def main():
initialize(parse_argv())
startTime = datetime.now()
print("START:")
current_gen = 0
staling = 0
prevAverage = 1
## Create initial population and calculate initial fitness
population = permutation(pop_size)
useHeuristic = False
useSwap = False
staled = False
heuristicThreshold = len(population) * 0.25
swapThreshold = len(population) * 5
stalingThreshold = len(population) * 10
currentThreshold = heuristicThreshold
while current_gen < gen_limit and not staled:
# Switch between tourney and multi-winner-tourney here
parents = tournament_selection(population, mating_pool_size,
tournament_size)
#parents = multi_winner_tourney_selection(population, mating_pool_size, mw_tournament_size, mw_tournament_winners)
r.shuffle(parents)
offspring = []
i = 0
while len(offspring) < mating_pool_size:
## Crossover
if r.random() < xover_rate:
xover_offspring = order_crossover(
parents[i].get_city_ids(), parents[i + 1].get_city_ids())
off0 = Route(xover_offspring[0])
off1 = Route(xover_offspring[1])
else:
off0 = parents[i].create_copy()
off1 = parents[i + 1].create_copy()
## mutation
if r.random() < mut_rate:
if (useSwap):
off0 = swap_mutation(off0.get_city_ids())
elif (useHeuristic):
off0 = heuristic_swap(off0)
else:
off0 = scramble_mutation(off0.get_city_ids())
if r.random() < mut_rate:
if (useSwap):
off1 = swap_mutation(off1.get_city_ids())
elif (useHeuristic):
off1 = heuristic_swap(off1)
else:
off1 = scramble_mutation(off1.get_city_ids())
## Add new offspring
offspring.append(off0)
offspring.append(off1)
i += 2
population = survivor_selection(population, offspring)
fitness = [x.get_fitness() for x in population]
average = sum(fitness) / len(fitness)
bestFit = min(fitness)
summaryAvgList.append(average)
summaryBestList.append(bestFit)
print("Generation ", current_gen)
print("Best fitness: ", bestFit)
print("Average fitness: ", average)
if (average / prevAverage > (currentThreshold - 1) / currentThreshold
and average / prevAverage <
(currentThreshold + 1 / currentThreshold)):
staling += 1
else:
staling = 0
if (staling == staling_limit):
if not useHeuristic:
useHeuristic = True
currentThreshold = swapThreshold
print("Changed from scramble to heuristic swap!")
elif not useSwap:
useSwap = True
currentThreshold = stalingThreshold
print("Changed from heuristic to random swap!")
else:
staled = True
staling = 0
prevAverage = average
current_gen += 1
# print the final best solution(s)
if (staled):
print("Population staled!")
else:
print("Gen limit reached!")
endTime = datetime.now()
deltaT = endTime - startTime
print("Completed in:", deltaT)
# output the graphs
summary.plot_avg_best_fit(summaryAvgList, summaryBestList)
summary.plot_route(population[fitness.index(min(fitness))])
summary.visualize()
def run(filename):
startTime = datetime.now()
print("START:")
current_gen = 0
staling = 0
prevAverage = 1
## Create initial population and calculate initial fitness
population = permutation(pop_size)
useHeuristic = False
useSwap = False
staled = False
heuristicThreshold = len(population)
swapThreshold = len(population) * 5
stalingThreshold = len(population) * 10
currentThreshold = heuristicThreshold
while current_gen < gen_limit and not staled:
# Switch between tourney and multi-winner-tourney here
parents = tournament_selection(population, mating_pool_size,
tournament_size)
#r.shuffle(parents)
offspring = []
i = 0
while len(offspring) < mating_pool_size:
## Crossover
if r.random() < xover_rate:
xover_offspring = order_crossover(
parents[i].get_city_ids(), parents[i + 1].get_city_ids())
off0 = Route(xover_offspring[0])
off1 = Route(xover_offspring[1])
else:
off0 = parents[i].create_copy()
off1 = parents[i + 1].create_copy()
## mutation
if r.random() < mut_rate:
if (useSwap):
off0 = swap_mutation(off0.get_city_ids())
elif (useHeuristic):
off0 = heuristic_swap(off0)
else:
off0 = scramble_mutation(off0.get_city_ids())
if r.random() < mut_rate:
if (useSwap):
off1 = swap_mutation(off1.get_city_ids())
elif (useHeuristic):
off1 = heuristic_swap(off1)
else:
off1 = scramble_mutation(off1.get_city_ids())
offspring.append(off0)
offspring.append(off1)
i += 2
population = survivor_selection(population, offspring)
fitness = [x.get_fitness() for x in population]
average = sum(fitness) / len(fitness)
bestFit = min(fitness)
if (current_gen % 50 == 0):
print("Generation ", current_gen)
print("Best fitness: ", bestFit)
print("Average fitness: ", average)
if (average / prevAverage > (currentThreshold - 1) / currentThreshold
and average / prevAverage <
(currentThreshold + 1 / currentThreshold)):
staling += 1
else:
staling = 0
if (staling == staling_limit):
if not useHeuristic:
useHeuristic = True
currentThreshold = swapThreshold
print("Changed from scramble to heuristic swap!")
elif not useSwap:
useSwap = True
currentThreshold = stalingThreshold
print("Changed from heuristic to random swap!")
else:
staled = True
staling = 0
prevAverage = average
current_gen += 1
# print the final best solution(s)
if (staled):
print("Population staled!")
endTime = datetime.now()
deltaT = endTime - startTime
data = {
"file": filename,
"elapsed_time": str(deltaT),
"generation": current_gen,
"best_fitness": bestFit,
"avg_fitness": average
}
return data
else:
print("Gen limit reached!")
endTime = datetime.now()
deltaT = endTime - startTime
data = {
"file": filename,
"elapsed_time": str(deltaT),
"generation": current_gen,
"best_fitness": bestFit,
"avg_fitness": average
}
return data
