ZED Run is, and will always be, a breeding game with a racing test on the side. There are very basic and fundamental steps in genetic algorithms that stables should know. We have an initial population of genesis ZED horses, we then calculate fitness using racing, we select fit individuals to then breed together with crossover of 1s and 0s and mutation to make the foal unique. Then, we start the process over. This article takes a deeper look genetic algorithms, and why we have multiple bloodline populations within the ZED ecosystem.
Don’t be Premature
A problem that genetic algorithms can face is one called premature convergence (link). This occurs when a large percentage of the population contains the same value for a particular gene, i.e. every horse preferring sunny days and straight tracks in their 1s and 0s DNA. There are many techniques that can be utilized to help prevent premature convergence such as a uniform crossover and incest prevention, but ZED leverages the popular technique of multiple populations (link).
Bloodlines
ZED Run has 4 bloodlines built into the genetic algorithm initial population. We have Nakamotos, Szabos, Finneys, and Buterins horses that act as sub-populations. The reason we do not have “half bloodlines” is because the genetic profile of each bloodline has unique traits to prevent premature convergence (link). For this reason, it is likely beneficial to cross bloodlines to have diverse DNA. However, crossing bloodlines may make assessing variables in the genetic algorithm more difficult.
Inbreeding
ZED Run was created to mimic the experience of IRL horse racing, and a strong component of thoroughbred horse racing is inbreeding. A genetic study of 500,000 thoroughbreds from 2005 found almost all horses descended from 28 horses and a single super stud (link). A technique that is also commonly used in genetic algorithms to avoid premature convergence is “incest prevention”. This would contradict real life horse breeding and the rules of ZED breeding, which makes the multiple bloodline populations even more important.
Conclusion
In conclusion, genetic algorithms can face the problem of premature convergence. This is when the value of the 1s and 0s within the DNA of the population is too similar. There are many techniques to prevent this, but ZED Run contains multiple populations. These sub-populations are the bloodlines, and have unique profiles to assist moving the progress of the game forward.
Share on facebook
Share on twitter
Share on linkedin
This is not financial advice, but merely my own experiences within the Zed Run. If this information has proven helpful, please consider donating to 0x4370c9D09D4AC084949b4ff101d59DFa75F9B6A3.
Visit our about page to learn more the team and read more in top articles!
Good luck on the on the track, Zed Heads! Any additional information will be on the Zed Community site!
The post Multi-Pop GA appeared first on Zed Gazette.