% % % % % % %pradines chromosomu populiacijos generavimas
% % % % % % chroms=genbin(6,5)
% % % % % % for i=1:5, rval(i)=bin2real(chroms(i,:),2,4);
% % % % % % end;
% % % % % % rval
% % % % % %
% % % % % % %Fitneso apskaiciavimas
% % % % % % [fit,fitot]=fitness('genf',chroms,2,4);
% % % % % % fit
% % % % % % fitot
% % % % % %
% % % % % % %Fitneso procentalumas
% % % % % % percent=fit/sum(fit)*100;
% % % % % % %percent
% % % % % % %sum(percent)
% % % % % %
% % % % % % %Nauja populiacija
% % % % % % cross=selectga('genf',chroms,2,4)
% % % % % % [fit,fitot]=fitness('genf',cross,2,4)
% % % % % % %percent=fit/sum(fit)*100;
% % % % % % %percent
% % % % % %
% % % % % % %Newgen kazkas nezinau po crossoverio
% % % % % % disp('newgen yra cia')
% % % % % % newgen=matesome(cross,.6)
% % % % % % [fit,fitot]=fitness('genf',newgen,2,4)
% % % % % %
% % % % % % %mutacija
% % % % % % disp('mutacija')
% % % % % % mutate(newgen,.005)

%main programike, kad nedegtu mums siknike
disp('main programa')
%[x f]=optga('genf',[2 4],6,5,5,.005,.6)
[x f]=optga1('genf',[0.001 0.3],6,5,10,.5,.8)

%Sugeneruoja random populiacija
function chromosome=genbin(bitl,numchrom)
% generation of a binary population
% word length bitl
% size of the population numchrom
maxchros=2^bitl;
if numchrom >= maxchros
    numchrom = maxchros;
end
chromosome = round(rand(numchrom,bitl));
end

%dvejetaine chromosoma pakeicia i desimtaini skaiciu(????)
function deci=bin2real(chrom,a,b)
% binary chromosome -> decimal number in the range from a to b
s=size(chrom);
[pop bitlength] = size(chrom);
maxchrom=2^bitlength-1;
% weights of binary digits
realel=chrom.*((2*ones(1,bitlength)).^fliplr([0:bitlength-1]));
% sumation
tot=sum(realel);
% range
deci=a+tot*(b-a)/maxchrom;
end

%Patrankos suvis
function target = genf(x)
% target value of x
%target = pi*x.*x.*(.666667*x+2);
target = 10 + (1/((x-0.160).^2 + 0.1))*sin(1/x);
end

%Apskaiciuojamas populiacijos fitneso lygis, reikia rasti
%labiausiai "fitnesuota" chromosoma ir ja pasirinkti reprodukcijai
%fitneso procentinis lygis nurodo, kad yra daugiau galimybiu patekti i
%kazkokia chromosomu populiacija
function [fit,fitot]=fitness(kriter,chrom,a,b)
% Fitness for the set of chromosomes in the range between a and b
% 'kriter' – the title of the external target function
[pop bitl]=size(chrom);
for k=1:pop
    v(k)=bin2real(chrom(k,:),a,b);
    fit(k)=feval(kriter,v(k));
end
fitot=sum(fit);
end

%Rekombinuoja nauja populiacija su "fitnesiskiausiom" chromosomom tam, kad
%gautu nauja, patobulinta populiacija
function newchrom=selectga(kriter,chrom,a,b)
% the title of the target function - ‘kriter’
[pop bitlength]=size(chrom);
fit=[];
% the assessment of chromosomes
[fit,fitot] = fitness(kriter,chrom,a,b);
for chromnum=1:pop
    sval(chromnum)=sum(fit(1,1:chromnum));
end
% roulette
parname=[];
for i=1:pop
    rval=floor(fitot*rand);
    if rval<sval(1)
        parname=[parname 1];
    else
        for j=1:pop-1
            sl=sval(j);
            su=sval(j)+fit(j+1);
            if (rval>=sl) & (rval<=su)
                parname=[parname j+1];
            end
        end
    end
end
newchrom(1:pop,:)=chrom(parname,:);
end

function chrom1=matesome(chrom,kiek)
% crossover procedure
% chrom – the initial set of genes
% chrom1 – genes after the crossover
% kiek – crossover degree from 0 to 1
mateind = [ ];
chrom1 = chrom;
[pop bitlength] = size(chrom);
ind = 1:pop;
u = floor(pop*kiek);
if floor(u/2) ~= u/2
    u = u-1;
end
% random crossover
while length(mateind)~=u
    i = round(rand*pop);
    if i == 0
        i = 1;
    end
    if ind(i) ~= -1
        mateind = [mateind i];
        i = -1;
    end
end
% crossover at a random point
for i = 1:2:u-1
    splitpos = floor(rand*bitlength);
    if splitpos == 0
        splitpos = 1;
    end
    i1 = mateind(i);
    i2 = mateind(i+1);
    tempgene = chrom(i1,splitpos+1:bitlength);
    chrom1(i1,splitpos+1:bitlength)=chrom(i2,splitpos+1:bitlength);
    chrom1(i2,splitpos+1:bitlength)=tempgene;
end
end

%mutacija
function chrom=mutate(chrom,mu)
% mutation of chromosomes
% mu – the mutation strength; usually assumed as a small positive number
[pop bitlength] = size(chrom);
for i=1:pop
    for j=1:bitlength
        if rand <= mu
            if chrom(i,j) == 1
                chrom(i,j) = 0;
            else
                chrom(i,j) = 1;
            end
        end
    end
end
end

function [xval,maxf]=optga(fun,range,bits,pop,gens,mu,matenum)
% GA optimization
% fun – the target function
% range[x1 x2] – the range of x: from x1 to x2
% bits – the number of bits in a word
% pop – the number of chromosomes in the population
% gens – the number of generations
% mu – the mutation parameter
newpop=[ ];
a=range(1); b=range(2);
newpop=genbin(bits,pop)
for i=1:gens
    selpop = selectga(fun,newpop,a,b);
    newgen = matesome(selpop,matenum);
    newgen1 = mutate(newgen,mu);
    newpop = newgen;
end
[fit,fitot] = fitness(fun,newpop,a,b)
[maxf,mostfit] = max(fit)
xval = bin2real(newpop(mostfit,:),a,b);
end


%dizaino savaite buvo labai gera, nes ismokome kurti burtazodi be ranku
function [xval,maxf]=optga1(fun,range,bits,pop,gens,mu,matenum)
mu
matenum

Y = ones(1,pop);
axis ([0 gens range(1) range(2)])
newpop=[ ];
a=range(1); b=range(2);
newpop=genbin(bits,pop);
for i=1:gens
    for iii = 1:pop
        XXX(iii) = bin2real(newpop(iii,:),a,b);
    end
    YYY = i*Y;
    [fit,fitot] = fitness(fun,newpop,a,b);
    fit
    plot(YYY,XXX,'o')
    hold on
    selpop = selectga(fun,newpop,a,b);
    newgen = matesome(selpop,matenum);
    newgen1 = mutate(newgen,mu);
    newpop = newgen;
end
[fit,fitot] = fitness(fun,newpop,a,b)
[maxf,mostfit] = max(fit)
xval = bin2real(newpop(mostfit,:),a,b);
end