Genetic Algorithm Initial Population Problem in Matlab

How can I generate initial population of 50*11 in Matlab? The Upper and Lower limits are as under;

ub = [80 75 35 60 50 1.1 1.1 1.1 1.1 1.1 1.1];  
lb = [20 0 10 0 0 0.95 0.95 0.95 0.95 0.95 0.95];          
dim=11;
Population=50;

Answer 1

[Star Strider]

This is what I would do:

opts = optimoptions('ga', 'InitialPopulationMatrix',[randi([20 80], 50, 5) randi([95 110], 50, 6)*1E-2]);

The ga function does not need to know all the ranges, however it is (in my experience) usually better to begin with lower values than higher values, since ga more easily increases than decreases the individual elements in the population ‘genes’.

Include whatever other options you want. Then be sure to include ‘opts’ as the last argument to the ga call, remembering to fill all the intervening arguments with appropriate arguments or the empty matrix [].

[Salah Djerouni]

hi ,

can you tell me ,how can define the pouplation in my case

objfnc=@first4;
nvar=7;
%% Bounds  %
% %% beta
    Lbeta=0;
    Ubeta=0.5;
% %% Mass
    m1_min=0; % TMD
    m1_max=20;  

    c1_min=0; % TMD
    c1_max=370;  
    
    k1_min=0; % TMD
    k1_max=39765;  
    
    m2_min=0; % TMDI
    m2_max=50;  
    
    c2_min=0; % TMDI
    c2_max=370; 
    k2_min=0; % TMDI
    k2_max=99413;      
LB=[Lbeta  m1_min  c1_min  k1_min  m2_min  c2_min  k2_min ];
UB=[Ubeta  m1_max  c1_max  k1_max  m2_max  c2_max  k2_max ];
% options = gaoptimset('PlotFcns',@gaplotbestf );
% % options = gaoptimset(@gamultiobj,'PlotFcn',{@gaplotpareto});
% [x,fval]=ga(objfnc,nvar,[],[],[],[],LB,UB,[],options);

[Walter Roberson]

What is the reason to supply your own random initial population? ga will construct a population inside of the bounds you give.

There are reasons to use an initial population, especially if there are constraints that might be difficult satisfy, but I do not see the reasoning in this situation?

[Christopher Jarrett]

What if you needed to run the population of genes(variables) row by row threw a loop that did some calculation with the genes that generates another vector? How could you make it do that?

[Walter Roberson]

You would probably code that through the selection function, mutation function, or crossover function.