variable=2; % definite the number of variables
bits=[20, 20]; %the bits for every variables used
%bound=[-100 100];
%pl=ones(variable,1);
bounds=[-100,100;-100,100]; % definite the bounds of every variable
MatrixLen=sum(bits)+1; %the length of the matrix: variables and fitness
popsize=100; % the inited population number, the row of the matrix
MaxGeneration=300; % run the loop MaxGeneration times
StartPop=zeros(popsize,MatrixLen);
Generation=1;
Done = 0;
scale=(bounds(:,2)-bounds(:,1))′./(2.^bits-1); % scale for the min bit 数据挖掘实验室
cs=[0 cumsum(bits)]; % cs: the start position of every variable
Pc=0.6; %set the probability of crossover
Pm=0.05; %set the probability of mutation
hold on;
axis([0 MaxGeneration 15 25]);
% 【Initialize the StartPop for SGA】
for i=1:popsize
for j=1:MatrixLen-1
if rand >= 0.5
StartPop(i,j)=1;
end;%endif
end;%endfor j 数据挖掘研究院
end;%endfor i
% evalue the fitness of individuals in the first generation
for i = 1:popsize
for j = 1:variable
a=StartPop(i,cs(j)+1:cs(j+1)); % the matrix for every variable
var(j)=sum(2.^(bits(j)-1:-1:0).*a)*scale(j)+bounds(j,1); % the value of every variable
end;
% plot(var(1),var(2),′r*′);
StartPop(i,MatrixLen) = 25 - (var(1).^2+var(2).^2).^0.25.*(sin(50.*(var(1).^2+var(2).^2).^0.1).^2 + 1);
end;
% 【End of initialization】
% plot the points of the first generation
% plot(1:popsize,StartPop(1:popsize,MatrixLen),′o′);
while(~Done)
[value,posit] = max(StartPop(:,MatrixLen));
Best = StartPop(posit,:); % store the best individual
TraceInfo(Generation,1) = value; % the best fitness before reproduction
TraceInfo(Generation,2) = mean(StartPop(:,MatrixLen)); % average fitness before reproduction
TraceInfo(Generation,3) = std(StartPop(:,MatrixLen)); % Standard Deviation
TraceInfo(Generation,4) = size(unique(StartPop(:,1:MatrixLen-1),′rows′),1); %diversity
% plot(Generation,TraceInfo(Generation,1),′r*′); % plot the best fitness 数据挖掘研究院
% plot(Generation,TraceInfo(Generation,2),′*′); % plot the average fitness
% 【Reproduction by Gambling】
EndPop=zeros(popsize,MatrixLen);
fitness=cumsum(StartPop(:,MatrixLen)′./sum(StartPop(:,MatrixLen))); % fitness matrix for every individual
for i = 1:popsize
b = rand;
j = 1;
while(b > fitness(j))
j = j+1;
end; %endwhile
EndPop(i,:)=StartPop(j,:);
%plot(j,EndPop(i,39),′x′) //plot the reproducted individuals
end; %endfor i
TraceInfo(Generation,5) = mean(EndPop(:,MatrixLen)); % average fitness after reproduction;before crossover
% 【End of Reproduction】
% 【Crossover Operation】
for i = 1:round(popsize.*Pc)
numA = 0;
numB = 0;
while(numA == numB) % numA numB stand for 2 individuals will cross
numA = round(rand*popsize+0.5);
numB = round(rand*popsize+0.5);
end;%endwhile
cPoint = round(rand*(MatrixLen-3)) + 1; % cPoint is the cross position
p1 = EndPop(numA,:);
p2 = EndPop(numB,:);
cPopA = [p1(1:cPoint) p2(cPoint+1:MatrixLen-1)];
cPopB = [p2(1:cPoint) p1(cPoint+1:MatrixLen-1)];
if cPopA(1:MatrixLen-1) == p1(1:MatrixLen-1)
cPopA(MatrixLen) = p1(MatrixLen);
cPopB(MatrixLen) = p2(MatrixLen);
elseif cPopA(1:MatrixLen-1) == p2(1:MatrixLen-1)
cPopA(MatrixLen) = p2(MatrixLen);
cPopB(MatrixLen) = p1(MatrixLen);
else
for j = 1:variable
a1=cPopA(cs(j)+1:cs(j+1));
a2=cPopB(cs(j)+1:cs(j+1));
var1(j)=sum(2.^(size(a1,2)-1:-1:0).*a1)*scale(j)+bounds(j,1); 数据挖掘研究院
var2(j)=sum(2.^(size(a2,2)-1:-1:0).*a2)*scale(j)+bounds(j,1);
end;
cPopA(MatrixLen) = 25 - (var1(1).^2+var1(2).^2).^0.25.*(sin(50.*(var1(1).^2+var1(2).^2).^0.1).^2 + 1);
cPopB(MatrixLen) = 25 - (var2(1).^2+var2(2).^2).^0.25.*(sin(50.*(var2(1).^2+var2(2).^2).^0.1).^2 + 1);
end;%endif
EndPop(numA,:) = cPopA;
EndPop(numB,:) = cPopB;
end;%endfor
TraceInfo(Generation,6) = mean(EndPop(:,MatrixLen)); % average fitness after crossover;before mutation
% 【End of Crossover】
% 【mutation】
for i = 1:round(Pm.*popsize)
numA = round(rand*popsize+0.5);
mPoint = round(rand*(MatrixLen-2)) + 1;
EndPop(numA,mPoint) = 1 - EndPop(numA,mPoint);
for j = 1:variable
a = EndPop(numA,cs(j)+1:cs(j+1));
var(j) = sum(2.^(size(a,2)-1:-1:0).*a).*scale(j)+bounds(j,1);
end;%endfor j
EndPop(numA,MatrixLen) = 25 - (var(1).^2+var(2).^2).^0.25.*(sin(50.*(var(1).^2+var(2).^2).^0.1).^2 + 1);
end;%endfor i
TraceInfo(Generation,7) = mean(EndPop(:,MatrixLen)); % average fitness after mutation
% 【End of mutation】
StartPop=EndPop;
% 精英保留策略
[value,posit] = min(StartPop(:,MatrixLen));
StartPop(posit,:) = Best;
% End of 精英保留策略
Generation = Generation + 1;
if Generation == MaxGeneration;
Done = 1;
end;%endif
end; %endwhile
% Record infomation of the last generation 数据挖掘研究院
[value,posit] = max(StartPop(:,MatrixLen));
Best = StartPop(posit,:); % store the best individual
TraceInfo(Generation,1) = value; % the best fitness before reproduction
TraceInfo(Generation,2) = mean(StartPop(:,MatrixLen)); % average fitness before reproduction
TraceInfo(Generation,3) = std(StartPop(:,MatrixLen)); % Standard Deviation
TraceInfo(Generation,4) = size(unique(StartPop(:,1:MatrixLen-1),′rows′),1); %diversity
% The information of matrix TraceInfo
% 1.Best fitness of per generation
% 2.Average fitness before repreduction
% 3.Standard Deviation
% 4.Diversity
% 5.Average fitness after reproduction;before crossover
% 6.Average fitness after crossover;before mutation
% 7.Average fitness after mutation
plot(TraceInfo(:,1),′r*′);
plot(TraceInfo(:,2),′*′);
%plot(TraceInfo(:,5),′r-′);
%plot(TraceInfo(:,6),′b-′);
%plot(TraceInfo(:,7),′g-′);
