Merge pull request #910 from SciML/datagen

Real valued datagen from GridTraining when boundary points are trivial
SciML · Nov 9, 2024 · ff6d507 · ff6d507
2 parents ccce83e + 21f5b01
commit ff6d507
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 0 deletions.
diff --git a/src/eltype_matching.jl b/src/eltype_matching.jl
@@ -9,6 +9,10 @@ function Adapt.adapt_storage(::EltypeAdaptor{T}, x::AbstractArray) where {T}
     return convert(AbstractArray{T}, x)
 end
 
+function Adapt.adapt_storage(::EltypeAdaptor{T}, x::AbstractArray{Union{}}) where {T}
+    return convert(AbstractArray{T}, x)
+end
+
 function Adapt.adapt_storage(::EltypeAdaptor{T}, x::AbstractArray{<:Complex}) where {T}
     return convert(AbstractArray{Complex{T}}, x)
 end
diff --git a/test/NNPDE_tests.jl b/test/NNPDE_tests.jl
@@ -2,6 +2,13 @@
 
 using NeuralPDE, Cubature, Integrals, QuasiMonteCarlo
 
+# DataGen is Real: https://github.com/SciML/NeuralPDE.jl/issues/906
+@parameters x
+@variables u(..)
+
+NeuralPDE.generate_training_sets(
+    [x ∈ (-1.0, 1.0)], 0.1, [u(x) ~ x], [0.0 ~ 0.0], Float64, [x], [:u])
+
 function callback(p, l)
     if p.iter == 1 || p.iter % 250 == 0
         println("Current loss is: $l after $(p.iter) iterations")