replicating excel solver using optimization linprog

I need to minimize the maximum value of buffer. Where Buffer(i)=buffer(i-1)+(gen-demand). Demand is given as a row matrix [10 1 5 2]. We need to vary the generation data (x) and initial value of buffer (y) under some linear inequality constraints ,to minimize the maximum value of buffer. I tried the below code, with [v] matrix as my initial guess of x and y: (i have not introduced constraints yet)

v=[0 0 0 0 1];

a=linprog(objective_fun(x,y) ,v);

function output=objective_fun (x,y)

demand=[10 1 5 2];

gen=ones(1,4)*x;

mis=gen-demand;

buffer=zeros(1,4);

buffer(1,1)=y;

for i=2:4

buffer(i)=buffer(i-1)+mis(i);

end

output=max(buffer);

end

I am getting error of not enough input. Please help.

Answer 1

[Alan Weiss]

I think that you would probably have an easier time formulating your problem by using the Problem-Based Optimization Workflow.

But to answer your specific question, the syntax for linprog requires objective_fun to be a vector, not a function of two variables as you have stated:

a=linprog(objective_fun(x,y) ,v); % objective_fun(x,y) looks like an error

I recommend again that you try the problem-based approach. It is much easier to user and less error-prone.

Alan Weiss

MATLAB mathematical toolbox documentation