Incorrect return type classification in Optimization.jl

Specific Domains Optimization (Mathematical)
1

Hello,
with version 4.0.0 Optimization.jl started to classify the return value “FTOL_REACHED” from NLOpt as failure. This produces a warning. The following example reproduces the problem:

using Optimization, OptimizationNLopt

function exec_mwe()
    f(x, p) = sum(0.1 * (x .+ p) .^ 2)
    f_opt = OptimizationFunction(f)
    u0 = [0.5]

    prob = Optimization.OptimizationProblem(f_opt, u0, [1.0], lb=[-10.0], ub=[10.0])
    sol = solve(prob, NLopt.LN_BOBYQA(), abstol=1e-8)
    println(sol)
end

exec_mwe()
2

Open an issue?

3

Well I don’t have a github account. But I guess it’s good reason to create one. I will open an issue.