Hi!

First and foremost, I want to express my gratitude for this incredible library and its clear documentation. Congratulations @ahmedfgad on the excellent work.
Thanks!

Now, version 3.3.0 is released... 👀 maybe we already have the variable chromossome feature? 👀

If not, my approach would simply have a big length gene_space and consider some special values (like -1) as the absent of that gene (I'm using "gene" as sinonym of "chromossome" here, but I know there's a small conceptual difference).

In my specific case, my optimization involves positioning weights in certain positions (which must be integers in a given range).
I don't know before hand how many weights I need, but I know it won't be bigger than, let's say, 5 weights. So I could have:

gene_space = [
range(-1, max_weight),
range(-1, max_weight),
range(-1, max_weight),
range(-1, max_weight),
range(-1, max_weight),
]

(I know I could just use gene_space=range(-1, max_weight),, but was to clarify that each individual would have 5 genes).

Then, on my fitness function I would have something like:

import pygad
from typing import Tuple, Literal
WeightPositionAsNone = Literal[-1]
MaxWeightPosition = Literal[5]
WeightPosition = Literal[WeightPositionAsNone,
 0, 1, 2, 3, 4, MaxWeightPosition]
Solution = Tuple[
 WeightPosition,
 WeightPosition,
 WeightPosition,
 WeightPosition,
 WeightPosition,
]
def fitness_func(
 ga_instance: pygad.GA,
 solution: Solution,
 solution_idx: int
):
 filtered_solution = [x for x in solution if x != -1]
 # Calculate fitness solution for the problem
 # desconsidering the weights that are not present
 fitness = sum(filtered_solution)
 return fitness

Does this seem like a valid approach?

This is absolutely a good idea. Still the feature is not fully merged. It is a major change.