andaviaco / abc Goto Github PK

Hello! First of all, thank the writer who really offered us a useful and efficient template of the artificial bee colony. But unfortunately, there are some little issues that are not correct after I confirmed the procedure proposed in the original paper in 2005. So here are the work I did to fix the code, mainly in "swarm.py":

1. During updating the positions of employed and onlooker bees, the fitness should be updated at the same time, which is missed in your code.

   def set_solution(self, food_source, new_solution):
        # greedy strategy is applied
        if np.array_equal(new_solution, food_source.solution):
            food_source.trials += 1
        else:
            food_source.solution = new_solution
            food_source.fitness = self.fitness(new_solution)
            food_source.trials = 0
        return food_source

2. An issue that a dude once pointed out is that one bee may go beyond the search range, which can be solved by a simple function under the NumPy frame-- numpy.clip() to constraint the agents inside the borders.
3. In "employed_bee_satge" and "onlooker_bee_stage" you, unfortunately, missed the return of function "set_solutions", which makes the optimal solution is not satble and far away from the actual one.

Hoped it will be helpful!

In generate_solution function, you may miss validating whether components of a new solution exceeds the limit boundaries

When my account was compromised a spam issue was created in this repo. I sincerely apologize. Cleaning up such issues via script.