Bump compat for OrdinaryDiffEq v7 / SciMLBase v3 ecosystem

.github/workflows/FormatCheck.yml

@@ -14,6 +14,9 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v6
+      - uses: julia-actions/setup-julia@v2
+        with:
+          version: '1'
       - uses: fredrikekre/runic-action@v1
         with:
           version: '1'

Project.toml

@@ -27,7 +27,7 @@ ModelingToolkitStandardLibrary = "2.24"
 OptimizationBase = "4.0.2, 5"
 OptimizationOptimJL = "0.4.8"
 OptimizationOptimisers = "0.3"
-OrdinaryDiffEqVerner = "1.2"
+OrdinaryDiffEqVerner = "1, 2"
 Random = "1.10"
 SafeTestsets = "0.1"
 SciCompDSL = "1"

docs/Project.toml

@@ -33,10 +33,10 @@ ModelingToolkitStandardLibrary = "2.7"
 Optimization = "4.0, 5"
 OptimizationOptimJL = "0.4"
 OptimizationOptimisers = "0.3"
-OrdinaryDiffEqTsit5 = "1"
-OrdinaryDiffEqVerner = "1"
+OrdinaryDiffEqTsit5 = "1, 2"
+OrdinaryDiffEqVerner = "1, 2"
 Plots = "1"
-SciMLBase = "2"
+SciMLBase = "2, 3"
 SciMLSensitivity = "7"
 SciMLStructures = "1.1.0"
 StableRNGs = "1"

src/ModelingToolkitNeuralNets.jl

@@ -208,7 +208,7 @@ function make_symbolic_nn_declaration(expr::Expr)
     end
 
     # Extracts individual component symbols.
-    nn,p = expr.args[1].args
+    nn, p = expr.args[1].args
     chain, rng = if Meta.isexpr(expr.args[2], :tuple)
         if (length(expr.args[2].args) > 2)
             error("@SymbolicNeuralNetwork accepts no more than 2 inputs on the right-hand side.")
@@ -219,9 +219,13 @@ function make_symbolic_nn_declaration(expr::Expr)
     end
 
     # Constructs the output expression.
-    snn_dec = :(($nn, $p) = SymbolicNeuralNetwork(; chain = $chain, nn_name = $(QuoteNode(nn)),
-        nn_p_name = $(QuoteNode(p)), n_input = ModelingToolkitNeuralNets._num_chain_inputs($chain),
-        n_output = ModelingToolkitNeuralNets._num_chain_outputs($chain)))
+    snn_dec = :(
+        ($nn, $p) = SymbolicNeuralNetwork(;
+            chain = $chain, nn_name = $(QuoteNode(nn)),
+            nn_p_name = $(QuoteNode(p)), n_input = ModelingToolkitNeuralNets._num_chain_inputs($chain),
+            n_output = ModelingToolkitNeuralNets._num_chain_outputs($chain)
+        )
+    )
     if !isnothing(rng)
         push!(snn_dec.args[2].args[2].args, Expr(:kw, :rng, rng))
     end

src/nn_par_accessors.jl

@@ -21,7 +21,7 @@ ModelingToolkitNeuralNets.isneuralnetwork(θ) # false
 """
 isneuralnetwork(p::Union{Symbolics.Num, Symbolics.Arr, Symbolics.CallAndWrap}) = isneuralnetwork(Symbolics.unwrap(p))
 function isneuralnetwork(p::Symbolics.SymbolicT)
-    getmetadata(p, NeuralNetworkParameter, false)
+    return getmetadata(p, NeuralNetworkParameter, false)
 end
 
 """
@@ -40,7 +40,7 @@ ModelingToolkitNeuralNets.isneuralnetworkps(θ) # true
 """
 isneuralnetworkps(p::Union{Symbolics.Num, Symbolics.Arr, Symbolics.CallAndWrap}) = isneuralnetworkps(Symbolics.unwrap(p))
 function isneuralnetworkps(p::Symbolics.SymbolicT)
-    getmetadata(p, NeuralNetworkParametrisation, false)
+    return getmetadata(p, NeuralNetworkParametrisation, false)
 end
 
 

test/nn_ps_accessors.jl

@@ -39,8 +39,8 @@ let
     @variables X(t) Y(t)
     @parameters d
     eqs = [
-        D(X) ~ NN([X], θ)[1] - d*X
-        D(Y) ~ X - d*Y
+        D(X) ~ NN([X], θ)[1] - d * X
+        D(Y) ~ X - d * Y
     ]
     @mtkcompile sys = System(eqs, t)
 
@@ -104,8 +104,8 @@ let
     @variables X(t) Y(t)
     @parameters d
     eqs = [
-        D(X) ~ NN([X], θ)[1] - d*X
-        D(Y) ~ X - d*Y
+        D(X) ~ NN([X], θ)[1] - d * X
+        D(Y) ~ X - d * Y
     ]
     @mtkcompile sys = System(eqs, t)
 

test/symbolicnn_macro.jl

@@ -24,7 +24,6 @@ let
 end
 
 
-
 # Checks that symbolic networks declared with/without the macro are identical (1).
 let
     # Declares the neural networks.
@@ -86,20 +85,20 @@ let
     @SymbolicNeuralNetwork NN, p = chain
     NN_func, p_func = SymbolicNeuralNetwork(; chain, n_input = 1, n_output = 1, nn_name = :NN, nn_p_name = :p)
 
-        # Checks that they are identical.
+    # Checks that they are identical.
     @test isequal(NN, NN_func)
     @test isequal(p, p_func)
 
     # Creates corresponding MTK models.
     @variables X(t) Y(t)
     @parameters d
     eqs_macro = [
-        D(X) ~ NN([X], p)[1] - d*X
-        D(Y) ~ X - d*Y
+        D(X) ~ NN([X], p)[1] - d * X
+        D(Y) ~ X - d * Y
     ]
     eqs_func = [
-        D(X) ~ NN_func([X], p_func)[1] - d*X
-        D(Y) ~ X - d*Y
+        D(X) ~ NN_func([X], p_func)[1] - d * X
+        D(Y) ~ X - d * Y
     ]
     @mtkcompile sys_macro = System(eqs_macro, t)
     @mtkcompile sys_func = System(eqs_func, t)
@@ -149,8 +148,8 @@ let
 
     # Checks that non-supported neural network architectures throw errors.
     @test_throws Exception @eval @SymbolicNeuralNetwork NN, p = Lux.Chain(
-        Lux.Conv((3, 3), 1 => 8, Lux.relu; pad=1),
-        Lux.Conv((3, 3), 8 => 16, Lux.relu; pad=1),
+        Lux.Conv((3, 3), 1 => 8, Lux.relu; pad = 1),
+        Lux.Conv((3, 3), 8 => 16, Lux.relu; pad = 1),
         Lux.Dense(28 * 28 * 16 => 10)
     )
 end