def steepest_ascent_hill_climbing_with_replacement(tasks,
graph,
skill_set,
iteration=1000,
extra_arg=100):
individual = create_individual(tasks, skill_set)
#individual = seeded_create_individual(tasks, graph, skill_set)
best_individual = individual
for i in range(iteration):
new_individual = mutate(graph, skill_set, individual)
for j in range(1, extra_arg):
w_individual = mutate(graph, skill_set, individual)
if fitness(graph, w_individual) < fitness(graph, new_individual):
new_individual = w_individual
# print("Iteration:", (i+1), fitness(graph, new_individual))
if fitness(graph, new_individual) < fitness(graph, best_individual):
best_individual = new_individual
individual = new_individual
'''
print("Best Individual:")
print("Tasks: ", best_individual["tasks"])
print("Team: ", best_individual["team"])
print("Fitness: ", fitness(graph, best_individual))
'''
return best_individual
def plain_ga(tasks, graph, skill_set, iteration=1000, pop_size=100):
P = initialize(tasks, skill_set, pop_size)
#P = seeded_initialize(tasks,graph,skill_set,pop_size)
best_individual = None
for i in range(iteration):
for p in P:
if best_individual is None or fitness(graph, p) < fitness(
graph, best_individual):
best_individual = p
Q = []
for j in range(pop_size // 2):
individual1 = tournament_selection(graph, P, tournament_size=7)
individual2 = tournament_selection(graph, P, tournament_size=7)
child1, child2 = crossover(skill_set, individual1, individual2)
child1 = mutate(graph, skill_set, child1)
child2 = mutate(graph, skill_set, child2)
Q.append(child1)
Q.append(child2)
P = Q
'''
print("Best Individual:")
print("Tasks: ", best_individual["tasks"])
print("Team: ", best_individual["team"])
print("Fitness: ", fitness(graph, best_individual))
'''
return best_individual
def execute_single_state(graph,
skill_set,
algorithm,
sample=100,
iteration=1000,
task_size=6):
all_tasks = []
for i in range(sample):
tasks = generate_random_tasks(skill_set, task_size)
all_tasks.append(tasks)
results = []
for i in range(sample):
result = algorithm(all_tasks[i], graph, skill_set, iteration)
results.append(result)
total_fitness = 0.0
total_team_size = 0.0
for i in range(sample):
total_fitness += fitness(graph, results[i])
total_team_size += len(set(results[i]["team"]))
avg_fitness = total_fitness / sample
avg_team_size = total_team_size / sample
print("Avg. Team Size: ", avg_team_size)
print("Avg. Fitness: ", avg_fitness)
return avg_team_size, avg_fitness
def hill_climbing(tasks, graph, skill_set, iteration=1000):
best_individual = create_individual(tasks, skill_set)
#best_individual = seeded_create_individual(tasks, graph, skill_set)
for i in range(iteration):
new_individual = mutate(graph, skill_set, best_individual)
#print("Iteration:", (i+1), fitness(graph, new_individual))
if fitness(graph, new_individual) < fitness(graph, best_individual):
best_individual = new_individual
'''
print("Best Individual:")
print("Tasks: ", best_individual["tasks"])
print("Team: ", best_individual["team"])
print("Fitness: ", fitness(graph, best_individual))
'''
return best_individual
def random_search(tasks, graph, skill_set, iteration=1000):
individual = create_individual(tasks, skill_set)
#individual = seeded_create_individual(tasks, graph, skill_set)
best_individual = individual
for i in range(iteration):
individual = create_individual(tasks, skill_set)
#individual = seeded_create_individual(tasks, graph, skill_set)
#print("Iteration:", (i+1), fitness(graph, individual))
if fitness(graph, individual) < fitness(graph, best_individual):
best_individual = individual
print("Best Individual:")
print("Tasks: ", best_individual["tasks"])
print("Team: ", best_individual["team"])
print("Fitness: ", fitness(graph, best_individual))
return best_individual
def elitist_ga(tasks,
graph,
skill_set,
iteration=1000,
pop_size=100,
elites_size=20):
P = initialize(tasks, skill_set, pop_size)
#P = seeded_initialize(tasks,graph,skill_set,pop_size)
best_individual = None
for i in range(iteration):
all_fitness = []
for p in P:
fp = fitness(graph, p)
all_fitness.append(fp)
if best_individual is None or fp < fitness(graph, best_individual):
best_individual = p
elites = numpy.array(all_fitness)
elites_index = numpy.argsort(elites)
elites_index = elites_index[:elites_size]
Q = []
for j in range(elites_size):
Q.append(P[elites_index[j]])
for j in range((pop_size - elites_size) // 2):
individual1 = tournament_selection(graph, P, tournament_size=7)
individual2 = tournament_selection(graph, P, tournament_size=7)
child1, child2 = crossover(skill_set, individual1, individual2)
child1 = mutate(graph, skill_set, child1)
child2 = mutate(graph, skill_set, child2)
Q.append(child1)
Q.append(child2)
P = Q
'''
print("Best Individual:")
print("Tasks: ", best_individual["tasks"])
print("Team: ", best_individual["team"])
print("Fitness: ", fitness(graph, best_individual))
'''
return best_individual