version 1.0.7

• Increased the upper limit of Pool Size to 100

• Replaced Reap Count with “Reap Rate,” which specifies the value as a percentage

• Added parameters for thinning matchups: “Match Range” and “Match Density”

• Fixed an issue where the number of wins for genomes marked as “retained” was being reset

The enemy in this game is “loss of diversity.” When diversity is lost, not only does the simulation become monotonous, but it also negatively impacts the search for optimal solutions.

In previous versions, implementations such as island models and self-incompatibility mechanisms were used to help maintain diversity. In V1.0.7, however, a more direct approach was introduced: the number of genomes that can be maintained within a single biome can now be increased.

This makes it easier to preserve diversity within the pool, but at the same time, the number of pairwise matches grows as O(N^2), causing a sharp increase in computation time per generation.

To address this, we introduced matchup thinning parameters: Match Range and Match Density. Matches defined by Match Range, which occur between nearby ranks, are expected to improve local optimization efficiency. On the other hand, additional random matches introduced by Match Density are less efficient, but help prevent premature convergence and contribute to maintaining diversity.

However, no one yet knows what the optimal values for Pool Size, Match Range, and Match Density are. This game is also an optimization problem left for the player to solve.