[Dev] Default options

[ocots]

function build_adnlp_model(
        initial_guess::CTModels.AbstractOptimalControlInitialGuess;
        backend,
        kwargs...
    )::ADNLPModels.ADNLPModel

    .....

        # build NLP
        nlp = ADNLPModel!(
            f,
            init,
            docp.bounds.var_l,
            docp.bounds.var_u,
            c!,
            docp.bounds.con_l,
            docp.bounds.con_u;
            minimize=(!docp.flags.max),
            backend_options...,
            unused_backends...,
            kwargs...,
        )

        return nlp
    end

    # NLP builder for ExaModels
    # +++ recheck kwargs passing / default with Olivier
    function build_exa_model(
        ::Type{BaseType}, 
        initial_guess::CTModels.AbstractOptimalControlInitialGuess; 
        backend
    )::ExaModels.ExaModel where {BaseType<:AbstractFloat}

        # recover discretization scheme and size
        scheme = Strategies.options(discretizer)[:scheme]
        grid_size = Strategies.options(discretizer)[:grid_size]

        # build initial guess
        init = get_docp_initial_guess(:exa, docp, initial_guess)

        # build Exa model and getters
        # +++ later try to call Exa constructor here if possible, reusing existing functions...
        build_exa = CTModels.get_build_examodel(ocp)
        nlp, exa_getter = build_exa(;
            grid_size=grid_size,
            backend=backend,
            scheme=scheme,
            init=init,
            base_type=BaseType,
        )

        return nlp
    end

[ocots]

And julia=1.10 in the Project.toml.

[PierreMartinon]

Cf #553

[PierreMartinon]

Minor question @ocots, @jbcaillau : if I understand correctly, the default options for the different strategies are independent, for instance the default ADNLP backend (:optimized) is set regardless of the discretizer (collocation) ?

Hypothetically, how would we handle the situation where another discretizer would have a different default value for the modeler ? Not sure the question will actually arise, just wondering. At worst I guess the builders in the discretizer could override the default values.

More generally, this would raise the question of the validity of certain combinations of options across the different strategies.

Edit: after discussion with @ocots, we can indeed override in the builder functions if needed. Moreover, the options system also provides the origin of a value (ie default or user), therefore we can choose to inform/warn/error in case of option substitution or incompatibility.

[jbcaillau]

@ocots ?

[PierreMartinon]

@jbcaillau see update above. I think we have what we need for now.

[jbcaillau]

@ocots please clarify why you reopened

[ocots]

Don't remember. We can close.