Trivial nonlinear system not simplifying to zero

[ChrisRackauckas]

MWE:

using ModelingToolkit, OrdinaryDiffEq
using ModelingToolkit: t_nounits as t
@variables x(t) y(t)
eqs = [
    0 ~ x - y
    0 ~ 2y - x
]

@named sys = System(eqs, t)
sys = mtkcompile(sys)

equations(sys)
# 1-element Vector{Equation}:
# 0 ~ x(t) - y(t)

prob = ODEProblem(sys, Pair[], (0, 10.0))
sol = solve(prob, Tsit5()) # ERROR: This solver is not able to use mass matrices. For compatible solvers see https://docs.sciml.ai/DiffEqDocs/stable/solvers/dae_solve/

[AayushSabharwal]

That is going to have an algebraic equation, though. There's no other way to break the loop

[ChrisRackauckas]

Bareiss doesn't solve this? It's a linear system.

[AayushSabharwal]

Alias elimination only considers equations where all coefficients are +-1

[AayushSabharwal]

It's been that way for ~4 years now

[oscardssmith]

We really should stop doing that. We should either allow arbitrary scalings or we should delete bareiss and use something dramatically simpler to find aliases.

[AayushSabharwal]

There's theoretically nothing wrong with arbitrary scalings.

[AayushSabharwal]

Fixing that should be a small PR