Argument passing issue.

Question

Ba Ba Black Sheep! 2017-1-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/318872-argument-passing-issue

编辑： Walter Roberson 2017-1-3

采纳的回答： Walter Roberson

在 MATLAB Online 中打开

I have a requirement that, I use the following modified Rosenbrock function,

where the coefficients a and b are to be read from a text file.

.

I tried the following,

 function out = rosenbrock(x)
    disp('rosenbrock()..........called');
    coeff = load('coeff.txt');
    a = coeff(1);
    b = coeff(2);
    xx = x(1);
    yy = x(2);
    out = (1 - xx + a)^2 + 100*(yy - b - (xx-a)^2)^2;
 end

But it is doing two things,

(1) Slowing down the performance of the optimization.

(2) The output is also not correct (the optimization isn't converging).

How can I solve this issue wile fulfilling my requirement?

Is it possible to pass the values of a and b as the arguments of rosenbrockwithgrad()?

Relevant Source Code

 function [x, fval, eflag, iter, fcount] = 
        Optimization_With_Analytic_Gradient(start_point)
    x0 = start_point;
    %   inline function defitions
    %fungrad = @(x)deal(fun(x),grad(x));
    % options setup
    options = optimoptions( 'fminunc', ...
                            'Display','off',...
                            'OutputFcn',@bananaout,...
                            'Algorithm','trust-region', ...
                            'GradObj','on');
    % calling fminunc
    [x, fval, eflag, output] = fminunc(@rosenbrockwithgrad, x0, options);
    iter = output.iterations;
    fcount = output.funcCount;
    %   plot window title
    title 'Rosenbrock with Analytic Gradient...'    
    disp('Optimization_With_Analytic_Gradient...');
end 
 function out = gradient( x )
    out = [-400*(x(2) - x(1)^2)*x(1) - 2*(1 - x(1)); 
            200*(x(2) - x(1)^2)];
end
 function [f,g] = rosenbrockwithgrad(x)
    % Calculate objective f
    f = rosenbrock(x);    
    if nargout > 1 % gradient required
        g = gradient(x);
    end
 end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2017-1-3

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/318872-argument-passing-issue#answer_249222

https://www.mathworks.com/help/matlab/math/parameterizing-functions.html

2 个评论
显示无隐藏无

Ba Ba Black Sheep! 2017-1-3

编辑：Ba Ba Black Sheep! 2017-1-3

在 MATLAB Online 中打开

I am extremely sorry. I could't figure it out.

 rosenbrockwithgrad_p = @(x)[rosenbrock(x, a, b); 
                            (nargout > 1) * gradient(x)];
    % calling fminunc
    [x, fval, eflag, output] = fminunc(rosenbrockwithgrad_p, x0, options);

Could you please help?

Walter Roberson 2017-1-3

编辑：Walter Roberson 2017-1-3

在 MATLAB Online 中打开

   function [x, fval, eflag, iter, fcount] = 
          Optimization_With_Analytic_Gradient(start_point, a, b)
      x0 = start_point;
      % options setup
      options = optimoptions( 'fminunc', ...
                              'Display','off',...
                              'OutputFcn',@bananaout,...
                              'Algorithm','trust-region', ...
                              'GradObj','on');
      % calling fminunc
      [x, fval, eflag, output] = fminunc(@(x) rosenbrockwithgrad(x, a, b), x0, options);
      iter = output.iterations;
      fcount = output.funcCount;
      %   plot window title
      title 'Rosenbrock with Analytic Gradient...'    
      disp('Optimization_With_Analytic_Gradient...');
  end 
   function out = gradient( x, a, b )
      %this routine probably needs to be changed to take a and b into account
      out = [-400*(x(2) - x(1)^2)*x(1) - 2*(1 - x(1)); 
              200*(x(2) - x(1)^2)];
   end
   function [f,g] = rosenbrockwithgrad(x, a, b)
      % Calculate objective f
      f = rosenbrock(x, a, b);    
      if nargout > 1 % gradient required
          g = gradient(x, a, b);
      end
   end