资源简介
多目标优化的一个性能指标-超立方体,它可以测试算法的收敛性也可以测试算法的多样性。
代码片段和文件信息
function res=exi2d(Prmus)
% P: approximation set 2xK r: reference point mu: mean vector s:stdev
% vector
% Example: exi2d([02;11;20][33][00][0.10.1]) should approximately
% result in 3
%
% determine all lower left corner cell coordinates
S=sortrows(P);
k=length(S);
c2 = sort(S(:2));
c1 = sort(S(:1));
for i=0:k % hoogte van beneden naar boven; hight from below to above
for j=0:(k-i) % first coordinate (length) of cell grid
% c1(i) c2(j) are now the cell coordinates according Fig. 2
% For coordinate j determine hight fMax2
if (j==0) fMax2=r(2); else fMax2 = c2(k+1-j); end;
% For coordinate i determine the width of the staircase fMax1
if (i==0) fMax1=r(1); else fMax1 = c1(k+1-i); end;
% get cell coordinates
if (j==0) cL1= -inf; else cL1 = c1(j); end;
if (i==0) cL2= -inf; else cL2 = c2(i); end;
if (j==k) cU1 = r(1); else cU1 = c1(j+1); end;
if (i==k) cU2 = r(2); else cU2 = c2(i+1); end;
% SM = points that are dominated or equal to upper cell bound
SM=[;];
for m=1:k
if (cU1 <= S(m1) && cU2 <= S(m2))
SM=[[S(m1)S(m2)];SM];
end
end
sPlus = hvolume2d(SM [fMax1fMax2]);
%Marginal integration over the length of a cell
Psi1 = exipsi(fMax1cU1mu(1)s(1)) - exipsi(fMax1cL1mu(1)s(1));
%Marginal integration over the height of a cell
Psi2 = exipsi(fMax2cU2mu(2)s(2)) - exipsi(fMax2cL2mu(2)s(2));
%Cumulative Gaussian over length for correction constant
GaussCDF1 = gausscdf((cU1-mu(1))/s(1)) - gausscdf((cL1-mu(1))/s(1));
%Cumulative Gaussian over length for correction constant
GaussCDF2 = gausscdf((cU2-mu(2))/s(2)) - gausscdf((cL2-mu(2))/s(2));
%ExI Kontribution fuer die aktuelle Zelle
c(i+1j+1)= Psi1*Psi2-sPlus*GaussCDF1*GaussCDF2;
end
end
res=sum(sum(max(c0)));
属性 大小 日期 时间 名称
----------- --------- ---------- ----- ----
目录 0 2010-04-28 15:05 HV_ba
文件 2033 2010-04-27 19:14 HV_ba
文件 80 2010-04-26 22:21 HV_ba
文件 51 2010-04-27 18:50 HV_ba
文件 59 2010-04-27 18:46 HV_ba
文件 339 2010-04-26 22:21 HV_ba
文件 1283 2010-04-27 19:29 HV_ba
- 上一篇:燕山大学编译原理实验报告.doc
- 下一篇:缠论笔段中枢第二版 for 通达信
评论
共有 条评论