Skip to content

Fix CI docs build #822

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Draft
wants to merge 6 commits into
base: master
Choose a base branch
from
Draft

Fix CI docs build #822

wants to merge 6 commits into from
+15 −10

Conversation

@abhro
Copy link
Contributor

@abhro abhro commented Nov 20, 2025

Checklist

  • Appropriate tests were added
  • Any code changes were done in a way that does not break public API
  • All documentation related to code changes were updated
  • The new code follows the contributor guidelines, in particular the SciML Style Guide and COLPRAC.
  • Any new documentation only uses public API

Additional context

Fixes errors in buildkite CI docs build.

Currently in draft mode: the example in tutorials/bvp_example.md (Ascher method for solving BVDAEs) fails on the newest LinearSolve.jl (3.47.0), but if pinned to LinearSolve.jl v3.40.0, it does produce results. Not sure what the underlying issues are yet.
@abhro
Copy link
Contributor Author

abhro commented Nov 20, 2025

The BVP Ascher thing fails on LinearSolve v3.46.0, but works with 3.45.0. MWE script:

using Pkg
Pkg.activate(temp=true)
ENV["JULIA_PKG_PRECOMPILE_AUTO"] = false
Pkg.add(name="LinearSolve", version=ARGS[1])
Pkg.add("BoundaryValueDiffEq")

import BoundaryValueDiffEq as BVP
Pkg.status()
function f!(du, u, p, t)
    e = 2.7
    du[1] = (1 + u[2] - sin(t)) * u[4] + cos(t)
    du[2] = cos(t)
    du[3] = u[4]
    du[4] = (u[1] - sin(t)) * (u[4] - e^t)
end
function bc!(res, u, p, t)
    res[1] = u[1]
    res[2] = u[3] - 1
    res[3] = u[2] - sin(1.0)
end
u0 = [0.0, 0.0, 0.0, 0.0]
tspan = (0.0, 1.0)
fun = BVP.BVPFunction(f!, bc!, mass_matrix = [1 0 0 0; 0 1 0 0; 0 0 1 0; 0 0 0 0])
prob = BVP.BVProblem(fun, u0, tspan)
sol = BVP.solve(prob,
    BVP.Ascher4(; zeta = [0.0, 0.0, 1.0], jac_alg = BVP.BVPJacobianAlgorithm(BVP.AutoForwardDiff()));
    dt = 0.01)

run as: julia script.jl X.Y.Z where X.Y.Z is the LinearSolve.jl version to use
@abhro
Copy link
Contributor Author

abhro commented Nov 20, 2025

The problems from DiffEqProblemLibrary.jl appear to be the older ones. I think a new version is needed for (ODEProblemLibrary, SDEProblemLibrary, DDEProblemLibrary, DAEProblemLibrary, BVProblemLibrary)
@ChrisRackauckas
Copy link
Member

ChrisRackauckas commented Nov 20, 2025

What's the error? We should make the docs build fail on error
@abhro
Copy link
Contributor Author

abhro commented Nov 21, 2025 •
edited
Loading

Running with 3.46 gives

$ julia mwe.jl 3.46.0
  Activating new project at `/tmp/jl_eW8Fpj`
   Resolving package versions...
   Installed LinearSolve ─ v3.46.0
    Updating `/tmp/jl_eW8Fpj/Project.toml`
⌃ [7ed4a6bd] + LinearSolve v3.46.0
    Updating `/tmp/jl_eW8Fpj/Manifest.toml`
[... Pkg output ...]
   Resolving package versions...
    Updating `/tmp/jl_eW8Fpj/Project.toml`
  [764a87c0] + BoundaryValueDiffEq v5.18.0
    Updating `/tmp/jl_eW8Fpj/Manifest.toml`
[... Pkg output ...]
Precompiling BoundaryValueDiffEq finished.
  15 dependencies successfully precompiled in 425 seconds. 198 already precompiled.
Status `/tmp/jl_eW8Fpj/Project.toml`
  [764a87c0] BoundaryValueDiffEq v5.18.0
⌃ [7ed4a6bd] LinearSolve v3.46.0
Info Packages marked with ⌃ have new versions available and may be upgradable.
┌ Error: BLAS/LAPACK dgetrf: Matrix is singular
│   Details: U(1,1) is exactly zero. The factorization has been completed, but U is singular and division by U will produce infinity.
│   Return code (info): 1
│   matrix_size: (303, 303)
│   rhs_type: Vector{Float64}
│   rhs_size: (303,)
│   memory_usage_MB: 0.7
│   matrix_type: Matrix{Float64}
│   element_type: Float64
└ @ LinearSolve ~/.julia/packages/LinearSolve/V1Dfx/src/mkl.jl:267
ERROR: LoadError: BLAS/LAPACK dgetrf: Matrix is singular
  Details: U(1,1) is exactly zero. The factorization has been completed, but U is singular and division by U will produce infinity.
  Return code (info): 1
  matrix_size: (303, 303)
  rhs_type: Vector{Float64}
  rhs_size: (303,)
  memory_usage_MB: 0.7
  matrix_type: Matrix{Float64}
  element_type: Float64
Stacktrace:
  [1] emit_message(message::String, level::Base.CoreLogging.LogLevel, file::String, line::Int64, _module::Module)
    @ SciMLLogging ~/.julia/packages/SciMLLogging/3uoc4/src/utils.jl:57
  [2] macro expansion
    @ ~/.julia/packages/SciMLLogging/3uoc4/src/utils.jl:147 [inlined]
  [3] solve!(cache::LinearSolve.LinearCache{Matrix{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, LinearSolve.DefaultLinearSolver, [types elided]...)
    @ LinearSolve ~/.julia/packages/LinearSolve/V1Dfx/src/mkl.jl:267
  [4] solve!
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/mkl.jl:239 [inlined]
  [5] macro expansion
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/default.jl:514 [inlined]
  [6] solve!(::LinearSolve.LinearCache{Matrix{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, LinearSolve.DefaultLinearSolver, [types elided]...)
    @ LinearSolve ~/.julia/packages/LinearSolve/V1Dfx/src/default.jl:503
  [7] solve!
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/default.jl:503 [inlined]
  [8] #solve!#21
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/common.jl:441 [inlined]
  [9] solve!
    @ ~/.julia/packages/LinearSolve/V1Dfx/src/common.jl:440 [inlined]
 [10] #_#1
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/ext/NonlinearSolveBaseLinearSolveExt.jl:24 [inlined]
 [11] LinearSolveJLCache
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/ext/NonlinearSolveBaseLinearSolveExt.jl:14 [inlined]
 [12] solve!(cache::NonlinearSolveBase.NewtonDescentCache{Vector{Float64}, Nothing, [types elided]...)
    @ NonlinearSolveBase ~/.julia/packages/NonlinearSolveBase/mJsVc/src/descent/newton.jl:119
 [13] solve! (repeats 2 times)
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/descent/newton.jl:94 [inlined]
 [14] step!(cache::NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithmCache{Vector{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, [types elided]...)
    @ NonlinearSolveFirstOrder ~/.julia/packages/NonlinearSolveFirstOrder/inFag/src/solve.jl:266
 [15] step!
    @ ~/.julia/packages/NonlinearSolveFirstOrder/inFag/src/solve.jl:244 [inlined]
 [16] #step!#180
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:559 [inlined]
 [17] step!
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:553 [inlined]
 [18] solve!(cache::NonlinearSolveFirstOrder.GeneralizedFirstOrderAlgorithmCache{Vector{Float64}, Vector{Float64}, Vector{Float64}, SciMLBase.NullParameters, [types elided]...)
    @ NonlinearSolveBase ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:287
 [19] #__solve#155
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:273 [inlined]
 [20] __solve
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:270 [inlined]
 [21] macro expansion
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:478 [inlined]
 [22] __generated_polysolve(::SciMLBase.NonlinearProblem{Vector{Float64}, true, SciMLBase.NullParameters, [types elided]...)
    @ NonlinearSolveBase ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:423
 [23] __generated_polysolve
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:423 [inlined]
 [24] #__solve#166
    @ ~/.julia/packages/NonlinearSolveBase/mJsVc/src/solve.jl:396 [inlined]
 [25] __perform_ascher_iteration(cache::BoundaryValueDiffEqAscher.AscherCache{true, Float64, SciMLBase.BVProblem{Vector{Float64}, Tuple{Float64, Float64}, [types elided]...)
    @ BoundaryValueDiffEqAscher ~/.julia/packages/BoundaryValueDiffEqAscher/CPeel/src/ascher.jl:162
 [27] #__solve#34
    @ ~/.julia/packages/BoundaryValueDiffEqCore/hFduL/src/BoundaryValueDiffEqCore.jl:38 [inlined]
 [28] __solve
    @ ~/.julia/packages/BoundaryValueDiffEqCore/hFduL/src/BoundaryValueDiffEqCore.jl:35 [inlined]
 [29] #solve_call#23
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:127 [inlined]
 [30] solve_call
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:84 [inlined]
 [31] solve_up(prob::SciMLBase.BVProblem{Vector{Float64}, Tuple{Float64, Float64}, true, Nothing, SciMLBase.NullParameters, SciMLBase.BVPFunction{true, [types elided]...)
    @ DiffEqBase ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:563
 [32] solve_up
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:540 [inlined]
 [33] #solve#29
    @ ~/.julia/packages/DiffEqBase/aB45d/src/solve.jl:530 [inlined]
 [34] top-level scope
    @ ~/mwe.jl:25
 [35] include(mod::Module, _path::String)
    @ Base ./Base.jl:306
 [36] exec_options(opts::Base.JLOptions)
    @ Base ./client.jl:317
 [37] _start()
    @ Base ./client.jl:550
in expression starting at mwe.jl:25
@abhro
Copy link
Contributor Author

abhro commented Nov 24, 2025

Created SciML/LinearSolve.jl#830 for the BVP example failing
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

2 participants

@abhro @ChrisRackauckas