This began as a Swift port of the wonderful project biosim4 by David R. Miller. I've always been interested in neural networks, but had never really done anything with them, after watching David's video about his project, I was inspired to get a better understanding for how these worked. As an iOS developer, I'm most familiar with Swift, and thought porting the project would be a good way to learn.

I originally considered calling this biosim4-swift to make the connection more obvious, but as I work on it more I've realized I don't want this to be a 1-to-1 port, and would like to have it "grow" in it's own direction.

Up Arrow - Output current genetic diversity Down Arrow - Toggle pause Left Arrow/Right Arrow - Decrement/Increment Steps per Refresh Left Arrow (when paused, and steps per refresh is 1) - Step backwards in history

"Barrier" UI Button - Enable placement of barriers via mouse press. By default these barriers are persisted across generations "Kill" UI Button - Enable deletion of individuals via mouse press "Select" UI Button - Enable selection of individuals via mouse press (selection does nothing at present)

Many values can be tweaked, and these can be found within Parameters.swift

Currently this only works on MacOS, but given that it uses SpriteKit, it should be fairly easy to port to other Apple OSes. Performance is an active area of development, and eventually I might move to something more low level than SpriteKit (ex. Metal)

Note that this is a side-project, so take everything in these plans with a grain of salt.

• Multiplatform (iOS, iPadOS, MacOS)
• Dynamic parameters (ex. change population, challenge etc. at runtime)
• Interacting with grid - drawing barriers and "safe areas" at runtime
• Interacting with individuals
  • View representation of neural net
  • Adjust lifespan (ex. make a specific individual immortal)
• Energy cost/reward (ex. movement takes X units of energy)
• Mutation rate based on proximity to radiation sources
• Heat sources
• Save/load parameters and populations
• Time controls (partially implemented, but currently no way to go back in time)