Error: Array indices must be positive integers or logical values. implementing fletcher reeves method

I am trying to implement the fletcher reeves method in matlab using "fminbnd". This is my code:

function[xmin,valormin] = fletcherreeves2(x0,funcao,grad,tol)
d0=-grad(x0);
x=x0;
d=d0;
k=0;
while norm(grad(x))>tol
    f_lambda=@(lambda) funcao(x+lambda*x);
    alpha=fminbnd(f_lambda,0,1000);
    xk=x+alpha*d;
    gradnovo=grad(xk);
    beta=(norm(gradnovo))^2/(norm(grad(x)))^2;
    dk=-gradnovo+beta*d;
    k=k+1;
    x=xk;
    d=dk;
    grad=gradnovo;
end
xmin=x;
valormin=funcao(xmin);
end

I am getting this error:

Error in fletcherreeves2 (line 6)

while norm(grad(x))>tol

I have tried using the following debug statements:

disp(d0)
disp(grad(x0))
disp(norm(grad(x0)))
disp(norm(grad(x0))>tol)

I am using the following input:

fletcherreeves2([5;5],@(x) x(1)^2-x(1)*x(2)+4*x(2)^2+3*x(1)+2,@(x) [2*x(1)-x(2)+3,-x(1)+8*x(2)],10^-(6))

and always receive the output i am supposed to, so I really don't know what my error is and I really have no idea what to do, any help is greatly appreciated!

Answer 1

[Walter Roberson]

fletcherreeves2([5;5],@(x) x(1)^2-x(1)*x(2)+4*x(2)^2+3*x(1)+2,@(x) [2*x(1)-x(2)+3,-x(1)+8*x(2)],10^-(6))

The third parameter there is an anonymous function

function[xmin,valormin] = fletcherreeves2(x0,funcao,grad,tol)

The anonymous function is being received under the name grad

d0=-grad(x0);

The anonymous function gets invoked.

while norm(grad(x))>tol

First time through, the anonymous function gets invoked.

        gradnovo=grad(xk);

There too, returning a numeric value.

            grad=gradnovo;

The numeric value is assigned to the variable grad

while norm(grad(x))>tol

Second time through, grad is a numeric variable that you are indexing at locations given by x

Answer 2

[Torsten]

[xmin,valormin]=fletcherreeves2([5;5],@(x) x(1)^2-x(1)*x(2)+4*x(2)^2+3*x(1)+2,@(x) [2*x(1)-x(2)+3;-x(1)+8*x(2)],10^-(6))

xmin = 2×1

-1.6000 -0.2000

valormin = -0.4000

function[xmin,valormin] = fletcherreeves2(x0,funcao,grad,tol)
  gradvelho = grad(x0);
  d0=-gradvelho;
  x=x0;
  d=d0;
  k=0;
  while norm(gradvelho)>tol
    f_lambda=@(lambda) funcao(x+lambda*d);
    alpha=fminbnd(f_lambda,0,1000);
    xk=x+alpha*d;
    gradnovo=grad(xk);
    beta=(norm(gradnovo))^2/(norm(gradvelho))^2;
    dk=-gradnovo+beta*d;
    k=k+1;
    x=xk;
    d=dk;
    gradvelho=gradnovo;
  end
  xmin=x;
  valormin=funcao(xmin);
end