Use DI.Cache

Author: devmotion

Project.toml

@@ -7,24 +7,25 @@ ADTypes = "47edcb42-4c32-4615-8424-f2b9edc5f35b"
 DifferentiationInterface = "a0c0ee7d-e4b9-4e03-894e-1c5f64a51d63"
 Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 FiniteDiff = "6a86dc24-6348-571c-b903-95158fe2bd41"
+LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 
 [compat]
 ADTypes = "1.11.0"
 DifferentiationInterface = "0.6.43, 0.7"
 FiniteDiff = "2.0"
 ForwardDiff = "0.10, 1.0"
+LinearAlgebra = "<0.0.1, 1"
 julia = "1.10"
 
 [extras]
 ADTypes = "47edcb42-4c32-4615-8424-f2b9edc5f35b"
 ComponentArrays = "b0b7db55-cfe3-40fc-9ded-d10e2dbeff66"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
-LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 OptimTestProblems = "cec144fc-5a64-5bc6-99fb-dde8f63e154c"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 RecursiveArrayTools = "731186ca-8d62-57ce-b412-fbd966d074cd"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["ADTypes", "ComponentArrays", "ForwardDiff", "LinearAlgebra", "OptimTestProblems", "Random", "RecursiveArrayTools", "SparseArrays", "Test"]
+test = ["ADTypes", "ComponentArrays", "ForwardDiff", "OptimTestProblems", "Random", "RecursiveArrayTools", "SparseArrays", "Test"]

src/NLSolversBase.jl

@@ -5,6 +5,7 @@ module NLSolversBase
 using ADTypes: AbstractADType, AutoFiniteDiff
 import DifferentiationInterface as DI
 using FiniteDiff: FiniteDiff
+using LinearAlgebra: LinearAlgebra
 import Distributed: clear!
 export AbstractObjective,
        NonDifferentiable,

src/objective_types/constraints.jl

@@ -163,13 +163,6 @@ function TwiceDifferentiableConstraints(c!, lx::AbstractVector, ux::AbstractVect
     x_example = zeros(T, nx)
     λ_example = zeros(T, nc)
     ccache = zeros(T, nc)
-   
-    function sum_constraints(_x, _λ)
-        # TODO: get rid of this allocation with DI.Cache
-        ccache_righttype = zeros(promote_type(T, eltype(_x)), nc)
-        c!(ccache_righttype, _x)
-        return sum(_λ[i] * ccache[i] for i in eachindex(_λ, ccache))
-    end
 
     jac_prep = DI.prepare_jacobian(c!, ccache, autodiff, x_example)
     con_jac! = let c! = c!, ccache = ccache, jac_prep = jac_prep, autodiff = autodiff
@@ -178,11 +171,15 @@ function TwiceDifferentiableConstraints(c!, lx::AbstractVector, ux::AbstractVect
             return _j
         end
     end
-
-    hess_prep = DI.prepare_hessian(sum_constraints, autodiff, x_example, DI.Constant(λ_example))
-    con_hess! = let sum_constraints = sum_constraints, hess_prep = hess_prep, autodiff = autodiff
+    
+    function sum_constraints(_x, _λ, _ccache)
+        c!(_ccache, _x)
+        return LinearAlgebra.dot(_λ, _ccache)
+    end
+    hess_prep = DI.prepare_hessian(sum_constraints, autodiff, x_example, DI.Constant(λ_example), DI.Cache(ccache))
+    con_hess! = let sum_constraints = sum_constraints, hess_prep = hess_prep, autodiff = autodiff, ccache = ccache
         function (_h, _x, _λ)
-            DI.hessian!(sum_constraints, _h, hess_prep, autodiff, _x, DI.Constant(_λ)) 
+            DI.hessian!(sum_constraints, _h, hess_prep, autodiff, _x, DI.Constant(_λ), DI.Cache(ccache)) 
             return _h
         end
     end