CODEWITHSUNDEEP
matlabgenetic-algorithm
Jake
Jake

Reputation: 7773

Sampling with replacement in Matlab

I have a set of n genomes, and n corresponding fitness values. I would like to sample the genomes with replacement according to their proportional fitness values, such that P(genome) = fitness(genome) / sum(fitness(genomes)). Is there a built-in mechanism for sampling like this in Matlab?

Upvotes: 0

Views: 696

Answers (3)

petrichor
petrichor

Reputation: 6569

You can use randsample for this purpose:

>> f = [2 5 7 8 1]; %# fitness values (not normalized)
>> g = {'g1', 'g2', 'g3', 'g4', 'g5'}; %# genome names
>> gSampled = randsample(g,100,true,f) %# sample 100 genomes with replacement

Upvotes: 0

Chris Taylor
Chris Taylor

Reputation: 47392

If you wanted to get 10 samples from a population of 4 whose fitnesses are 0.4, 0.3, 0.2 and 0.1 respectively, then you can do:

>> fitness = [0.4 0.3 0.2 0.1];
>> mnrnd(10,fitness)
ans =
     3     5     1     1

Upvotes: 1

jkt
jkt

Reputation: 2578

Matlab has built in multinomial random number generator. You can use mnrnd for your purpose.

Upvotes: 0

Related Questions