Enzyme + SparseArrays + LinearSolve = Not Happy

[AbdAlazezAhmed]

Describe the bug 🐞

When differentiating a solve with A being sparse, the structure of A changes (rowvals changes size and values, colptr changes values)

Expected behavior

The sparsity pattern (or A in general) should not be mutated iiuc

Minimal Reproducible Example 👇

using SparseArrays, Enzyme, LinearSolve, LinearAlgebra

function do_sth(_K, some_scalar, ret)
    # nonzeros(_K) .+= some_scalar # We're not even doing anything here
    f = rand(4)
    m  = _K
    prob = LinearProblem(m, f)
    linsolve = init(prob, KrylovJL_CG())
    sol = solve!(linsolve)
    ret[1] = sol|>sum
    return nothing
end


K = sprand(4, 4, 0.3) + SparseArrays.I
K = make_zero(K) .+ K
display(K) # to see if it changes later
@show K.rowval 
@show K.nzval
@show K.m
@show K.n
@show K.colptr
y = [0.0]
# Enzyme.API.printall!(true)
Enzyme.autodiff(Reverse |> set_runtime_activity, do_sth , Duplicated(K, make_zero(K)), Active(1.0), Duplicated(y, [1.0]))
display(K[1:end, 1:end]) # This looks like as if it concatinated the nonzeros and added grabage afterwards
@show K.rowval 
@show K.nzval
@show K.m
@show K.n
@show K.colptr
display(K) # Warum?

Error & Stacktrace ⚠️
It's not about the error tho

4×4 SparseMatrixCSC{Float64, Int64} with 7 stored entries:
 1.0       0.196913   ⋅    ⋅ 
  ⋅        1.0        ⋅    ⋅ 
  ⋅         ⋅        1.0  0.0132919
 0.715638   ⋅         ⋅   1.75787
K.rowval = [1, 4, 1, 2, 3, 3, 4]
K.nzval = [1.0, 0.715638225985187, 0.19691255689136355, 1.0, 1.0, 0.013291905495585454, 1.7578727660564724]
K.m = 4
K.n = 4
K.colptr = [1, 3, 5, 6, 8]
┌ Warning: Verbosity toggle: max_iters 
│  Solver reached maximum number of iterations
└ @ LinearSolve ~/.julia/dev/LinearSolve/src/iterative_wrappers.jl:328
┌ Warning: Verbosity toggle: max_iters 
│  Solver reached maximum number of iterations
└ @ LinearSolve ~/.julia/dev/LinearSolve/src/iterative_wrappers.jl:328
4×4 SparseMatrixCSC{Float64, Int64} with 16 stored entries:
 1.0       1.0           6.92923e-310  5.0e-324
 0.715638  0.0132919     6.92923e-310  6.92923e-310
 0.196913  1.75787       3.546e-320    5.0e-324
 1.0       6.92923e-310  1.3251e-319   6.92888e-310
K.rowval = [1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4]
K.nzval = [1.0, 0.715638225985187, 0.19691255689136355, 1.0, 1.0, 0.013291905495585454, 1.7578727660564724]
K.m = 4
K.n = 4
K.colptr = [1, 5, 9, 13, 17]
ERROR: AssertionError: _goodbuffers(S)
Stacktrace:
  [1] _checkbuffers
    @ ~/.julia/juliaup/julia-1.11.5+0.x64.linux.gnu/share/julia/stdlib/v1.11/SparseArrays/src/sparsematrix.jl:168 [inlined]
  [2] array_summary
    @ ~/.julia/juliaup/julia-1.11.5+0.x64.linux.gnu/share/julia/stdlib/v1.11/SparseArrays/src/sparsematrix.jl:339 [inlined]
  [3] summary
    @ ./show.jl:3152 [inlined]
  [4] show(io::IOContext{Base.TTY}, ::MIME{Symbol("text/plain")}, X::SparseMatrixCSC{Float64, Int64})
    @ Base ./arrayshow.jl:368
  [5] (::REPL.var"#68#69"{REPL.REPLDisplay{REPL.LineEditREPL}, MIME{Symbol("text/plain")}, Base.RefValue{Any}})(io::Any)
    @ REPL ~/.julia/juliaup/julia-1.11.5+0.x64.linux.gnu/share/julia/stdlib/v1.11/REPL/src/REPL.jl:394
  [6] with_repl_linfo(f::Any, repl::REPL.LineEditREPL)
    @ REPL ~/.julia/juliaup/julia-1.11.5+0.x64.linux.gnu/share/julia/stdlib/v1.11/REPL/src/REPL.jl:678
  [7] display(d::REPL.REPLDisplay, mime::MIME{Symbol("text/plain")}, x::Any)
    @ REPL ~/.julia/juliaup/julia-1.11.5+0.x64.linux.gnu/share/julia/stdlib/v1.11/REPL/src/REPL.jl:380
  [8] display(d::REPL.REPLDisplay, x::Any)
    @ REPL ~/.julia/juliaup/julia-1.11.5+0.x64.linux.gnu/share/julia/stdlib/v1.11/REPL/src/REPL.jl:399
  [9] display(x::Any)
    @ Base.Multimedia ./multimedia.jl:340
 [10] top-level scope
    @ ~/Vorlagen/mwe.jl:38

The error is related to displaying broken CSC arrays

Environment (please complete the following information):

• Output of using Pkg; Pkg.status()

⌃ [a0c0ee7d] DifferentiationInterface v0.7.4
  [7da242da] Enzyme v0.13.108 `https://github.com/EnzymeAD/Enzyme.jl.git#main`
  [c061ca5d] Ferrite v1.2.0 `https://github.com/Ferrite-FEM/Ferrite.jl.git#master`
  [7ed4a6bd] LinearSolve v3.48.0 `~/.julia/dev/LinearSolve`
⌅ [da2b9cff] Mooncake v0.4.139
  [2f01184e] SparseArrays v1.11.0

• Output of using Pkg; Pkg.status(; mode = PKGMODE_MANIFEST)

  [47edcb42] ADTypes v1.20.0
  [7d9f7c33] Accessors v0.1.42
  [79e6a3ab] Adapt v4.4.0
  [ec485272] ArnoldiMethod v0.4.0
  [4fba245c] ArrayInterface v7.22.0
  [4c555306] ArrayLayouts v1.12.2
  [fa961155] CEnum v0.5.0
  [082447d4] ChainRules v1.72.6
  [d360d2e6] ChainRulesCore v1.26.0
  [944b1d66] CodecZlib v0.7.8
  [38540f10] CommonSolve v0.2.4
  [bbf7d656] CommonSubexpressions v0.3.1
  [34da2185] Compat v4.18.1
  [a33af91c] CompositionsBase v0.1.2
  [2569d6c7] ConcreteStructs v0.2.3
  [187b0558] ConstructionBase v1.6.0
  [9a962f9c] DataAPI v1.16.0
  [864edb3b] DataStructures v0.19.3
  [e2d170a0] DataValueInterfaces v1.0.0
  [163ba53b] DiffResults v1.1.0
  [b552c78f] DiffRules v1.15.1
⌃ [a0c0ee7d] DifferentiationInterface v0.7.4
  [8d63f2c5] DispatchDoctor v0.4.26
  [b4f34e82] Distances v0.10.12
  [ffbed154] DocStringExtensions v0.9.5
  [4e289a0a] EnumX v1.0.5
  [7da242da] Enzyme v0.13.108 `https://github.com/EnzymeAD/Enzyme.jl.git#main`
  [f151be2c] EnzymeCore v0.8.17
  [e2ba6199] ExprTools v0.1.10
  [55351af7] ExproniconLite v0.10.14
  [c061ca5d] Ferrite v1.2.0 `https://github.com/Ferrite-FEM/Ferrite.jl.git#master`
  [1a297f60] FillArrays v1.15.0
  [f6369f11] ForwardDiff v1.3.0
  [069b7b12] FunctionWrappers v1.1.3
  [77dc65aa] FunctionWrappersWrappers v0.1.3
  [46192b85] GPUArraysCore v0.2.0
  [61eb1bfa] GPUCompiler v1.7.5
  [86223c79] Graphs v1.13.1
  [d25df0c9] Inflate v0.1.5
  [3587e190] InverseFunctions v0.1.17
  [92d709cd] IrrationalConstants v0.2.6
  [82899510] IteratorInterfaceExtensions v1.0.0
  [692b3bcd] JLLWrappers v1.7.1
  [ae98c720] Jieko v0.2.1
  [ba0b0d4f] Krylov v0.10.3
  [929cbde3] LLVM v9.4.4
  [5078a376] LazyArrays v2.9.3
  [9c8b4983] LightXML v0.9.3
  [7ed4a6bd] LinearSolve v3.48.0 `~/.julia/dev/LinearSolve`
  [2ab3a3ac] LogExpFunctions v0.3.29
  [e6f89c97] LoggingExtras v1.2.0
  [1914dd2f] MacroTools v0.5.16
  [dbe65cb8] MistyClosures v2.1.0
⌅ [da2b9cff] Mooncake v0.4.139
  [2e0e35c7] Moshi v0.3.7
  [77ba4419] NaNMath v1.1.3
  [b8a86587] NearestNeighbors v0.4.24
  [d8793406] ObjectFile v0.5.0
  [bac558e1] OrderedCollections v1.8.1
  [d236fae5] PreallocationTools v0.4.34
⌅ [aea7be01] PrecompileTools v1.2.1
  [21216c6a] Preferences v1.5.0
  [c1ae055f] RealDot v0.1.0
  [3cdcf5f2] RecipesBase v1.3.4
  [731186ca] RecursiveArrayTools v3.39.0
  [189a3867] Reexport v1.2.2
  [ae029012] Requires v1.3.1
  [7e49a35a] RuntimeGeneratedFunctions v0.5.16
  [fdea26ae] SIMD v3.7.2
  [0bca4576] SciMLBase v2.127.0
  [a6db7da4] SciMLLogging v1.6.0
  [c0aeaf25] SciMLOperators v1.13.0
  [431bcebd] SciMLPublic v1.0.0
  [53ae85a6] SciMLStructures v1.7.0
  [6c6a2e73] Scratch v1.3.0
  [efcf1570] Setfield v1.1.2
  [699a6c99] SimpleTraits v0.9.5
  [dc90abb0] SparseInverseSubset v0.1.2
  [276daf66] SpecialFunctions v2.6.1
  [90137ffa] StaticArrays v1.9.15
  [1e83bf80] StaticArraysCore v1.4.4
  [10745b16] Statistics v1.11.1
  [82ae8749] StatsAPI v1.7.1
  [09ab397b] StructArrays v0.7.2
  [53d494c1] StructIO v0.3.1
  [2efcf032] SymbolicIndexingInterface v0.3.46
  [3783bdb8] TableTraits v1.0.1
  [bd369af6] Tables v1.12.1
  [48a634ad] Tensors v1.16.2
  [e689c965] Tracy v0.1.6
  [3bb67fe8] TranscodingStreams v0.11.3
  [3a884ed6] UnPack v1.0.2
  [4004b06d] VTKBase v1.0.1
  [64499a7a] WriteVTK v1.21.2
⌅ [7cc45869] Enzyme_jll v0.0.221+0
  [1d5cc7b8] IntelOpenMP_jll v2025.2.0+0
  [dad2f222] LLVMExtra_jll v0.0.38+0
  [ad6e5548] LibTracyClient_jll v0.9.1+6
  [94ce4f54] Libiconv_jll v1.18.0+0
  [856f044c] MKL_jll v2025.2.0+0
  [efe28fd5] OpenSpecFun_jll v0.5.6+0
  [02c8fc9c] XML2_jll v2.15.1+0
  [1317d2d5] oneTBB_jll v2022.0.0+1
  [0dad84c5] ArgTools v1.1.2
  [56f22d72] Artifacts v1.11.0
  [2a0f44e3] Base64 v1.11.0
  [ade2ca70] Dates v1.11.0
  [8ba89e20] Distributed v1.11.0
  [f43a241f] Downloads v1.6.0
  [7b1f6079] FileWatching v1.11.0
  [9fa8497b] Future v1.11.0
  [b77e0a4c] InteractiveUtils v1.11.0
  [4af54fe1] LazyArtifacts v1.11.0
  [b27032c2] LibCURL v0.6.4
  [76f85450] LibGit2 v1.11.0
  [8f399da3] Libdl v1.11.0
  [37e2e46d] LinearAlgebra v1.11.0
  [56ddb016] Logging v1.11.0
  [d6f4376e] Markdown v1.11.0
  [a63ad114] Mmap v1.11.0
  [ca575930] NetworkOptions v1.2.0
  [44cfe95a] Pkg v1.11.0
  [de0858da] Printf v1.11.0
  [9a3f8284] Random v1.11.0
  [ea8e919c] SHA v0.7.0
  [9e88b42a] Serialization v1.11.0
  [1a1011a3] SharedArrays v1.11.0
  [6462fe0b] Sockets v1.11.0
  [2f01184e] SparseArrays v1.11.0
  [4607b0f0] SuiteSparse
  [fa267f1f] TOML v1.0.3
  [a4e569a6] Tar v1.10.0
  [8dfed614] Test v1.11.0
  [cf7118a7] UUIDs v1.11.0
  [4ec0a83e] Unicode v1.11.0
  [e66e0078] CompilerSupportLibraries_jll v1.1.1+0
  [deac9b47] LibCURL_jll v8.6.0+0
  [e37daf67] LibGit2_jll v1.7.2+0
  [29816b5a] LibSSH2_jll v1.11.0+1
  [c8ffd9c3] MbedTLS_jll v2.28.6+0
  [14a3606d] MozillaCACerts_jll v2023.12.12
  [4536629a] OpenBLAS_jll v0.3.27+1
  [05823500] OpenLibm_jll v0.8.5+0
  [bea87d4a] SuiteSparse_jll v7.7.0+0
  [83775a58] Zlib_jll v1.2.13+1
  [8e850b90] libblastrampoline_jll v5.11.0+0
  [8e850ede] nghttp2_jll v1.59.0+0
  [3f19e933] p7zip_jll v17.4.0+2

• Output of versioninfo()

Julia Version 1.11.5
Commit 760b2e5b739 (2025-04-14 06:53 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 20 × 13th Gen Intel(R) Core(TM) i5-13500
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, goldmont)
Threads: 20 default, 0 interactive, 10 GC (on 20 virtual cores)
Environment:
  JULIA_LOAD_PATH = /usr/share/gmsh/api/julia:
  JULIA_EDITOR = code
  JULIA_VSCODE_REPL = 1

Additional context

AD not happy

[ChrisRackauckas]

Does the sparsity pattern only change with Enzyme, or can this be reduced?