R help - May 2025 - How to get the same result for GA optimization?

I am using *Genetic Algorithm* to maximize some function which use data.

I use GA package in R for this (
https://cran.r-project.org/web/packages/GA/index.html)

Below is my code

library(GA)
set.seed(1)
Dat = data.frame(rnorm(1000), matrix(rnorm(1000 * 30), nc = 30))

Fitness_Fn = function(x) {
    return(cor(Dat[, 1], as.matrix(Dat[, -1]) %*% matrix(x, nc = 1))[1,1])
}

ga(type = 'real-valued', fitness = Fitness_Fn, seed = 1, lower rep(0,
30), upper = rep(1, 30), elitism = 10, popSize = 200, maxiter 1, pcrossover =
0.9, pmutation = 0.9, run = 1)

However now I alter the columns of my data and rerun the GA

Dat = Dat[, c(1, 1 + sample(1:30, 30, replace = F))]
Fitness_Fn = function(x) {
    return(cor(Dat[, 1], as.matrix(Dat[, -1]) %*% matrix(x, nc = 1))[1,1])
}

ga(type = 'real-valued', fitness = Fitness_Fn, seed = 1, lower rep(0,
30), upper = rep(1, 30), elitism = 10, popSize = 200, maxiter 1, pcrossover =
0.9, pmutation = 0.9, run = 1)

Surprisingly, I get different result from above 2 implementations.

In first case, I get

GA | iter = 1 | Mean = 0.01534124 | Best = 0.04351926

In second case,

GA | iter = 1 | Mean = 0.01705027 | Best = 0.04454167

I have fixed the random number generation using seed = 1 in the ga() function,
So I am expecting I would get exactly same result.

Could you please help identify the issue here? I want to get exactly same
result irrespective of the order of the columns of dat for above function.

Thanks for your time.

	[[alternative HTML version deleted]]

Dear Daniel

As far as I can see you have re-generated the data before calling ga() 
so the data is not just a permutation of the first set.

Michael

On 07/05/2025 15:36, Daniel Lobo wrote:
> I am using *Genetic Algorithm* to maximize some function which use data.
> 
> I use GA package in R for this (
> https://cran.r-project.org/web/packages/GA/index.html)
> 
> Below is my code
> 
> library(GA)
> set.seed(1)
> Dat = data.frame(rnorm(1000), matrix(rnorm(1000 * 30), nc = 30))
> 
> Fitness_Fn = function(x) {
>      return(cor(Dat[, 1], as.matrix(Dat[, -1]) %*% matrix(x, nc = 1))[1,1])
> }
> 
> ga(type = 'real-valued', fitness = Fitness_Fn, seed = 1, lower >
rep(0, 30), upper = rep(1, 30), elitism = 10, popSize = 200, maxiter > 1,
pcrossover = 0.9, pmutation = 0.9, run = 1)
> 
> However now I alter the columns of my data and rerun the GA
> 
> Dat = Dat[, c(1, 1 + sample(1:30, 30, replace = F))]
> Fitness_Fn = function(x) {
>      return(cor(Dat[, 1], as.matrix(Dat[, -1]) %*% matrix(x, nc = 1))[1,1])
> }
> 
> ga(type = 'real-valued', fitness = Fitness_Fn, seed = 1, lower >
rep(0, 30), upper = rep(1, 30), elitism = 10, popSize = 200, maxiter > 1,
pcrossover = 0.9, pmutation = 0.9, run = 1)
> 
> Surprisingly, I get different result from above 2 implementations.
> 
> In first case, I get
> 
> GA | iter = 1 | Mean = 0.01534124 | Best = 0.04351926
> 
> In second case,
> 
> GA | iter = 1 | Mean = 0.01705027 | Best = 0.04454167
> 
> I have fixed the random number generation using seed = 1 in the ga()
function,
> So I am expecting I would get exactly same result.
> 
> Could you please help identify the issue here? I want to get exactly same
> result irrespective of the order of the columns of dat for above function.
> 
> Thanks for your time.
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
https://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
-- 
Michael Dewey