Revamp CI

.github/workflows/CI.yml

@@ -10,28 +10,34 @@ on:
       - main
     paths-ignore:
       - 'docs/**'
+  schedule:
+    - cron: '41 0 * * 5'
 jobs:
   test:
-    runs-on: ubuntu-latest
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        group:
+          - Core
+        version:
+          - '1'
+        os:
+          - ubuntu-latest
+          - macos-latest
+          - windows-latest
     steps:
       - uses: actions/checkout@v4
       - uses: julia-actions/setup-julia@v1
         with:
-          version: 1
-      - uses: actions/cache@v4
-        env:
-          cache-name: cache-artifacts
+          version: ${{ matrix.version }}
+      - uses: julia-actions/cache@v1
         with:
-          path: ~/.julia/artifacts
-          key: ${{ runner.os }}-test-${{ env.cache-name }}-${{ hashFiles('**/Project.toml') }}
-          restore-keys: |
-            ${{ runner.os }}-test-${{ env.cache-name }}-
-            ${{ runner.os }}-test-
-            ${{ runner.os }}-
+          token: ${{ secrets.GITHUB_TOKEN }}
       - uses: julia-actions/julia-buildpkg@v1
       - uses: julia-actions/julia-runtest@v1
-        env:
-          GROUP: ${{ matrix.group }}
+        with:
+          depwarn: error
       - uses: julia-actions/julia-processcoverage@v1
       - uses: codecov/codecov-action@v3
         with:

.github/workflows/Downgrade.yml

@@ -0,0 +1,41 @@
+name: Downgrade
+on:
+  pull_request:
+    branches:
+      - main
+    paths-ignore:
+      - 'docs/**'
+  push:
+    branches:
+      - main
+    paths-ignore:
+      - 'docs/**'
+  schedule:
+    - cron: '41 0 * * 5'
+jobs:
+  test:
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        group:
+          - Core
+        version:
+          - '1'
+        os:
+          - ubuntu-latest
+          - macos-latest
+          - windows-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: julia-actions/setup-julia@v1
+        with:
+          version: ${{ matrix.version }}
+      - uses: cjdoris/julia-downgrade-compat-action@v1
+        with:
+          skip: Pkg,TOML
+      - uses: julia-actions/cache@v1
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+      - uses: julia-actions/julia-buildpkg@v1
+      - uses: julia-actions/julia-runtest@v1

.github/workflows/documentation.yml

@@ -6,15 +6,16 @@ on:
       - main
     tags: '*'
   pull_request:
-
+  schedule:
+    - cron: '41 0 * * 5'
 jobs:
   build:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
       - uses: julia-actions/setup-julia@latest
         with:
-          version: '1.10'
+          version: '1'
       - name: Install dependencies
         run: julia --project=docs/ -e 'using Pkg; Pkg.develop(PackageSpec(path=pwd())); Pkg.instantiate()'
       - name: Build and deploy

Project.toml

@@ -7,7 +7,6 @@ version = "1.2.1"
 CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
 DiffEqBase = "2b5f629d-d688-5b77-993f-72d75c75574e"
 DocStringExtensions = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
-ExprTools = "e2ba6199-217a-4e67-a87a-7c52f15ade04"
 Flux = "587475ba-b771-5e3f-ad9e-33799f191a9c"
 Functors = "d9f16b24-f501-4c13-a1f2-28368ffc5196"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
@@ -18,19 +17,35 @@ SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StochasticDiffEq = "789caeaf-c7a9-5a7d-9973-96adeb23e2a0"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
-TestEnv = "1e6cf692-eddd-4d53-88a5-2d735e33781b"
 Tracker = "9f7883ad-71c0-57eb-9f7f-b5c9e6d3789c"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 cuDNN = "02a925ec-e4fe-4b08-9a7e-0d78e3d38ccd"
 
 [compat]
-CUDA = "3, 4, 5"
-DiffEqBase = "6"
-DocStringExtensions = "0.8, 0.9"
-ExprTools = "0.1"
-Flux = "0.13, 0.14"
-Functors = "0.2, 0.3, 0.4"
+Aqua = "0.8"
+CUDA = "4.4, 5"
+DiffEqBase = "6.137"
+DocStringExtensions = "0.9"
+Flux = "0.13.12, 0.14"
+Functors = "0.4"
+LinearAlgebra = "1.10"
+Random = "1.10"
 Reexport = "1"
-Statistics = "1"
-Zygote = "0.6"
-julia = "1.8, 1.9"
+SafeTestsets = "0.1"
+SciMLSensitivity = "7.49"
+SparseArrays = "1.10"
+Statistics = "1.10"
+StochasticDiffEq = "6.63"
+Test = "1.10"
+Tracker = "0.2.18"
+Zygote = "0.6.61"
+cuDNN = "1.1"
+julia = "1.10"
+
+[extras]
+Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
+SafeTestsets = "1bc83da4-3b8d-516f-aca4-4fe02f6d838f"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["Aqua", "Test", "SafeTestsets"]

src/DeepBSDE.jl

@@ -8,7 +8,7 @@
 ## Arguments
 - `u0`: a Flux.jl `Chain` with a d-dimensional input and a 1-dimensional output for the solytion guess.
 - `σᵀ∇u`: a Flux.jl `Chain` for the BSDE value guess.
-- `opt`: the optimization algorithm to be used to optimize the neural networks. Defaults to `ADAM(0.1)`.
+- `opt`: the optimization algorithm to be used to optimize the neural networks. Defaults to `Flux.Optimise.Adam(0.1)`.
 
 ## Example
 Black-Scholes-Barenblatt equation
@@ -28,7 +28,7 @@
 σ_f(X,p,t) = Diagonal(sigma*X) #Matrix d x d
 prob = PIDEProblem(g, f, μ_f, σ_f, x0, tspan)
 
-hls  = 10 + d #hiden layer size
+hls  = 10 + d #hidden layer size
 opt = Flux.Optimise.Adam(0.001)
 u0 = Flux.Chain(Dense(d,hls,relu),
                 Dense(hls,hls,relu),
@@ -188,7 +188,7 @@
     Flux.train!(loss_n_sde, ps, data, opt; cb = cb)
 
     if !limits
-        # Returning iters or simply u0(x0) and the tained neural network approximation u0
+        # Returning iters or simply u0(x0) and the trained neural network approximation u0
         if save_everystep
             sol = PIDESolution(x0, tspan[1]:dt:tspan[2], losses, iters, re1(p3))
         else
@@ -254,8 +254,8 @@
             true && println("Current loss is: $l")
             l < 1e-6 && Flux.stop()
         end
-        dataS = Iterators.repeated((), maxiters_limits)
-        Flux.train!(loss_, ps, dataS, ADAM(0.01); cb = cb)
+        dataS = Iterators.repeated((), maxiters_upper)
+        Flux.train!(loss_, ps, dataS, Flux.Optimise.Adam(0.01); cb = cb)
         u_high = loss_()
 
         verbose && println("Lower limit")

src/DeepSplitting.jl

@@ -1,10 +1,11 @@
-Base.copy(t::Tuple) = t # required for below
-function Base.copy(opt::O) where {O <: Flux.Optimise.AbstractOptimiser}
-    return O([copy(getfield(opt, f)) for f in fieldnames(typeof(opt))]...)
+_copy(t::Tuple) = t
+_copy(t) = t
+function _copy(opt::O) where  O<:Flux.Optimise.AbstractOptimiser
+    return O([_copy(getfield(opt,f)) for f in fieldnames(typeof(opt))]...)
 end
 
 """
-    DeepSplitting(nn, K=1, opt = ADAM(0.01), λs = nothing, mc_sample =  NoSampling())
+    DeepSplitting(nn, K=1, opt = Flux.Optimise.Adam(0.01), λs = nothing, mc_sample =  NoSampling())
 
 Deep splitting algorithm.
 
@@ -25,7 +26,7 @@ nn = Flux.Chain(Dense(d, hls, tanh),
                 Dense(hls,hls,tanh),
                 Dense(hls, 1, x->x^2))
 
-alg = DeepSplitting(nn, K=10, opt = ADAM(), λs = [5e-3,1e-3],
+alg = DeepSplitting(nn, K=10, opt = Flux.Optimise.Adam(), λs = [5e-3,1e-3],
                     mc_sample = UniformSampling(zeros(d), ones(d)) )
 ```
 """
@@ -39,7 +40,7 @@ end
 
 function DeepSplitting(nn;
         K = 1,
-        opt::O = ADAM(0.01),
+        opt::O = Flux.Optimise.Adam(0.01),
         λs::L = nothing,
         mc_sample = NoSampling()) where {
         O <: Flux.Optimise.AbstractOptimiser,
@@ -167,7 +168,7 @@ function DiffEqBase.solve(prob::PIDEProblem,
         _maxiters = length(maxiters) > 1 ? maxiters[min(net, 2)] : maxiters[]
 
         for λ in λs
-            opt_net = copy(opt) # starting with a new optimiser state at each time step
+            opt_net = _copy(opt) # starting with a new optimiser state at each time step
             opt_net.eta = λ
             verbose &&
                 println("Training started with ", typeof(opt_net), " and λ :", opt_net.eta)

test/DeepSplitting.jl

@@ -44,7 +44,7 @@ end
             Dense(hls, hls, relu),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(0.01) #optimiser
+        opt = Flux.Optimise.Adam(0.01) #optimiser
         alg = DeepSplitting(nn, opt = opt)
 
         f(y, z, v_y, v_z, ∇v_y, ∇v_z, p, t) = 0.0f0 .* v_y
@@ -88,7 +88,7 @@ end
             Dense(hls, hls, relu),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(0.01) #optimiser
+        opt = Flux.Optimise.Adam(0.01) #optimiser
         alg = DeepSplitting(nn, opt = opt)
 
         f(y, z, v_y, v_z, ∇v_y, ∇v_z, p, t) = 0.0f0 .* v_y #TODO: this fix is not nice
@@ -107,7 +107,7 @@ end
         u1_anal = [u_anal(x, tspan[end]) for x in eachcol(xs)]
         e_l2 = mean(rel_error_l2.(u1, u1_anal))
         println("rel_error_l2 = ", e_l2, "\n")
-        @test e_l2 < 0.13
+        @test e_l2 < 0.185
     end
 end
 
@@ -134,7 +134,7 @@ end
             Dense(hls, hls, relu),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(0.01) #optimiser
+        opt = Flux.Optimise.Adam(0.01) #optimiser
         alg = DeepSplitting(nn, opt = opt)
 
         f(y, z, v_y, v_z, ∇v_y, ∇v_z, p, t) = 0.0f0 .* v_y #TODO: this fix is not nice
@@ -192,7 +192,7 @@ end
             Dense(hls, hls, relu),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(0.01) #optimiser
+        opt = Flux.Optimise.Adam(0.01) #optimiser
         alg = DeepSplitting(nn, opt = opt)
 
         f(y, z, v_y, v_z, ∇v_y, ∇v_z, p, t) = r * v_y #TODO: this fix is not nice
@@ -234,8 +234,8 @@ end
             Dense(hls, hls, tanh),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(1e-3) #optimiser
-        alg = DeepSplitting(nn, opt = opt)
+        opt = Flux.Optimise.Adam(1e-3) #optimiser
+        alg = DeepSplitting(nn, opt = opt )
 
         X0 = fill(0.0f0, d)  # initial point
         g(X) = 1.0f0 ./ (2.0f0 .+ 4.0f-1 * sum(X .^ 2, dims = 1))   # initial condition
@@ -257,7 +257,7 @@ end
         u1 = sol.us[end]
         # value coming from \cite{Beck2017a}
         e_l2 = rel_error_l2(u1, 0.30879)
-        @test e_l2 < 0.5 # this is quite high as a relative error. 
+        @test e_l2 < 0.5 # this is quite high as a relative error.
         println("d = $d, rel_error_l2 = $e_l2")
     end
 end
@@ -281,8 +281,8 @@ end
             Dense(hls, hls, relu),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(1e-2) #optimiser
-        alg = DeepSplitting(nn, opt = opt)
+        opt = Flux.Optimise.Adam(1e-2) #optimiser
+        alg = DeepSplitting(nn, opt = opt )
 
         X0 = fill(0.0f0, d)  # initial point
         g(X) = exp.(-0.25f0 * sum(X .^ 2, dims = 1))   # initial condition
@@ -329,8 +329,8 @@ if false
                 Dense(hls, hls, tanh),
                 Dense(hls, 1)) # Neural network used by the scheme
 
-            opt = ADAM(1e-3) #optimiser
-            alg = DeepSplitting(nn, opt = opt)
+            opt = Flux.Optimise.Adam(1e-3) #optimiser
+            alg = DeepSplitting(nn, opt = opt )
 
             X0 = repeat([1.0f0, 0.5f0], div(d, 2))  # initial point
             g(X) = sum(X .^ 2, dims = 1) # initial condition
@@ -381,8 +381,8 @@ if false
             Dense(hls, hls, tanh),
             Dense(hls, 1)) # Neural network used by the scheme
 
-        opt = ADAM(1e-3) #optimiser
-        alg = DeepSplitting(nn, opt = opt)
+        opt = Flux.Optimise.Adam(1e-3) #optimiser
+        alg = DeepSplitting(nn, opt = opt )
 
         X0 = fill(0.0f0, d)  # initial point
         g(X) = log.(5.0f-1 .+ 5.0f-1 * sum(X .^ 2, dims = 1)) # initial condition
@@ -430,8 +430,8 @@ end
         Dense(hls, hls, tanh),
         Dense(hls, 1)) # Neural network used by the scheme
 
-    opt = ADAM()
-    alg = DeepSplitting(nn, opt = opt, λs = [1e-2, 1e-3])
+    opt = Flux.Optimise.Adam()
+    alg = DeepSplitting(nn, opt = opt, λs = [1e-2,1e-3] )
 
     X0 = fill(100.0f0, d)  # initial point
     g(X) = minimum(X, dims = 1) # initial condition
@@ -519,8 +519,8 @@ end
             # BatchNorm(hls, affine = true, dim = 1),
             Dense(hls, 1, x -> x^2)) # positive function
 
-        opt = ADAM(1e-2)#optimiser
-        alg = DeepSplitting(nn_batch, K = K, opt = opt, mc_sample = x0_sample)
+        opt = Flux.Optimise.Adam(1e-2)#optimiser
+        alg = DeepSplitting(nn_batch, K=K, opt = opt, mc_sample = x0_sample)
 
         function g(x)
             Float32((2 * π)^(-d / 2)) * ss0^(-Float32(d) * 5.0f-1) *
@@ -575,8 +575,8 @@ end
                 Dense(hls, hls, tanh),
                 Dense(hls, 1)) # Neural network used by the scheme
 
-            opt = ADAM(1e-2) #optimiser
-            alg = DeepSplitting(nn, K = K, opt = opt, mc_sample = UniformSampling(-∂, ∂))
+            opt = Flux.Optimise.Adam(1e-2) #optimiser
+            alg = DeepSplitting(nn, K=K, opt = opt, mc_sample = UniformSampling(-∂, ∂) )
 
             x0 = fill(0.0f0, d)  # initial point
             g(X) = exp.(-0.25f0 * sum(X .^ 2, dims = 1))   # initial condition

test/qa.jl

@@ -0,0 +1,11 @@
+using HighDimPDE, Aqua
+@testset "Aqua" begin
+    Aqua.find_persistent_tasks_deps(HighDimPDE)
+    Aqua.test_ambiguities(HighDimPDE, recursive = false)
+    Aqua.test_deps_compat(HighDimPDE)
+    Aqua.test_piracies(HighDimPDE)
+    Aqua.test_project_extras(HighDimPDE)
+    Aqua.test_stale_deps(HighDimPDE)
+    Aqua.test_unbound_args(HighDimPDE)
+    Aqua.test_undefined_exports(HighDimPDE)
+end