I read about the Demons of Cyclic Space in the Computer Recreations column of Scientific American written by A.K. Dewdney. The article described a simple cellular automata. I found the idea a bit fascinating so I spent some time and implemented a simple version on my PC. I kept writing new versions as I upgrade my computer but sometime around 1995 I put it aside and haven’t touched it in a long time.
Recently I was digging around old disks and backup files and I came across the Windows version I wrote using MFC and Visual C++. Surprisingly it worked without changes and is so fast that I can’t watch the changes as they occur.
I think it’s time to dust off my old friend and experiment some more. What happens if I increase or decrease the number of cell types? What about restricting growth in certain directions? How would giving one type of cell a higher starting population change the outcome?
Below the cut I’ve placed some images of the simulation as it runs. It is a pale imitation because you can’t see the transformation from noise to ordered little cells marching along.
- Location:98004
- Music:Kansas - Two Cents Worth


Comments
In this simulator I created rules for Life (of course), knight life (life rules but the neighborhood are the 8 cells a knights move away from the current cell instead of the 8 adjacent cells) and wireworld (an automata which mimics electricity along wires and which can have or/and/nor/xor etc gates built out of it).
I still have the source code around here somewhere, but the X11 APIs have changed enough (and my code was crufty enough in college) that it no longer compiles.
I don't know the rules for 'demons' however, but it'd be semi interesting to see if I could fix up Xcella (the simulator) and run the demon rules under it :)