Merge branch 'SciML:master' into master

Project.toml

@@ -1,7 +1,7 @@
 name = "NeuralPDE"
 uuid = "315f7962-48a3-4962-8226-d0f33b1235f0"
 authors = ["Chris Rackauckas <[email protected]>"]
-version = "5.11.0"
+version = "5.12.0"
 
 [deps]
 Adapt = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
@@ -33,25 +33,24 @@ Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 RuntimeGeneratedFunctions = "7e49a35a-f44a-4d26-94aa-eba1b4ca6b47"
 SciMLBase = "0bca4576-84f4-4d90-8ffe-ffa030f20462"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
-StochasticDiffEq = "789caeaf-c7a9-5a7d-9973-96adeb23e2a0"
 SymbolicUtils = "d1185830-fcd6-423d-90d6-eec64667417b"
 Symbolics = "0c5d862f-8b57-4792-8d23-62f2024744c7"
 UnPack = "3a884ed6-31ef-47d7-9d2a-63182c4928ed"
 Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [compat]
-Adapt = "3, 4"
+Adapt = "4"
 AdvancedHMC = "0.6"
-ArrayInterface = "6, 7"
 Aqua = "0.8"
-CUDA = "4"
+ArrayInterface = "7"
+CUDA = "5.1"
 ChainRulesCore = "1"
-ComponentArrays = "0.13.2, 0.14, 0.15"
+ComponentArrays = "0.15"
 Cubature = "1.5"
 DiffEqBase = "6"
 DiffEqNoiseProcess = "5.1"
-Distributions = "0.23, 0.24, 0.25"
-DocStringExtensions = "0.8, 0.9"
+Distributions = "0.25"
+DocStringExtensions = "0.9"
 DomainSets = "0.6, 0.7"
 Flux = "0.14"
 ForwardDiff = "0.10"
@@ -60,15 +59,15 @@ Integrals = "4"
 LineSearches = "7.2"
 LinearAlgebra = "1"
 LogDensityProblems = "2"
-Lux = "0.4, 0.5"
+Lux = "0.5"
 LuxCUDA = "0.3"
 MCMCChains = "6"
 ModelingToolkit = "8"
 MonteCarloMeasurements = "1"
 Optim = "1.7.8"
 Optimization = "3"
-OptimizationOptimJL = "0.1"
-OptimizationOptimisers = "0.1"
+OptimizationOptimJL = "0.2"
+OptimizationOptimisers = "0.2"
 OrdinaryDiffEq = "6"
 Pkg = "1"
 QuasiMonteCarlo = "0.3.2"
@@ -78,7 +77,6 @@ RuntimeGeneratedFunctions = "0.5"
 SafeTestsets = "0.1"
 SciMLBase = "2"
 Statistics = "1"
-StochasticDiffEq = "6.13"
 SymbolicUtils = "1"
 Symbolics = "5"
 Test = "1"

docs/Project.toml

@@ -26,19 +26,19 @@ SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
 AdvancedHMC = "0.6"
 Cubature = "1.5"
 DiffEqBase = "6.106"
-Distributions = "0.23, 0.24, 0.25"
+Distributions = "0.25"
 Documenter = "1"
 DomainSets = "0.6, 0.7"
-Flux = "0.13, 0.14"
+Flux = "0.14"
 Integrals = "4"
-Lux = "0.4, 0.5"
+Lux = "0.5"
 ModelingToolkit = "8.33"
 MonteCarloMeasurements = "1"
 NeuralPDE = "5.3"
 Optimization = "3.9"
-OptimizationOptimJL = "0.1"
-OptimizationOptimisers = "0.1"
-OptimizationPolyalgorithms = "0.1"
+OptimizationOptimJL = "0.2"
+OptimizationOptimisers = "0.2"
+OptimizationPolyalgorithms = "0.2"
 OrdinaryDiffEq = "6.31"
 Plots = "1.36"
 QuasiMonteCarlo = "0.3"

lib/NeuralPDELogging/test/adaptive_loss_log_tests.jl

@@ -3,7 +3,6 @@ using Test, NeuralPDE
 using Optimization, OptimizationOptimisers
 import ModelingToolkit: Interval, infimum, supremum
 using Random, Lux
-#using Plots
 @info "Starting Soon!"
 
 nonadaptive_loss = NeuralPDE.NonAdaptiveLoss(pde_loss_weights = 1, bc_loss_weights = 1)
@@ -70,7 +69,7 @@ function test_2d_poisson_equation_adaptive_loss(adaptive_loss, run, outdir, hasl
         if haslogger
             log_value(logger, "outer_error/loss", l, step = iteration[1])
             if iteration[1] % 30 == 0
-                u_predict = reshape([first(phi([x, y], p)) for x in xs for y in ys],
+                u_predict = reshape([first(phi([x, y], p.u)) for x in xs for y in ys],
                                     (length(xs), length(ys)))
                 diff_u = abs.(u_predict .- u_real)
                 total_diff = sum(diff_u)

src/NeuralPDE.jl

@@ -9,7 +9,7 @@ using Reexport, Statistics
 @reexport using ModelingToolkit
 
 using Zygote, ForwardDiff, Random, Distributions
-using Adapt, DiffEqNoiseProcess, StochasticDiffEq
+using Adapt, DiffEqNoiseProcess
 using Optimization
 using OptimizationOptimisers
 using Integrals, Cubature

src/pinn_types.jl

@@ -507,10 +507,6 @@ function (f::Phi{<:Lux.AbstractExplicitLayer})(x::AbstractArray, θ)
     y
 end
 
-function (f::Phi{<:Optimisers.Restructure})(x, θ)
-    f.f(θ)(adapt(parameterless_type(θ), x))
-end
-
 function get_u()
     u = (cord, θ, phi) -> phi(cord, θ)
 end

test/NNPDE_tests.jl

@@ -173,8 +173,8 @@ function test_2d_poisson_equation(chain_, strategy_)
         u(x, 0) ~ 0.0, u(x, 1) ~ -sin(pi * x) * sin(pi * 1)]
     # Space and time domains
     domains = [x ∈ Interval(0.0, 1.0), y ∈ Interval(0.0, 1.0)]
-
-    discretization = PhysicsInformedNN(chain_, strategy_)
+    ps = Lux.setup(Random.default_rng(), chain_)[1]
+    discretization = PhysicsInformedNN(chain_, strategy_; init_params = ps)
     @named pde_system = PDESystem(eq, bcs, domains, [x, y], [u(x, y)])
     prob = discretize(pde_system, discretization)
     res = solve(prob, OptimizationOptimisers.Adam(0.1); maxiters = 500, cb = callback)
@@ -251,8 +251,8 @@ end
 
     cb_ = function (p, l)
         println("loss: ", l)
-        println("pde_losses: ", map(l_ -> l_(p), pde_inner_loss_functions))
-        println("bcs_losses: ", map(l_ -> l_(p), bcs_inner_loss_functions))
+        println("pde_losses: ", map(l_ -> l_(p.u), pde_inner_loss_functions))
+        println("bcs_losses: ", map(l_ -> l_(p.u), bcs_inner_loss_functions))
         return false
     end
 
@@ -352,7 +352,7 @@ end
 
     cb_ = function (p, l)
         println("loss: ", l)
-        println("losses: ", map(l -> l(p), loss_functions))
+        println("losses: ", map(l -> l(p.u), loss_functions))
         return false
     end
 

test/additional_loss_tests.jl

@@ -57,9 +57,9 @@ using ComponentArrays
     phi = discretization.phi
     cb_ = function (p, l)
         println("loss: ", l)
-        println("pde_losses: ", map(l_ -> l_(p), pde_inner_loss_functions))
-        println("bcs_losses: ", map(l_ -> l_(p), bcs_inner_loss_functions))
-        println("additional_loss: ", norm_loss_function(phi, p, nothing))
+        println("pde_losses: ", map(l_ -> l_(p.u), pde_inner_loss_functions))
+        println("bcs_losses: ", map(l_ -> l_(p.u), bcs_inner_loss_functions))
+        println("additional_loss: ", norm_loss_function(phi, p.u, nothing))
         return false
     end
     res = solve(prob, OptimizationOptimJL.LBFGS(), maxiters = 400, callback = cb_)
@@ -82,9 +82,9 @@ using ComponentArrays
     phi = discretization.phi
     cb_ = function (p, l)
         println("loss: ", l)
-        println("pde_losses: ", map(l_ -> l_(p), pde_inner_loss_functions))
-        println("bcs_losses: ", map(l_ -> l_(p), bcs_inner_loss_functions))
-        println("additional_loss: ", norm_loss_function(phi, p, nothing))
+        println("pde_losses: ", map(l_ -> l_(p.u), pde_inner_loss_functions))
+        println("bcs_losses: ", map(l_ -> l_(p.u), bcs_inner_loss_functions))
+        println("additional_loss: ", norm_loss_function(phi, p.u, nothing))
         return false
     end
     res = solve(prob, OptimizationOptimJL.LBFGS(), maxiters = 400, callback = cb_)