Optimization Problem where Objective is not a direct function of Variable

Hello !

I need to solve an optimization problem by minimizing the value of a variable B by iterating on the value of a variable x as follows:

prob = optimproblem;
x = optimvar('x','LowerBound',0,'UpperBound',100);

A = ComputeA (x);
B = ComputeB(A,x);

prob.Objective = B;
prob.Constraints.cons1 = B < ref;;

I can not manage to have this work. I have read through documentation but, in every example I came across, the problem Objective is a direct function of the optimvariable, wether linear or non-linear. In my case, I have to go through multiple calculations to go from x to B and so I can not get my prob.Objective to be a simple equation.

Has anyone ever coded something similar ?

Thank you in advance,

Regards.

[Walter Roberson]

In some situations, if you have the symbolic toolbox, you can combine the equations by using something like

syms X
A = ComputeA(X);
B = ComputeB(A, X);
obj = matlabFunction(B, X);

and use solver-based optimization.

This will not work in all cases, and it is not uncommon that it needs a bit of tweaking of the code to make it work. For example, code such as

p = 0;
for K = 1 : 5; p = p + x.^p; end

might need to be change to instead use

 p = zeros(1,1,'like', x);