From 0c20f1f71edfdcf044c4b134f95df8a6c989d2e1 Mon Sep 17 00:00:00 2001
From: jwilson <jwilson@psl.nmsu.edu>
Date: Fri, 28 Jul 2023 23:19:02 -0600
Subject: [PATCH] Added simple CuArray patch

---
 Project.toml      |  1 +
 src/derivative.jl |  7 +++++++
 test/cuda.jl      | 19 +++++++++++++++++++
 3 files changed, 27 insertions(+)
 create mode 100644 test/cuda.jl

diff --git a/Project.toml b/Project.toml
index 7ad1b28c..a41e00e7 100644
--- a/Project.toml
+++ b/Project.toml
@@ -4,6 +4,7 @@ authors = ["Songchen Tan <i@tansongchen.com>"]
 version = "0.2.1"
 
 [deps]
+CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
 ChainRules = "082447d4-558c-5d27-93f4-14fc19e9eca2"
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
 ChainRulesOverloadGeneration = "f51149dc-2911-5acf-81fc-2076a2a81d4f"
diff --git a/src/derivative.jl b/src/derivative.jl
index ef46a1e0..4b442bdb 100644
--- a/src/derivative.jl
+++ b/src/derivative.jl
@@ -1,5 +1,6 @@
 
 export derivative
+using CUDA
 
 """
     derivative(f, x::T, order::Int64)
@@ -23,6 +24,12 @@ end
     derivative(f, x, l, Val{order + 1}())
 end
 
+# add CUDA support
+@inline function derivative(f, x::CuArray, l::CuArray,
+    order::Int64) 
+derivative(f, x, l, Val{order + 1}())
+end
+
 @inline function derivative(f, x::T, ::Val{N}) where {T <: Number, N}
     t = TaylorScalar{T, N}(x, one(x))
     return extract_derivative(f(t), N)
diff --git a/test/cuda.jl b/test/cuda.jl
new file mode 100644
index 00000000..0ebf6699
--- /dev/null
+++ b/test/cuda.jl
@@ -0,0 +1,19 @@
+using CUDA
+
+# cpu
+f(x,y) = x^2 * y^2
+input_cpu = [1e0,1e0]
+derivative(temp -> f(temp[1],temp[2]),input_cpu,[1e0,0e0],2)
+
+# gpu
+f(x,y) = x^2 * y^2
+input_gpu = CuArray([1e0,1e0])
+direction_gpu = CuArray([1e0,0e0])
+derivative(temp -> f(temp[1],temp[2]),input_gpu,direction_gpu,2)
+
+
+
+
+
+
+