CODEWITHSUNDEEP
matlaboptimizationlinear-programming
Ahmed Mohmed
Ahmed Mohmed

Reputation: 1

Affine scaling method in Matlab

I am trying to code the affine scaling method for maximization LP problem and I used the following code. 

function [X,y,k] = affsm(A,x,c)
xx(:,1)=x';
er=1e-5
 k=1;
test=1;
r=2/3;
theta=1
while test>er/theta
D = diag(xx(:,k).^2);
AD = A*D;
dx = -(D-AD'*(AD*A')^(-1)*AD)*c';
 theta =r*min([xx(:,k)./abs(dx);1]);
 xx(:,k+1) = xx(:,k) + theta*dx ;
 test=max([c*dx ,norm(dx)]);
 k=k+1;
end
y=xx(:,k);
X=xx;
end

I am trying to solve the following problem 

\begin{array}{c}{\max \quad Z=3 x+5 y} \\ {x+3 y \leq 60} \\ {3 x+4 y \leq 120} \\ {x \geq 10} \\ {x, y \geq 0}\end{array}

but it gives me wrong results what I am doing wrong in the above code ?.

Note: The optimal solution must be $(x,y)=(12,24)$ with $z=132$

Upvotes: 0

Views: 601

Answers (1)

obchardon
obchardon

Reputation: 10790

I don't know what your function is supposed to do since there is no explaination, but to resolve this kind of linear equation with constraints you can use linprog

% define the constraints
% Every constraints have this format: x1*x(1) + x2*x(2) ≤ n
% Where x(1) and x(2) are your variable and x1,x2 and n are integers.
A = [1 3
    3 4
    -1 0
     0 -1
    -1 0];

b = [60 120 -10 0 0];
% define the objective function
f = [-3 -5];
% solve
x = linprog(f,A,b)

Result:

x = 
    24
    12

Noticed that I've reversed the sign of the objective function in order to maximize the solution (by default linprog will minimize the solution)

Upvotes: 1

Related Questions