Got "ERROR: LoadError: AssertionError: islinear" when using symbolic_linear_solve

[ysaereve]

Issue

I've run into a bit of a puzzle with symbolic_linear_solve, and I was hoping you could lend your expertise. I'm working with two sets of linear equations (included below), and while one solves without a hitch, the other is giving me some trouble. I've been trying to understand why using Symbolics.linear_expansion, and it seems that Test Case 1 is returning true, while Test Case 2 is returning false. I'm a bit stuck, so any advice or suggestions you might have would be fantastic!

Test Case 1: (Works)

eq = 8.85 - 8.85 * mp ~ (1+0.13) * er

Test Case 2: (Got "ERROR: LoadError: AssertionError: islinear" error)

eq = 8.85 ~ er / (1-mp) * (1+0.13)

Code

#!/usr/bin/env -S julia +release
using Symbolics
using Pkg

function printpackageversion(names::Vector{String})
    deps = Pkg.dependencies()
    for (uuid, dep) in deps
        if dep.name in names
            println("$(dep.name): v$(dep.version)")
        end
    end
end

println("Julia: v$VERSION")
printpackageversion(["Symbolics"])

# er: exchange rate
# mp: margin percent
@variables er mp

# Test Case #1
begin
    println("\n# Test Case #1")
    @show eq = 8.85 - 8.85 * mp ~ (1+0.13) * er 
    @show Symbolics.linear_expansion(eq, mp)

    @show n = symbolic_linear_solve(eq, er, simplify = true)
    @show substitute(n, Dict(mp => 0.05))
    @show n = symbolic_linear_solve(eq, mp, simplify = true)
    @show substitute(n, Dict(er => 7.75))
end

# Test Case #2
begin
    println("\n# Test Case #2")
    @show eq = 8.85 ~ er / (1-mp) * (1+0.13) 
    @show Symbolics.linear_expansion(eq, mp)

    @show n = symbolic_linear_solve(eq, er, simplify = true)
    @show substitute(n, Dict(mp => 0.05))
    @show n = symbolic_linear_solve(eq, mp, simplify = true)
    @show substitute(n, Dict(er => 7.75))
end

Error

Julia: v1.11.2
Symbolics: v6.22.1

# Test Case #1
eq = (8.85 - 8.85mp ~ (1 + 0.13) * er) = 8.85 - 8.85mp ~ 1.13er
Symbolics.linear_expansion(eq, mp) = (8.85, -8.85 + 1.13er, true)
n = symbolic_linear_solve(eq, er, simplify = true) = 0.8849557522123894(8.85 - 8.85mp)
substitute(n, Dict(mp => 0.05)) = 7.4402654867256635
n = symbolic_linear_solve(eq, mp, simplify = true) = 0.11299435028248588(8.85 - 1.13er)
substitute(n, Dict(er => 7.75)) = 0.010451977401130071

# Test Case #2
eq = (8.85 ~ (er / (1 - mp)) * (1 + 0.13)) = 8.85 ~ (1.13er) / (1 - mp)
Symbolics.linear_expansion(eq, mp) = (0, 0, false)
n = symbolic_linear_solve(eq, er, simplify = true) = 7.831858407079646(1 - mp)
substitute(n, Dict(mp => 0.05)) = 7.4402654867256635
ERROR: LoadError: AssertionError: islinear
Stacktrace:
  [1] symbolic_linear_solve(eq::Equation, var::Num; simplify::Bool, check::Bool)
    @ Symbolics ~/.julia/packages/Symbolics/PxO3a/src/linear_algebra.jl:105
  [2] macro expansion
    @ show.jl:1232 [inlined]
  [3] top-level scope
    @ ~/Projects/Babel/Julia/TDSymbolics/TDSymbolics-TDislinear.jl:41
  [4] include(fname::String)
    @ Main ./sysimg.jl:38
  [5] run(debug_session::VSCodeDebugger.DebugAdapter.DebugSession, error_handler::VSCodeDebugger.var"#3#4"{String})
    @ VSCodeDebugger.DebugAdapter ~/.vscode-insiders/extensions/julialang.language-julia-1.127.2/scripts/packages/DebugAdapter/src/packagedef.jl:122
  [6] startdebugger()
    @ VSCodeDebugger ~/.vscode-insiders/extensions/julialang.language-julia-1.127.2/scripts/packages/VSCodeDebugger/src/VSCodeDebugger.jl:45
  [7] top-level scope
    @ ~/.vscode-insiders/extensions/julialang.language-julia-1.127.2/scripts/debugger/run_debugger.jl:12
  [8] include(mod::Module, _path::String)
    @ Base ./Base.jl:557
  [9] exec_options(opts::Base.JLOptions)
    @ Base ./client.jl:323
 [10] _start()
    @ Base ./client.jl:531
in expression starting at ~/Projects/Babel/Julia/TDSymbolics/TDSymbolics-TDislinear.jl:34
 *  Terminal will be reused by tasks, press any key to close it. 

[ChrisRackauckas]

Case 2 isn't a linear expression. You can pre-process to multiply by 1-mp to make it one, but currently this solver function is not capable of doing that.