Question

我有一个权重向量w，我希望为其设置一个值范围。

尤其是以下内容：

当w（i）> 0

0.0001 <= w（i）<= 0.01
当w（i）<= 0

-0.01 <= w（i）<= -0.0001

如何在fmincon中翻译？我尝试了以下方法：

% problem
fnhandle                    = @(x)in_objectfun(x, ymu);
nonlcon                     = @(x)in_nonlconstr(x);
[A, b, Aeq, beq, lb, ub]    =     in_linconstr(imu);
x0                          =     in_startvalue(imu); % equal weighted with sign respected

% optimization
[woptimal,fval,exitflag,output,lambda,grad,hessian] = fmincon(fnhandle, x0, A, b, Aeq, beq, lb, ub, nonlcon, options); 


% the core script where I have difficulties with
function [A, b, Aeq, beq, lb, ub] = in_linconstr(imu)


        A   = [];
        b   = [];
        Aeq = [];
        beq = [];
        lb  = [];
        ub  = [];

        % the sign of the weights is determined by imu
        temp              = zeros(1,nbAssets);
        temp(imu >= 0)    = 1;
        temp(imu < 0)     = -1;

        tempPos           = zeros(1,nbAssets);
        tempPos(imu>=0)   = 1;

        tempNeg           = zeros(1,nbAssets);
        tempNeg(imu<0)    = -1;


        % -----------------------------------------------------------------
        % A: Sign of the Weight

        Asign  = diag(-temp);
        bsign  = zeros(nbAssets,1);

        A      = [A; Asign];
        b      = [b; bsign];


        % -----------------------------------------------------------------
        % B: Single exposures


        % A x >= 0.0001
        AposMin           = -diag(tempPos);
        bposMin           = zeros(nbAssets,1);
        bposMin(imu>=0)   = -0.0001;

        % A x < 0.01
        AposMax           = diag(tempPos);
        bposMax           = zeros(nbAssets,1);
        bposMax(imu>=0)   = 0.01;

        % A x < -0.0001 
        AnegMin           = diag(tempNeg);
        bnegMin           = zeros(nbAssets,1);
        bnegMin(imu<0)    = 0.0001;

        % A x >= -0.01
        AnegMax           = -diag(tempNeg);
        bnegMax           = zeros(nbAssets,1);
        bnegMax(imu<0)    = -0.01;

        A = [A; AposMin];
        b = [b; bposMin];

        A = [A; AposMax];
        b = [b; bposMax];

        A = [A; AnegMin];
        b = [b; bnegMin];

        A = [A; AnegMax];
        b = [b; bnegMax];

end

运行以下命令时，我发现其中一个约束未得到满足，因此我猜想我的公式出了点问题：

[wraw,fval,exitflag,output,lambda,grad,hessian] = fmincon(fnhandle, x0, A, b, Aeq, beq, lb, ub, nonlcon, options); %#ok<*ASGLU>

正值和负值的线性约束范围不同

0 个答案: