Problem with optimization using fminunc + GradObj

Hello,

I have problem with optimization using fminunc with GradObj set to on. I'm trying to find minimum in Rosenbrock function - which have only one minimum at [1.0, 1.0]. Without GradObj fminunc is ale to find optimum. With GradObj function stops after 3 iterations with message:

-----

                                                          First-order 
   Iteration  Func-count       f(x)        Step-size       optimality
       0           1             24.2                           211
       1           3          4.25728    0.000871384           17.8  
       2           4          4.12914              1           3.16  

Local minimum possible.

fminunc stopped because it cannot decrease the objective function
along the current search direction.

<stopping criteria details>

X =

     -1.0306    1.0697

FVAL =

      4.1291

EXITFLAG =

       5

OUTPUT = 

         iterations: 3
          funcCount: 22
           stepsize: 1
      firstorderopt: 3.1610
          algorithm: 'medium-scale: Quasi-Newton line search'
            message: [1x362 char]

-----

Here is code I'm running:

function [f,G] = func1(x)
    f = (1 - x(1))*(1 - x(1)) + 100*(x(2) - x(1)*x(1))*(x(2) - x(1)*x(1));
    % Gradient of the objective function
    if nargout > 1
        G = [-2 - 2*x(1) - 400*x(2)*x(1) + 400*x(1)*x(1)*x(1), 
                200*x(2) - 200*x(1)*x(1)];
    end
end

...

options = optimset('LargeScale', 'off', 'InitialHessType', 'Scaled-Identity', 'GradObj', 'on', 'Display', 'iter');
% where startX = -1.2 and startY = 1.0
[X,FVAL,EXITFLAG,OUTPUT] = fminunc(@func1,[startX, startY], options)

Please, help me solve this problem.

Answer 1

[Andrew Newell]

It's a sign error. The gradient should be

G = [-2 + 2*x(1) - 400*x(2)*x(1) + 400*x(1)^3
              200*x(2) - 200*x(1)^2];