Steady State Heat without Time Dependence is Broken

[avik-pal]

https://docs.sciml.ai/MethodOfLines/dev/tutorials/heatss/ seems to be broken with the latest versions of the packages.

ERROR: ArgumentError: Equations (12), states (12), and initial conditions (121) are of different lengths. To allow a different number of equations than states use kwarg check_length=false.
Stacktrace:
  [1] check_eqs_u0(eqs::Vector{Equation}, dvs::Vector{Any}, u0::Vector{Float64}; check_length::Bool, kwargs::@Kwargs{})
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/abstractsystem.jl:1745
  [2] process_NonlinearProblem(constructor::Type, sys::NonlinearSystem, u0map::Vector{…}, parammap::SciMLBase.NullParameters; version::Nothing, jac::Bool, checkbounds::Bool, sparse::Bool, simplify::Bool, linenumbers::Bool, parallel::Symbolics.SerialForm, eval_expression::Bool, use_union::Bool, tofloat::Bool, kwargs::@Kwargs{…})
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/nonlinear/nonlinearsystem.jl:338
  [3] process_NonlinearProblem
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/nonlinear/nonlinearsystem.jl:322 [inlined]
  [4] (NonlinearProblem{…})(sys::NonlinearSystem, u0map::Vector{…}, parammap::SciMLBase.NullParameters; check_length::Bool, kwargs::@Kwargs{})
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/nonlinear/nonlinearsystem.jl:367
  [5] NonlinearProblem (repeats 2 times)
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/nonlinear/nonlinearsystem.jl:364 [inlined]
  [6] #NonlinearProblem#740
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/nonlinear/nonlinearsystem.jl:361 [inlined]
  [7] NonlinearProblem(sys::NonlinearSystem, args::Vector{Float64})
    @ ModelingToolkit /mnt/julia/packages/ModelingToolkit/oIgbi/src/systems/nonlinear/nonlinearsystem.jl:360
  [8] discretize(pdesys::PDESystem, discretization::MOLFiniteDifference{…}; analytic::Nothing, kwargs::@Kwargs{})
    @ PDEBase /mnt/julia/packages/PDEBase/aRCcl/src/discretization_state.jl:63
  [9] discretize
    @ Main /mnt/julia/packages/PDEBase/aRCcl/src/discretization_state.jl:55 [inlined]
 [10] construct_heatss_nlproblem(δ::Float64)
    @ Main /mnt/research/benchmarks/SciMLBenchmarks.jl/benchmarks/NonlinearProblem/steady_state_heateq.jmd:48
 [11] top-level scope
    @ /mnt/research/benchmarks/SciMLBenchmarks.jl/benchmarks/NonlinearProblem/steady_state_heateq.jmd:1
Some type information was truncated. Use `show(err)` to see complete types.

The last working versions we could figure out were:

  [5b8099bc] DomainSets v0.6.7
  [23fbe1c1] Latexify v0.16.1
⌃ [94925ecb] MethodOfLines v0.10.0
⌃ [961ee093] ModelingToolkit v8.64.0
⌃ [a7812802] PDEBase v0.1.4
⌅ [0bca4576] SciMLBase v1.98.1
  [9672c7b4] SteadyStateDiffEq v1.16.1

cc @ChrisRackauckas

[YingboMa]

From the error message, it's probably trying to construct the u0 from the unsimplified system by mistake.

julia> s = symbolic_discretize(pdesys, discretization)[1];
JuliaSimCompiler: generate_system

julia> ss = structural_simplify(s);

julia> NonlinearProblem(structural_simplify(s), rand(length(states(ss))))
NonlinearProblem with uType Vector{Float64}. In-place: true
u0: 12-element Vector{Float64}:
 0.5658693983716057
 0.730452210930952
 0.7999093647009754
 0.34068068695778175
 0.6066626175654133
 0.49832174589693123
 0.985735570069928
 0.10313229637404808
 0.5761788103850009
 0.019531876564822248
 0.4600618070405691
 0.5048377501152175

julia> length(states(s))
121

[xtalax]

Afraid that this isn't the case:

        simpsys = structural_simplify(sys)
        if tspan === nothing
            add_metadata!(get_metadata(sys), sys)
            return prob = NonlinearProblem(simpsys, ones(length(simpsys.states));
                                           discretization.kwargs..., kwargs...)

[avik-pal]

Digging a bit further 8.71.2 for Modeling Toolkit works

[ChrisRackauckas]

I thought you said you were on latest?

[avik-pal]

I meant 8.71.2 works, but the latest 8.72 doesn't