Improve CompressedSensingIPM.jl (#25)

Author: amontoison

Project.toml

@@ -17,10 +17,10 @@ SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 [compat]
 AMDGPU = "2.0.1"
 CUDA = "5.8.2"
-FFTW = "1.8.1"
+FFTW = "1.9.0"
 Krylov = "0.10.1"
 LinearAlgebra = "1.10"
-MadNLP = "0.8.9"
+MadNLP = "0.8.10"
 MadNLPGPU = "0.7.13"
 NLPModels = "0.21.5"
 Random = "1.10"

benchmarks/Manifest.toml

@@ -12,9 +12,9 @@ version = "0.5.3"
 
 [[deps.AMDGPU]]
 deps = ["AbstractFFTs", "AcceleratedKernels", "Adapt", "Atomix", "CEnum", "ExprTools", "GPUArrays", "GPUCompiler", "GPUToolbox", "KernelAbstractions", "LLD_jll", "LLVM", "LLVM_jll", "Libdl", "LinearAlgebra", "Pkg", "Preferences", "PrettyTables", "Printf", "ROCmDeviceLibs_jll", "Random", "Random123", "RandomNumbers", "SparseArrays", "SpecialFunctions", "StaticArraysCore", "Statistics", "UnsafeAtomics"]
-git-tree-sha1 = "0cd47f7233bea4cf0201eb0047f532176da2621d"
+git-tree-sha1 = "7508a4e161f28689740a7ef9dcc907e679d1a735"
 uuid = "21141c5a-9bdb-4563-92ae-f87d6854732e"
-version = "1.3.4"
+version = "2.0.1"
 
     [deps.AMDGPU.extensions]
     AMDGPUChainRulesCoreExt = "ChainRulesCore"
@@ -39,17 +39,15 @@ version = "1.5.0"
     Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [[deps.AcceleratedKernels]]
-deps = ["ArgCheck", "GPUArraysCore", "KernelAbstractions", "Markdown"]
-git-tree-sha1 = "bf0d8cca80cdf7f0f6d9851d94d42a5db862f1b9"
+deps = ["ArgCheck", "GPUArraysCore", "KernelAbstractions", "Markdown", "UnsafeAtomics"]
+git-tree-sha1 = "0de01460ed11e90b42ce666c8ed0265bad59aa6a"
 uuid = "6a4ca0a5-0e36-4168-a932-d9be78d558f1"
-version = "0.4.2"
+version = "0.4.3"
 
     [deps.AcceleratedKernels.extensions]
-    AcceleratedKernelsMetalExt = "Metal"
     AcceleratedKernelsoneAPIExt = "oneAPI"
 
     [deps.AcceleratedKernels.weakdeps]
-    Metal = "dde4c033-4e86-420c-a63e-0dd931031962"
     oneAPI = "8f75cd03-7ff8-4ecb-9b8f-daf728133b1b"
 
 [[deps.Adapt]]
@@ -84,9 +82,9 @@ version = "1.11.0"
 
 [[deps.Atomix]]
 deps = ["UnsafeAtomics"]
-git-tree-sha1 = "b5bb4dc6248fde467be2a863eb8452993e74d402"
+git-tree-sha1 = "29bb0eb6f578a587a49da16564705968667f5fa8"
 uuid = "a9b6321e-bd34-4604-b9c9-b65b8de01458"
-version = "1.1.1"
+version = "1.1.2"
 
     [deps.Atomix.extensions]
     AtomixCUDAExt = "CUDA"
@@ -189,9 +187,9 @@ version = "0.13.1"
 
 [[deps.Compat]]
 deps = ["TOML", "UUIDs"]
-git-tree-sha1 = "3a3dfb30697e96a440e4149c8c51bf32f818c0f3"
+git-tree-sha1 = "0037835448781bb46feb39866934e243886d756a"
 uuid = "34da2185-b29b-5c13-b0c7-acf172513d20"
-version = "4.17.0"
+version = "4.18.0"
 weakdeps = ["Dates", "LinearAlgebra"]
 
     [deps.Compat.extensions]
@@ -220,15 +218,15 @@ version = "1.16.0"
 
 [[deps.DataFrames]]
 deps = ["Compat", "DataAPI", "DataStructures", "Future", "InlineStrings", "InvertedIndices", "IteratorInterfaceExtensions", "LinearAlgebra", "Markdown", "Missings", "PooledArrays", "PrecompileTools", "PrettyTables", "Printf", "Random", "Reexport", "SentinelArrays", "SortingAlgorithms", "Statistics", "TableTraits", "Tables", "Unicode"]
-git-tree-sha1 = "fb61b4812c49343d7ef0b533ba982c46021938a6"
+git-tree-sha1 = "a37ac0840a1196cd00317b57e39d6586bf0fd6f6"
 uuid = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
-version = "1.7.0"
+version = "1.7.1"
 
 [[deps.DataStructures]]
-deps = ["Compat", "InteractiveUtils", "OrderedCollections"]
-git-tree-sha1 = "4e1fe97fdaed23e9dc21d4d664bea76b65fc50a0"
+deps = ["OrderedCollections"]
+git-tree-sha1 = "6c72198e6a101cccdd4c9731d3985e904ba26037"
 uuid = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"
-version = "0.18.22"
+version = "0.19.1"
 
 [[deps.DataValueInterfaces]]
 git-tree-sha1 = "bfc1187b79289637fa0ef6d4436ebdfe6905cbd6"
@@ -368,9 +366,9 @@ uuid = "34004b35-14d8-5ef3-9330-4cdb6864b03a"
 version = "0.3.28"
 
 [[deps.InlineStrings]]
-git-tree-sha1 = "8594fac023c5ce1ef78260f24d1ad18b4327b420"
+git-tree-sha1 = "8f3d257792a522b4601c24a577954b0a8cd7334d"
 uuid = "842dd82b-1e85-43dc-bf29-5d0ee9dffc48"
-version = "1.4.4"
+version = "1.4.5"
 
     [deps.InlineStrings.extensions]
     ArrowTypesExt = "ArrowTypes"
@@ -382,9 +380,9 @@ version = "1.4.4"
 
 [[deps.IntelOpenMP_jll]]
 deps = ["Artifacts", "JLLWrappers", "LazyArtifacts", "Libdl"]
-git-tree-sha1 = "0f14a5456bdc6b9731a5682f439a672750a09e48"
+git-tree-sha1 = "ec1debd61c300961f98064cfb21287613ad7f303"
 uuid = "1d5cc7b8-4909-519e-a0f8-d0f5ad9712d0"
-version = "2025.0.4+0"
+version = "2025.2.0+0"
 
 [[deps.InteractiveUtils]]
 deps = ["Markdown"]
@@ -408,9 +406,9 @@ version = "1.0.0"
 
 [[deps.JLLWrappers]]
 deps = ["Artifacts", "Preferences"]
-git-tree-sha1 = "a007feb38b422fbdab534406aeca1b86823cb4d6"
+git-tree-sha1 = "0533e564aae234aff59ab625543145446d8b6ec2"
 uuid = "692b3bcd-3c85-4b1f-b108-f13ce0eb3210"
-version = "1.7.0"
+version = "1.7.1"
 
 [[deps.JSON]]
 deps = ["Dates", "Mmap", "Parsers", "Unicode"]
@@ -426,9 +424,9 @@ version = "0.2.1+0"
 
 [[deps.KernelAbstractions]]
 deps = ["Adapt", "Atomix", "InteractiveUtils", "MacroTools", "PrecompileTools", "Requires", "StaticArrays", "UUIDs"]
-git-tree-sha1 = "4efa9cec6f308e0f492ea635421638bff81cf6f8"
+git-tree-sha1 = "83c617e9e9b02306a7acab79e05ec10253db7c87"
 uuid = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
-version = "0.9.36"
+version = "0.9.38"
 
     [deps.KernelAbstractions.extensions]
     EnzymeExt = "EnzymeCore"
@@ -493,7 +491,7 @@ version = "1.4.0"
 deps = ["LinearAlgebra", "SparseArrays"]
 git-tree-sha1 = "71297e2bb9f92e675dfd218cc1b309ae180ffa15"
 repo-rev = "master"
-repo-url = "https://github.com/vishwas1984/LaplaceInterpolation.jl.git"
+repo-url = "https://github.com/vishwas1984/LaplaceInterpolation.jl"
 uuid = "afb1b68e-e51f-4e49-a799-0e085170423d"
 version = "0.1.1"
 
@@ -544,9 +542,9 @@ version = "1.11.0"
 
 [[deps.LinearOperators]]
 deps = ["FastClosures", "LinearAlgebra", "Printf", "Requires", "SparseArrays", "TimerOutputs"]
-git-tree-sha1 = "1894a798ed8887895c5ae70f1fe8331c0c1d8480"
+git-tree-sha1 = "db137007d2c4ed948aa5f2518a2b451851ea8bda"
 uuid = "5c8ed15e-5a4c-59e4-a42b-c7e8811fb125"
-version = "2.10.0"
+version = "2.11.0"
 
     [deps.LinearOperators.extensions]
     LinearOperatorsAMDGPUExt = "AMDGPU"
@@ -592,9 +590,9 @@ version = "5.1.3+0"
 
 [[deps.MKL_jll]]
 deps = ["Artifacts", "IntelOpenMP_jll", "JLLWrappers", "LazyArtifacts", "Libdl", "oneTBB_jll"]
-git-tree-sha1 = "5de60bc6cb3899cd318d80d627560fae2e2d99ae"
+git-tree-sha1 = "282cadc186e7b2ae0eeadbd7a4dffed4196ae2aa"
 uuid = "856f044c-d86e-5d09-b602-aeab76dc8ba7"
-version = "2025.0.1+1"
+version = "2025.2.0+0"
 
 [[deps.MacroTools]]
 git-tree-sha1 = "1e0228a030642014fe5cfe68c2c0a818f9e3f522"
@@ -603,9 +601,9 @@ version = "0.5.16"
 
 [[deps.MadNLP]]
 deps = ["LDLFactorizations", "LinearAlgebra", "Logging", "NLPModels", "Pkg", "Printf", "SolverCore", "SparseArrays", "SuiteSparse"]
-git-tree-sha1 = "70ae9a1212ae748a19926db8b59642044709af13"
+git-tree-sha1 = "6b8cda243ee56472853b92bd131da5998ce95d71"
 uuid = "2621e9c9-9eb4-46b1-8089-e8c72242dfb6"
-version = "0.8.8"
+version = "0.8.10"
 
     [deps.MadNLP.extensions]
     MadNLPMOI = "MathOptInterface"
@@ -615,9 +613,9 @@ version = "0.8.8"
 
 [[deps.MadNLPGPU]]
 deps = ["AMD", "CUDA", "CUDSS", "KernelAbstractions", "LinearAlgebra", "MadNLP", "Metis", "SparseArrays"]
-git-tree-sha1 = "fb72ac1ee9ff608670315dd00514e9b423d1ff48"
+git-tree-sha1 = "760bc61aa70d536729ec6d08ed6799726b126f3b"
 uuid = "d72a61cc-809d-412f-99be-fd81f4b8a598"
-version = "0.7.10"
+version = "0.7.13"
 
 [[deps.Markdown]]
 deps = ["Base64"]
@@ -673,15 +671,15 @@ version = "0.4.3"
 
 [[deps.NVTX]]
 deps = ["Colors", "JuliaNVTXCallbacks_jll", "Libdl", "NVTX_jll"]
-git-tree-sha1 = "1a24c3430fa2ef3317c4c97fa7e431ef45793bd2"
+git-tree-sha1 = "6b573a3e66decc7fc747afd1edbf083ff78c813a"
 uuid = "5da4648a-3479-48b8-97b9-01cb529c0a1f"
-version = "1.0.0"
+version = "1.0.1"
 
 [[deps.NVTX_jll]]
 deps = ["Artifacts", "JLLWrappers", "Libdl"]
-git-tree-sha1 = "cd475b587ff77910789a18e68da789fc446a2a05"
+git-tree-sha1 = "af2232f69447494514c25742ba1503ec7e9877fe"
 uuid = "e98f9f5b-d649-5603-91fd-7774390e6439"
-version = "3.2.1+0"
+version = "3.2.2+0"
 
 [[deps.NetworkOptions]]
 uuid = "ca575930-c2e3-43a9-ace4-1e988b2c1908"
@@ -745,9 +743,9 @@ version = "1.2.1"
 
 [[deps.Preferences]]
 deps = ["TOML"]
-git-tree-sha1 = "9306f6085165d270f7e3db02af26a400d580f5c6"
+git-tree-sha1 = "0f27480397253da18fe2c12a4ba4eb9eb208bf3d"
 uuid = "21216c6a-2e73-6563-6e65-726566657250"
-version = "1.4.3"
+version = "1.5.0"
 
 [[deps.PrettyTables]]
 deps = ["Crayons", "LaTeXStrings", "Markdown", "PrecompileTools", "Printf", "Reexport", "StringManipulation", "Tables"]
@@ -833,9 +831,9 @@ version = "0.7.0"
 
 [[deps.ScopedValues]]
 deps = ["HashArrayMappedTries", "Logging"]
-git-tree-sha1 = "1147f140b4c8ddab224c94efa9569fc23d63ab44"
+git-tree-sha1 = "c3b2323466378a2ba15bea4b2f73b081e022f473"
 uuid = "7e506255-f358-4e82-b7e4-beb19740aa63"
-version = "1.3.0"
+version = "1.5.0"
 
 [[deps.Scratch]]
 deps = ["Dates"]
@@ -861,9 +859,9 @@ version = "0.3.8"
 
 [[deps.SortingAlgorithms]]
 deps = ["DataStructures"]
-git-tree-sha1 = "66e0a8e672a0bdfca2c3f5937efb8538b9ddc085"
+git-tree-sha1 = "64d974c2e6fdf07f8155b5b2ca2ffa9069b608d9"
 uuid = "a2af1166-a08f-5f64-846c-94a0d3cef48c"
-version = "1.2.1"
+version = "1.2.2"
 
 [[deps.SparseArrays]]
 deps = ["Libdl", "LinearAlgebra", "Random", "Serialization", "SuiteSparse_jll"]
@@ -884,9 +882,9 @@ version = "2.5.1"
 
 [[deps.StaticArrays]]
 deps = ["LinearAlgebra", "PrecompileTools", "Random", "StaticArraysCore"]
-git-tree-sha1 = "0feb6b9031bd5c51f9072393eb5ab3efd31bf9e4"
+git-tree-sha1 = "cbea8a6bd7bed51b1619658dec70035e07b8502f"
 uuid = "90137ffa-7385-5640-81b9-e52037218182"
-version = "1.9.13"
+version = "1.9.14"
 
     [deps.StaticArrays.extensions]
     StaticArraysChainRulesCoreExt = "ChainRulesCore"
@@ -919,9 +917,9 @@ version = "1.7.1"
 
 [[deps.StatsBase]]
 deps = ["AliasTables", "DataAPI", "DataStructures", "LinearAlgebra", "LogExpFunctions", "Missings", "Printf", "Random", "SortingAlgorithms", "SparseArrays", "Statistics", "StatsAPI"]
-git-tree-sha1 = "b81c5035922cc89c2d9523afc6c54be512411466"
+git-tree-sha1 = "2c962245732371acd51700dbb268af311bddd719"
 uuid = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
-version = "0.34.5"
+version = "0.34.6"
 
 [[deps.StatsFuns]]
 deps = ["HypergeometricFunctions", "IrrationalConstants", "LogExpFunctions", "Reexport", "Rmath", "SpecialFunctions"]

benchmarks/crystal.jl

@@ -42,14 +42,14 @@ function crystal(z3d; variant::Bool=false, gpu::Bool=false, gpu_arch::String="cu
         punched_pmn = copy(z3d)
         punched_pmn = punched_pmn[1:210, 1:310, 1:310]
 
-        index_missing_2D = CartesianIndex{3}[]
+        index_missing_3D = CartesianIndex{3}[]
         for i=0:4.
             for j=0:6.
                 for k = 0:6.
                     center =[i,j,k]
                     absolute_indices1 = punch_3D_cart(center, radius, x, y, z)
                     punched_pmn[absolute_indices1] .= 0
-                    append!(index_missing_2D, absolute_indices1)
+                    append!(index_missing_3D, absolute_indices1)
                 end
             end
         end
@@ -68,51 +68,39 @@ function crystal(z3d; variant::Bool=false, gpu::Bool=false, gpu_arch::String="cu
         punched_pmn = copy(z3d)
         punched_pmn = punched_pmn[1:310, 1:410, 1:410]
 
-        index_missing_2D = CartesianIndex{3}[]
+        index_missing_3D = CartesianIndex{3}[]
         for i=0:6.
             for j=0:8.
                 for k = 0:8.
                     center =[i,j,k]
                     absolute_indices1 = punch_3D_cart(center, radius, x, y, z)
                     punched_pmn[absolute_indices1] .= 0
-                    append!(index_missing_2D, absolute_indices1)
+                    append!(index_missing_3D, absolute_indices1)
                 end
             end
         end
     end
 
     DFTsize = size(punched_pmn)  # problem dim
     DFTdim = length(DFTsize)  # problem size
-    M_perptz = M_perp_tz_wei(DFTdim, DFTsize, punched_pmn)
-    S = Vector{Float64}
     if gpu
         if gpu_arch == "cuda"
-            M_perptz = CuArray(M_perptz)
-            S = CuVector{Float64}
+            AT = CuArray{Float64}
+            VT = CuVector{Float64}
         elseif gpu_arch == "rocm"
-            M_perptz = ROCArray(M_perptz)
-            S = ROCVector{Float64}
+            AT = ROCArray{Float64}
+            VT = ROCVector{Float64}
         else
             error("Unsupported GPU architecture \"$gpu_arch\".")
         end
+    else
+        AT = Array{Float64}
+        VT = Vector{Float64}
     end
-    Nt = prod(DFTsize)
 
     lambda = 1
-
-    alpha_LS = 0.1
-    gamma_LS = 0.8
-    eps_NT = 1e-6
-    eps_barrier = 1e-6
-    mu_barrier = 10
-
-    parameters = FFTParameters(DFTdim, DFTsize, M_perptz, lambda, index_missing_2D, alpha_LS, gamma_LS, eps_NT, mu_barrier, eps_barrier)
-
-    t_init = 1
-    beta_init = ones(Nt) ./ 2
-    c_init = ones(Nt)
-
-    nlp = FFTNLPModel{Float64, S}(parameters; rdft, preconditioner=true)
+    parameters = FFTParameters(DFTdim, DFTsize, punched_pmn |> AT, lambda, index_missing_3D)
+    nlp = FFTNLPModel{VT}(parameters; rdft, preconditioner=true)
 
     # Solve with MadNLP/CG
     t1 = time()