GalacticOptim with AmplNLWriter optimizer

Specific Domains Optimization (Mathematical)
1

Hello!

I’m trying to use GalacticOptim.jl with AmplNLWriter.jl optimizer but it’s not working.

using GalacticOptim, AmplNLWriter, Ipopt_jll

rosenbrock(x,p) = (p[1] - x[1])^2 + p[2] * (x[2] - x[1]^2)^2

x0 = zeros(2)

p = [1.0,100.0]

prob = OptimizationProblem(rosenbrock,x0,p)

optimizer = GalacticOptim.MOI.OptimizerWithAttributes(AmplNLWriter.Optimizer(Ipopt_jll.amplexe))

sol = solve(prob, optimizer)

println(sol)

Here I use Ipopt_jll just for example, but it’s not working even with it. Could you please help me to find an error?

Thank you

2

Can you open this as an issue on the repo?

3

Yes, sure

4

AmplNLWriter requires the expression graph, which galacticoptim probably doesn’t provide.

Why not just use Ipopt directly via ipopt.jl?

5

I used Ipopt here only as an example, to check whether it will work. But I’d like to use AmplNLWriter later with another local solver on my laptop.

1 Like
6

Use JuMP instead of galacticoptim

7

Thank you for this recommendation. JuMP is very good but I thought that GalacticOptim and ModelingToolkit will be easier to use with nonlinearities.

8

JuMP supports nonlinear:
https://jump.dev/JuMP.jl/stable/manual/nlp/

9

We can just do it through MTK though

10

I remember MTK can find minima and maxima, but how is this combination for non-linear optimisation in general? I’m going to be taking an econ class, with a lot of optimisation of Cartesian planes.

11

It might be best to start a new post for this. It’s unrelated to the ExprGraph problem.

1 Like