarp4(autodiff = AutoFiniteDiff()),
+    ]
+    u = [0.0, 0.0]
+    function f1(u, p, t)
+        #display(typeof(t))
+        du = zeros(2)
+        du[1] = 0.1 * u[1] + 0.2 * u[2]
+        du[2] = 0.1 * t
+        return du
+    end
+    prob = ODEProblem(f1, u, (0.0, 1.0))
+    sol = solve(prob, alg)
+
+    function f2(du, u, p, t)
+        #display(typeof(t))
+        du2 = zeros(2)
+        du2[1] = 0.1 * u[1] + 0.2 * u[2]
+        du2[2] = 0.1 * t
+        return du .= du2
+    end
+    prob = ODEProblem(f2, u, (0.0, 1.0))
+    sol = solve(prob, alg)
+end

lib/OrdinaryDiffEqDifferentiation/test/differentiation_traits_tests.jl

@@ -0,0 +1,45 @@
+using OrdinaryDiffEqRosenbrock, Test, ADTypes
+
+jac_called = Ref(false)
+tgrad_called = Ref(false)
+
+function Lotka(du, u, p, t)
+    du[1] = u[1] - u[1] * u[2] # REPL[7], line 3:
+    du[2] = -3 * u[2] + 1 * u[1] * u[2]
+    return nothing
+end
+
+function Lotka_jac(J, u, p, t)
+    jac_called.x = true
+    J[1, 1] = 1.0 - u[2]
+    J[1, 2] = -u[1]
+    J[2, 1] = 1 * u[2]
+    J[2, 2] = -3 + u[1]
+    return nothing
+end
+
+function Lotka_tgrad(grad, u, p, t)
+    tgrad_called.x = true
+    grad[1] = 1 * 0
+    grad[2] = 1 * 0
+    return nothing
+end
+
+Lotka_f = ODEFunction(Lotka, jac = Lotka_jac, tgrad = Lotka_tgrad)
+prob = ODEProblem(Lotka_f, ones(2), (0.0, 10.0))
+
+good_sol = solve(prob, Rosenbrock23())
+
+@test jac_called[]
+@test tgrad_called[]
+
+prob2 = ODEProblem(Lotka, ones(2), (0.0, 10.0))
+
+sol = solve(prob2, Rosenbrock23(autodiff = AutoForwardDiff()))
+@test ≈(good_sol[:, end], sol[:, end], rtol = 1.0e-2)
+
+sol = solve(prob2, Rosenbrock23(autodiff = AutoForwardDiff(chunksize = 1)))
+@test ≈(good_sol[:, end], sol[:, end], rtol = 1.0e-2)
+
+sol = solve(prob2, Rosenbrock23(autodiff = AutoFiniteDiff()))
+@test ≈(good_sol[:, end], sol[:, end], rtol = 1.0e-2)