Storing value of loss

[swilliamson7]

Hello! Small idea for a new feature, but it could be nice to add an option to store the value of the objective every n iterations during the optimization. This is something I tend to look at later for figures to discuss the performance of the optimization over time. Maybe it wouldn't be too much work to add?

[sshin23]

Thanks @swilliamson7 for the suggestion.

One possibility is to take a user callback function and call user_callback(solver) somewhere, so that user can do something like

function my_callback(solver)
    push!(result, solver.obj)
end

Any idea @amontoison @frapac?

[frapac]

I am in to implement user callback in MadNLP, I think it would be a great feature.

Another option is to implement a NLPModels that stores the intermediate values.