Add_minimalsurface

[indyalardjane]

No description provided.

[tmigot]

Thanks @indyalardjane for the PR!

I added a couple of comments.

In the functions, you don't want to convert variables with T as it will break the derivatives. We want Julia to infer the type.

Do you have bound constraints in your model as well? I.e. constraints of the form x >= lvar or x <= uvar? If so, you should prepare vector lvar and uvar and use the following constructor

ADNLPModel(f, x0, lvar, uvar, c, lcon, ucon)

[tmigot]

A couple of new comments.
Could you remove the two files minimalsurface.jl ? Also, can you compare your model with the Ampl version that looks simpler for some reason? Thanks!!

[tmigot]

Suggested change

	using Plots
	using ADNLPModels, NLPModels, NLPModelsIpopt, DataFrames, LinearAlgebra, Distances, SolverCore, PyPlot

[tmigot]

Suggested change

	return ADNLPModel(f, v, constraints, lcon, ucon)
	return ADNLPModels.ADNLPModel(f, v, constraints, lcon, ucon)

[tmigot]

Suggested change

	s = hx * hy * (sqrt(1 + (gi^2) +(gj)^2)) # Approximation of the derivative with the method of rectangles
	S = S + s # Update the value of S
	S = S + hx * hy * (sqrt(1 + (gi^2) +(gj)^2)) # Approximation of the derivative with the method of rectangles

[tmigot]

Is there a reason to choose 20? I suggest we do: nx = Int(round(sqrt(n))) and ny = nx

[dpo]

@indyalardjane It looks like there are still a few comments to address here, please.