# How to obtain a contour plot from given function?

1 view (last 30 days)
Utku Senel on 2 Nov 2020
Commented: Mario Malic on 3 Nov 2020
function cost = obj_function(x)
cost =(x(1)).^2+(x(2)-3).^2;
end
%% the lines after comments are new files but in the same folder(the below is the constraints)
function [c, ceq] = nonlcon(x)
ceq = (x(1).^2)+(3*x(2))-3;
c = x(1).^2.*4 + x(2).^2.*9 - 36;
%%below is the rest of it
LB = [];
UB = [];
X0 = [-1, 1000000];
A = [];
B = [];
Aeq = [];
Beq = [];
[x, fval] = fmincon(@obj_function, X0, A, B, Aeq, Beq, UB, LB, @nonlcon)
My entire function is given above. It maybe confusing to understand but how can i implement my function into contour function?

Mario Malic on 2 Nov 2020
Edited: Mario Malic on 2 Nov 2020
You need to obtain all x values and their corresponding objective function values. Reference to the answer that the idea is taken from.
function cost = obj_function(x)
peristent X_Vals F_Vals
if ~nargin
cost.X_Vals = X_Vals;
cost.F_Vals = F_Vals;
return
end
% Initialising values
X_Vals(1,:) = [0 0];
F_Vals(1,1) = 0;
cost =(x(1)).^2+(x(2)-3).^2;
X_Vals(end+1,:) = x;
F_Vals(end+1,1) = cost;
end
After optimisation is finished call objective function without arguments, it will return a structure containing the needed values.
cost = obj_function();
You'll get only points with the values, I am not sure if that is enough for the contour plot, if not, then check out scatteredInterpolant function.
Also, make sure you delete first row from both variables, since those are just values to initialise the variable. There's also another approach to run your optimisation nested, that way you can have variables in your workspace when optimisation is finished. You can find an example of it if you search the documentation or this website.

Show 1 older comment
Mario Malic on 3 Nov 2020
Replace your cost function with my code, add this after you call fmincon
Values = obj_function();
Utku Senel on 3 Nov 2020
Mario Malic on 3 Nov 2020
P.S. Don't forget to remove the first row in X_Vals and F_Vals.