jinnzest/genetic-algorithm-scala

Researching of applying different performance optimizations while implementing a genetic algorithm. Master branch contains initial unoptimized version while Develop branch contains few steps of optimization. https://github.com/jinnzest/genetic-algorithm-rust follows the same approach but using Rust lang.

Researching of applying different performance optimizations while implementing a genetic algorithm. Optimizations will be applied step by step and each improvement will be measured and compared to initial exec time. The initial version is written to follow functional programming idioms as much as possible. One of the goals of the researching is to understand how functional style of programming in scala affects performance.

CPU of the computer used to run measurement: Intel(R) Core(TM) i9-10900K CPU @ 3.70GHz.

OS: headless linux, version 6.1.77-2-MANJARO,

JDK: openjdk version "23-ea" 2024-09-17

To try it yourself run: sbt clean assembly && java -jar ./target/scala-2.13/genetic-algorithm.

Optimization steps and measurements:

1. Without optimization it runs 100000 generations for about 5797327 ms (~ 96 minutes)
2. Array[Gen] is replaced by Array[Long], : 99853 ms, ~ 58 times faster than initial one
3. Gray code is optimized: 64225 ms, ~ 90 times faster than initial one
4. Move all objects to pools to reuse them: 29444 ms, ~196 times faster than initial one
5. Replace Random by ThreadLocalRandom: 23844 ms, ~243 times faster than initial one