def hearthstone_tester1():
createConfig("experiments", "best_deck", "random", 1)
opponents = 50
population = init(opponents, "Mage")
print(population[0])
wins = 0
for opponent in population:
pytohs("random", opponent.get_class(), opponent.get_deck())
system("gradlew runSim") #run simulation
#if getResults("config")["P0"] == 1:
# wins += 1
fp = open("experiments/experiments.hsres")
text = fp.read()
# print("run")
while text.find('winner":') == -1:
system(
"gradlew runSim --configure-on-demand --no-rebuild --offline")
fp = open("experiments/experiments.hsres")
text = fp.read()
if text[text.find('winner":') + 8] == "0":
wins += 1
fp.close()
fit = wins / opponents
print(fit)
def hearthstone_GA_single_class():
createConfig("experiments", "protagonist", "antagonist", 1)
pop_size = 6
population = init(pop_size, "Mage")
print(population[0])
fitness_all(population) #give each member of the population a fitness
# population = initialization(pop_size)
# best_fitness = 0
# best_model = None
# fitness_evals = 0
generation = 0
#fitnesses = []
#each iteration of this loop is a generation
while generation < 10:
print(generation)
# save(population, generation)
generation += 1
population = selection(population)
#mutate population
for element in population:
if random.random() < 0.8:
mutate(element)
population.sort(key=attrgetter('fitness'), reverse=True)
print(population[0])
print("Generation:", generation)
def hearthstone_tester2():
createConfig("experiments", "best_deck", "random", 1)
# population.append(init(pop_size, "Warlock"))
# population.append(init(pop_size, "Mage"))
# population.append(init(pop_size, "Druid"))
# population.append(init(pop_size, "Hunter"))
# population.append(init(pop_size, "Paladin"))
# population.append(init(pop_size, "Priest"))
# population.append(init(pop_size, "Rogue"))
# population.append(init(pop_size, "Shaman"))
# population.append(init(pop_size, "Warrior"))
opponents = 40
types = [
"Warlock", "Druid", "Hunter", "Paladin", "Priest", "Rogue", "Shaman",
"Warrior"
] #8 types
population = []
for my_type in types:
sub_pop = init(5, my_type)
for model in sub_pop:
population.append(model)
print(population[0])
wins = 0
for opponent in population:
pytohs("random", opponent.get_class(), opponent.get_deck())
system("gradlew runSim") #run simulation
#if getResults("config")["P0"] == 1:
# wins += 1
fp = open("experiments/experiments.hsres")
text = fp.read()
# print("run")
while text.find('winner":') == -1:
system(
"gradlew runSim --configure-on-demand --no-rebuild --offline")
fp = open("experiments/experiments.hsres")
text = fp.read()
if text[text.find('winner":') + 8] == "0":
wins += 1
fp.close()
fit = wins / opponents
print(fit)
def hearthstone_GA():
createConfig("experiments", "protagonist", "antagonist", 5)
pop_size = 10
population = init(pop_size, "Mage")
generation = 0
print("Generation: " + str(generation))
calculateFitness(
population) # give each member of the population a fitness
print(str(population[0]))
# population = initialization(pop_size)
# best_fitness = 0
# best_model = None
# fitness_evals = 0
fitnesses = []
# each iteration of this loop is a generation
while generation < 10:
print("Generation: " + str(generation))
# save(population, generation)
generation += 1
# fitness_evals += 100#each generation performs 100 fitness evaluations
# cross over population
population = crossoverPopulation(population)
# mutate population
for element in population:
if random.random() < 0.8:
mutate(element)
population.sort(key=attrgetter('fitness'), reverse=True)
bestIndividual = population[0]
pytohs(bestIndividual.get_class(), bestIndividual.get_class(),
bestIndividual.get_deck())
print(
str(bestIndividual.get_class()) +
"'s best individual has a fitness of: " +
str(bestIndividual.get_fitness()))
print("Generation:", generation)
def hearthstone_GA():
createConfig("experiments", "protagonist", "antagonist", 1)
pop_size = 5
# warlordPop = init(pop_size, "warlord")
# magePop = init(pop_size, "mage")
# druidPop = init(pop_size, "druid")
# hunterPop = init(pop_size, "hunter")
# paladinPop = init(pop_size, "paladin")
# priestPop = init(pop_size, "priest")
# roguePop = init(pop_size, "rogue")
# shamanPop = init(pop_size, "shaman")
# warriorPop = init(pop_size, "warrior")
#
# population = {
# "warlord": warlordPop,
# "druid": druidPop,
# "mage": magePop,
# "hunter": hunterPop,
# "paladin": paladinPop,
# "priest": priestPop,
# "rogue": roguePop,
# "shaman": shamanPop,
# "warrior": warriorPop
# }
population = []
population.append(init(pop_size, "Warlock"))
population.append(init(pop_size, "Mage"))
population.append(init(pop_size, "Druid"))
population.append(init(pop_size, "Hunter"))
population.append(init(pop_size, "Paladin"))
population.append(init(pop_size, "Priest"))
population.append(init(pop_size, "Rogue"))
population.append(init(pop_size, "Shaman"))
population.append(init(pop_size, "Warrior"))
# warlordPop = init(pop_size, "warlord")
# magePop = init(pop_size, "mage")
# druidPop = init(pop_size, "druid")
# hunterPop = init(pop_size, "hunter")
# paladinPop = init(pop_size, "paladin")
# priestPop = init(pop_size, "priest")
# roguePop = init(pop_size, "rogue")
# shamanPop = init(pop_size, "shaman")
# warriorPop = init(pop_size, "warrior")
calculateFitness(population) #give each member of the population a fitness
print(str(population[0][0]))
# population = initialization(pop_size)
# best_fitness = 0
# best_model = None
# fitness_evals = 0
generation = 0
fitnesses = []
#each iteration of this loop is a generation
while generation < 10:
print(generation)
# save(population, generation)
generation += 1
# fitness_evals += 100#each generation performs 100 fitness evaluations
#cross over population
population = crossoverPopulation(population)
#mutate population
for subPopulation in population:
for element in subPopulation:
if random.random() < 0.8:
mutate(element)
#
# index = 0
# while index < pop_size:
# model = population[index]
# roll = random.random()
# #80% chance of mutation for each member of the population
# if roll < 0.80:
# mutate(model)
# index += 1
#evaluate and find best model
"""
models_fitness = []
for model in population:
models_fitness.append((model,fitness(model)))
models_fitness.sort(key=itemgetter(1), reverse=True)
fitnesses.append(models_fitness[0][1])
if models_fitness[0][1] > best_fitness:
best_fitness = models_fitness[0][1]
best_model = models_fitness[0][0]"""
#after termination condition, display best model, best fitness of said model, how many generations ran, and how many fitness evaluations
# print(best_model)
# print(best_fitness)
for subPopulation in population:
subPopulation.sort(key=attrgetter('fitness'), reverse=True)
bestIndividual = subPopulation[0]
for element in subPopulation:
print(str(element.get_deck()) + str(element.get_fitness()))
print(str(bestIndividual))
pytohs(
bestIndividual.get_class(), bestIndividual.get_class(),
bestIndividual.get_deck()
) #outputs the best performing individual to heroclass.hsdeck to analyse
print(
str(bestIndividual.get_class()) +
"'s best individual has a fitness of: " +
str(bestIndividual.get_fitness()))
print("Generation:", generation)