CEC2010_F8 函数与论文表述不一致

[evibhm]

根据论文 Mallipeddi R, Suganthan P N. Problem definitions and evaluation criteria for the CEC 2010 competition on constrained real-parameter optimization[J]. Nanyang Technological University, Singapore, 2010, 24: 910. 中的描述，CEC2010_F8的约束函数应该是：

而PlatEMO平台代码中CEC2010_F8的约束函数是：

https://github.com/BIMK/PlatEMO/blob/69a0e5bb046de3d63afc89360bad4f2ba3472a85/PlatEMO/Problems/Single-objective%20optimization/CEC%202010/CEC2010_F8.m#L47C9-L53C12

%% Calculate constraint violations
function PopCon = CalCon(obj,PopDec)
    Z = PopDec - repmat(obj.O(1:size(PopDec,2)),size(PopDec,1),1);
    Y = Z*obj.Mat;
    PopCon(:,1) = abs(mean((-Y.*sin(sqrt(abs(Y)))),2)) - 1e-4;
    PopCon(:,2) = 0.5 - exp(-0.1*sqrt(mean(Y.^2,2))) - 3*exp(mean(cos(0.1*Y),2)) + exp(1);
end

请问多出来的这个约束函数是写错了还是有什么其他的考量？

[2ivers]

the paper, the formula and the mean function used in the code implementation both calculate the mean, which corresponds to taking the column-wise average of all matrices. 1/D notation in the literature refers to summing each component and then averaging, which is mathematically equivalent. use of the mean function in the code is accurate and actually simplifies the process.

[evibhm]

@czichiy I think PopCon(:,1) is redundant

[2ivers]

PopCon(:,1) = abs(mean((-Y.*sin(sqrt(abs(Y)))),2)) - 1e-4;
PopCon(:,2) = 0.5 - exp(-0.1*sqrt(mean(Y.^2,2))) - 3*exp(mean(cos(0.1*Y),2)) + exp(1);

if the result of PopCon(:,1) is not used in other parts of the code, then it is indeed redundant. 😁