Accessing parameter values in a finalize_solution function

[rseydam]

Hi,

I am trying to use a finalize_solution inside MPC.continuation(...) .

I assume that whenever the function put in for finalize_solution returns true, it is accepted and the solution will be calculated; if it returns false, it is rejected.

In my function, I want to check several things. First, I am checking if all variables are positive. Secondly, I want to constrain the calculation to certain parameter ranges.

function boundary_finalizer(u, p)
    # Check dynamical variables > 0
    if any(x -> x <= 0, u)
        return false
    end

    # Read parameters from p
    γ = getproperty(p, :γ)
    n = getproperty(p, :n)
    ε_glu = getproperty(p, :ε_glu)

	@info "Called finalizer (accept) γ=$(γ), n=$(n), ε_glu=$(ε_glu)"

    # Check parameter ranges
    if !(0.01 <= γ <= 3.5) || !(0.01 <= n <= 1.0) || !(1e-7 <= ε_glu <= 0.2)
        return false
    end

    return true  # all conditions met
end

When I access the parameters in this manner, they remain constant during the calculation of all charts. How can I access the actual parameters on the current chart?

[rveltz]

Hi,

It is not supposed to change during continuation, which MPC.continuation are you talking about?

[rseydam]

Hey,

I am using it to calculate an atlas_hopf like in the example ABC problem.

In the meantime, I had another look at the example and saw that there is a solution to my question already there. Like in this part of the ABC problem, I am recording the parameters using the appropriate record_from_solution and then checking those in the finalize_solution function. In this way, one can create all kinds of useful finalizers. Great example!

prob = MPC.ManifoldProblem_BK(
                        prob_bk, br.sol[1].x, (@optic _.D), (@optic _.β);
                        record_from_solution = (X, p; k...) -> begin
                            return (β = X[end], D = X[end-1], u3 = X[3])
                        end,
                        finalize_solution = (X,p) -> begin
                            D = X[end-1]
                            β = X[end]
                            keep = (0.01 <= D <= 0.5) && (1.5 <= β <= 1.65)
                            return keep
                        end,
                        )

Thank you! I think I will close this issue.

[rveltz]

I recently add the surface plot for the Hopf surface instead of scatter plot, look a the tutorials.