Callback function in IPOP for catching inner iterations

Specific Domains Optimization (Mathematical)
, ,
1

Hi, I am currently using a callback function to get the number of inner iterations.
Firstly, I have this: set_optimizer_attributes(model, “mu_target” => μ, “mu_init” => μ);
Where μ is updated outside of IPOPT ( I have my reasons for doing so) However, I am calling the callback function like this,

iters = []
function my_callback(
            alg_mod::Cint,
            iter_count::Cint,
            obj_value::Float64,
            inf_pr::Float64,
            inf_du::Float64,
            mu::Float64,
            d_norm::Float64,
            regularization_size::Float64,
            alpha_du::Float64,
            alpha_pr::Float64,
            ls_trials::Cint)

            append!(iters, iter_count)
            [...]

After IPOPT has successfully found the solution for my specific barrier parameter, I want to know the number of iterations IPOPT used. As I set the barrier parameter myself there will be no “outer” iterations. I believe that I should get the number of iterations by taking: j += last(iters). However, when I debug my way into the iters vector I see this,
iterations
where the purple color represent the element in the vector. I expected that length(iters) -1 = last(iters) (-1 because the iteration starts at 0 which is the initial value and therefore not counted as an iteration) Why is this not the case?

2

I’ve moved this question to the “Optimization (Mathematical)” section.

It’s hard to say anything without a reproducible example. We’re not doing anything special in Ipopt.jl, so the behavior you’re seeing is what is coming from the Ipopt algorithm.

x-ref Increasing the barrier parameter manuelly, IPOPT - #2 by odow