Question

有两件事值得注意：

fmincon只接受只有1个参数的函数。所以我需要每次只定义一个只有一个参数的新函数。更多详细信息，请参阅http://www.mathworks.com/help/optim/ug/passing-extra-parameters.html
但是每次都无法定义匿名函数。更多细节由Change Matlab function in loop

function V_t = backward_recursion(W_0_L,W_0_U,Q_L,Q_U,T,grid_point_num,N,M,R_f,K,alpha,omega_0,mu_R,Sigma_R)
    grid = create_grid(W_0_L,W_0_U,Q_L,Q_U,T,grid_point_num) ;
    v_t_plus_one_list = [0,1,0] ;
    for t = T-1:-1:0
        grid_t = grid(t+1,:) ;
        bellman_right_hand_side_with_max = zeros(1,grid_point_num) ; 
        for W_index = size(grid_t,1)
            W_t = grid_t(W_index,1) ;
            disp(['W_t=',num2str(W_t)])
            bellman_right_hand_side_with_max_opt = @(omega) bellman_right_hand_side_without_max(omega, N, M, R_f, alpha, W_t, mu_R, Sigma_R, v_t_plus_one_list) ;
            [x,fval] = fmincon(bellman_right_hand_side_with_max_opt,omega_0,ones(1,N),1,[],[],[0,0,0],[1,1,1]);
            bellman_right_hand_side_with_max(W_index,1) = - fval ;
        end
        V_t = - log( - bellman_right_hand_side_with_max ) / alpha ; 
        v_t_plus_one_list = polyfit(grid_t, V_t, K) ;
    end
end

我尝试了嵌套功能，但它没有工作

function [x,fval] = runnested(omega_0, N, M, R_f, alpha, W_t, mu_R, Sigma_R, v_t_plus_one_list)
    [x,fval] = fmincon(nested_fun,omega_0,ones(1,N),1,[],[],[0,0,0],[1,1,1]) ;
    function y = nested_fun(omega)
        y = bellman_right_hand_side_without_max(omega, N, M, R_f, alpha, W_t, mu_R, Sigma_R, v_t_plus_one_list) ;
    end
end

和

    [~,bellman_right_hand_side_with_max_result] = runnested(omega_0, N, M, R_f, alpha, W_t, mu_R, Sigma_R, v_t_plus_one_list) ;

    bellman_right_hand_side_with_max(W_index,1) = - bellman_right_hand_side_with_max_result ;

说

Not enough input arguments.

Error in runnested/nested_fun (line 4)
        y = bellman_right_hand_side_without_max(omega, N, M, R_f, alpha, W_t, mu_R, Sigma_R, v_t_plus_one_list) ;

Error in runnested (line 2)
    [x,fval] = fmincon(nested_fun,omega_0,ones(1,N),1,[],[],[0,0,0],[1,1,1]) ;

如何在循环中使用带有额外参数的fmincon？

0 个答案: