Odd behavior with stuck population

[mahrossi]

@mmarianski reported an odd behavior of populations having same members after a while (all low energy conformers? Must provide examples and check this behaviour.

[mmarianski]

I took a look at a GlcNAc ga run (30 generations). These are energies of the molecules forming a population at the end of the GA:
Structure 51:, energy: -22297.7064, is temporary added to the population
Structure 140:, energy: -22297.7058, is temporary added to the population
Structure 151:, energy: -22297.7058, is temporary added to the population
Structure 124:, energy: -22297.7053, is temporary added to the population
Structure 146:, energy: -22297.6852, is temporary added to the population
Structure 42:, energy: -22297.6851, is temporary added to the population
Structure 89:, energy: -22297.6851, is temporary added to the population
Structure 121:, energy: -22297.6849, is temporary added to the population
Structure 84:, energy: -22297.6844, is temporary added to the population
Structure 109:, energy: -22297.6844, is temporary added to the population (edited)

as you can see, these are basically two energy ranges. Looking at the geometries, it confirms that these are two conformations only.

What I understand from the fafoom output file, it it performs the xyz comparison only before the relaxation occurs. I see few instances that the relaxation was discarded because "Geomerty fine, but the structure already known". However, once the molecule is relaxed, there is no sign that the comparison occurs once again. And the molecule is added to the population, the least stable removed, and then the pop's get saturated by one-two conformers.

output.txt
last_population.zip