1

Im struggling to solve an open problem, namely, when building a cellular automata like Conways GOL, how would one go about implementing teams? Instead of just one Automata, I want two or three competing over the cells that are available.

The real problem is how do you let the Automata decide which team wins a 'battle' when blue cells for example meet red cells, how do we determine which colored cells will eventually overcome the other. Using a coin flip to randomly determine the winner will result in a equilibrium where both colors keep switching cells but it is extremely unlikely one will completely replace the other as every turn they both have a 50/50 chance.

So thats not a solution. How would one go about getting a 'winner' out of 2 contestants who ar basically the same? I was thinking about giving every cell a "power (int)" of 1, and adding up all those power variables (basically counting all same colored cells and comparing that with the other team, whichever team/color has more cells in the game at that point will continue to dominate and eventually win the game, but thats not a solution either.

Can someone point me in the right direction here?

Basically how do you determine a winner when both parties are exactly equal, and it cannot be 50/50 random either, it has to be able to evolve like in Conways Game of Life.

EDIT: This Automaton wont take into account overpopulation (cells dying from too many adjacent neighbours). It is a Cellular Automaton but it will have different rulles than CGOL. For now only expansion from 2 or more neighbours, essentially the cells will try to fill the entire grid with their color.

Thanks for your insight !

-Erik

eriks jud
  • 73
  • 6
  • You could make it so that each cell will either settle on red or blue, depending on its coordinates. E.g. all even coordinates will settle on red, or use a PRNG (with a fixed, per-game seed) to determine whether any given cell will settle on red or blue. Or widen this a bit and have regions of adjacent cells which favour a particular color? E.g. the bottom right corner is more favourable to red. Again you could come up with a pseudo-random way of creating these regions. – canton7 Apr 06 '22 at 10:13
  • Or you could pick a metric such as, when two colors compete for a cell, the color which has the largest connected "army" behind it wins, or a connected cell of the same color the furthest distance away, or the largest number of cells of the same color within some radius – canton7 Apr 06 '22 at 10:16
  • @canton7 Your second solution, is what im thinking about aswell, but I dont expect it to work in this iteration, I feel as the solution is somewhere in that direction though. A problem I see with this solution is that whichever team/color started with the most cells at the beginning of the game will win 99% of the time. – eriks jud Apr 06 '22 at 10:19
  • 2
    I think this is a case of Try It And See. And then tweak, iterate, and experiment. You don't tend to get large connected regions in CGOL: things have a habit of dying. So I'm not really sure what "win" means here – canton7 Apr 06 '22 at 10:22
  • Sorry @canton7 , I realise my question was not a 100% clear so I have added an edit! Basically cells wont die like in CGOL, I might add a rule where cells die, but for now they only 'die' when another colored cell replaces them. – eriks jud Apr 06 '22 at 10:31
  • Are there separate rules for each color? Or could it be like: A cell with 1 red neighbor and 2 blue neighbors becomes red in the next generation (and vice versa)? – Heiko Theißen Jun 12 '22 at 12:46

0 Answers0