simpsus / pygemtd Goto Github PK

to enable a first naive approach to the maze of optimal length, we need a function that returns the fitness (i.e. the length of the path) for a grid. In pseudocode:

fitness(grid):
  result = 0
  for pair of waypoints:
    result += A_star(pair)
  return result

https://stackoverflow.com/a/21271416/6068036

Let's just share ideas about longest path strategies that may inspire algorithms, as the gradient-free methods (e.g. genetic algorithm) doesn't seem promising for this type of problem.

• https://github.com/techwithtim/Tower-Defense-Game
• https://github.com/dilumich/Tower-Defense-Game
• https://www.pygame.org/tags/towerdefense

Design question: at the moment, a creep notices it is breached and sets a flag. The game needs to process a breach (deduct live, etc), but the game does not know the creeps. The wave knows the creeps and can detect a (newly) breached creep.

Option one: flag in the wave. negative: double flagging and twice detection (wave and game)
Option two: register a callback (in the game) with the creep. pro: would be an active call. negative: currently the creep generator is a lambda, so is another thing to do after creation, and the wave does not know the game, so the game would have to register the callback at the wave as well. again, double passing.
Option three: ??

from here

https://github.com/hjweide/a-star

to make the evaluation/fitness function ready for large scale comptuation

100x100 is way too big for anything ... especially for any algorithm trying to process it

maybe 40x40?

https://stackoverflow.com/a/41336876/6068036