What about collapsing the search space by using only one neuron in the chromosome and letting it clone, etc during an initial unsupervised ramp up period?

[TimeDelta]

This would probably have to be instead of using pre-existing algorithms though so might not be worth the reduction in search space because the headstart from pre-existing algorithms might be a bigger search space savings.

[TimeDelta]

Alternatively, this might be useful for mitigating the search space increase (Efficient combinatorial optimization by quantum-inspired parallel annealing in analogue memristor crossbar): https://www.nature.com/articles/s41467-023-41647-2