def test_regenerate_species(self):
model = PrimaVita(log_dir_info="test")
model.regenerate_species()
self.assertEqual(len(model.players), 20)
shutil.rmtree("Players_Data_test")
def test_players_in_env_get_idx(self):
model = PrimaVita(log_dir_info="test",
params_dic={"initial_population": 2})
# Test players nearby
model.players = [
Player(i=0,
log_dir="Players_Data_test",
tob=10,
x=50,
y=500,
energy=200),
Player(i=1,
log_dir="Players_Data_test",
tob=12,
x=48,
y=505,
energy=200)
]
vec, distances, env = model.players_in_env(model.players[0],
get_idx=True)
self.assertTrue(vec[:-1], [2, -5])
self.assertTrue(distances, [5.385164807134504])
self.assertTrue(env, [1])
shutil.rmtree("Players_Data_test")
def test_update_time(self):
model = PrimaVita(log_dir_info="test")
model.update_time()
self.assertEqual(model.time, 0)
shutil.rmtree("Players_Data_test")
def test_check_particles(self):
model = PrimaVita(log_dir_info="test")
model.food_particles = [
Particle(x=64, y=514),
Particle(x=10, y=502),
Particle(x=60, y=514),
Particle(x=12, y=502)
]
model.check_particles()
self.assertEqual(model.food_particles[2], 0)
self.assertEqual(model.food_particles[3], 0)
shutil.rmtree("Players_Data_test")
def test_pad_state(self):
model = PrimaVita(log_dir_info="test")
# Greater than maxlen
padded_state = model.pad_state([4, 3, 2, 1], 2)
self.assertEqual(padded_state, [4, 3])
# Less than maxlen
padded_state = model.pad_state([4, 3], 4)
self.assertEqual(list(padded_state), [4, 3, 0, 0])
# Equal to maxlen
padded_state = model.pad_state([4, 3, 2, 1], 4)
self.assertEqual(padded_state, [4, 3, 2, 1])
shutil.rmtree("Players_Data_test")
def test_food_in_env_no_get_idx(self):
model = PrimaVita(log_dir_info="test",
params_dic={"initial_population": 1})
# Test dead player
model.players = [1]
food_env = model.food_in_env(model.players[0])
self.assertEqual(food_env, -1)
# Test food nearby
model.players = [
Player(i=0,
log_dir="Players_Data_test",
tob=10,
x=50,
y=500,
energy=200)
]
model.food_particles = [Particle(x=64, y=514), Particle(x=45, y=502)]
vec, distances = model.food_in_env(model.players[0])
self.assertEqual(vec, [14, 14, -5, 2])
self.assertEqual(distances, [19.79898987322333, 5.385164807134504])
shutil.rmtree("Players_Data_test")
def test_food_nearby(self):
model = PrimaVita(log_dir_info="test",
params_dic={"initial_population": 1})
# Test dead player
model.players = [1]
food = model.food_nearby(model.players[0])
self.assertEqual(food, -1)
# Test food nearby
model.players = [
Player(i=0,
log_dir="Players_Data_test",
tob=10,
x=50,
y=500,
energy=200)
]
model.food_particles = [Particle(x=64, y=514), Particle(x=45, y=502)]
food = model.food_nearby(model.players[0])
self.assertEqual(food, 0)
shutil.rmtree("Players_Data_test")
def test_initializer(self):
model = PrimaVita(log_dir_info="test")
self.assertTrue(os.path.exists("Players_Data_test"))
self.assertTrue(os.path.exists("Players_Data_test/Embeddings"))
shutil.rmtree("Players_Data_test")
# Testing init method (called from initialiazer)
# Testing regenerate_species inside init method
self.assertEqual(len(model.players), model.initial_population)
def test_write_data(self):
"""
Test write_data method which is used to write to log file once agent dies
"""
model = PrimaVita(log_dir_info="test")
if not os.path.exists(model.log_dir):
os.mkdir(model.log_dir)
player = Player(i=1, log_dir=model.log_dir, tob=10, energy=200)
player.write_data(time=10, alive_count=10)
self.assertTrue(os.path.exists(os.path.join(model.log_dir,
"10-1.npy")))
shutil.rmtree(model.log_dir)
def test_search_mate(self):
model = PrimaVita(log_dir_info="test",
params_dic={"initial_population": 3})
# Test dead player
model.players[0] = 1
player = model.search_mate(model.players[0])
self.assertEqual(player, -1)
# Test find mate in radius
model.players = [
Player(i=0,
log_dir="Players_Data_test",
tob=10,
x=50,
y=500,
energy=200),
Player(i=1,
log_dir="Players_Data_test",
tob=12,
x=48,
y=505,
energy=200),
Player(i=2,
log_dir="Players_Data_test",
tob=12,
x=40,
y=530,
energy=200)
]
model.players[0].is_impotent = False
model.players[0].gender = "Male"
model.players[1].is_impotent = False
model.players[1].gender = "Female"
model.time = 30
player = model.search_mate(model.players[0])
self.assertEqual(player, 1)
shutil.rmtree("Players_Data_test")
def test_get_current_state(self):
model = PrimaVita(log_dir_info="test",
params_dic={"initial_population": 2})
# Test state shape
state, _ = model.get_current_state()
self.assertEqual(state.shape, (2, 21))
# Test stopping condition
model.killed = [0, 1, 2]
model.players = [0, 1, 2]
_, running = model.get_current_state()
self.assertEqual(running, False)
shutil.rmtree("Players_Data_test")
def test_search_enemy(self):
model = PrimaVita(log_dir_info="test",
params_dic={"initial_population": 3})
# Test dead player
model.players[0] = 1
player = model.search_enemy(model.players[0])
self.assertEqual(player, -1)
# Test find opponent in radius
model.players = [
Player(i=0,
log_dir="Players_Data_test",
tob=10,
x=50,
y=500,
energy=200),
Player(i=1,
log_dir="Players_Data_test",
tob=12,
x=40,
y=530,
energy=200),
Player(i=2,
log_dir="Players_Data_test",
tob=12,
x=48,
y=505,
energy=200)
]
player = model.search_enemy(model.players[0])
self.assertEqual(player, 2)
shutil.rmtree("Players_Data_test")
from pygeneses.envs.prima_vita import PrimaVita
params_dic = {
"initial_population": 20,
"state_size": 32,
"speed": 30,
"model_updates": 5
}
model = PrimaVita(mode="human", params_dic=params_dic)
model.run(stop_at=100)
def test_take_action(self):
model = PrimaVita(log_dir_info="test")
shutil.rmtree("Players_Data_test")
from pygeneses.envs.prima_vita import PrimaVita rl_model = PrimaVita(log_dir_info="profile") rl_model.run(stop_at=20)
import os
os.putenv('SDL_VIDEODRIVER', 'fbcon')
os.environ["SDL_VIDEODRIVER"] = "dummy"
from pygeneses.envs.prima_vita import PrimaVita
rl_model = PrimaVita()
def training_loop(n_episodes=1000, max_t=1000, gamma=1.0, print_every=100):
rl_model.init()
states = rl_model.get_current_state()
running = True
while running:
rl_model.update_time()
if(rl_model.time == 20):
running = False
for i, agent in enumerate(rl_model.model.agents):
if(type(rl_model.players[i]) != int):
rl_model.take_action(i, states[i])
states = rl_model.get_current_state()
if(type(states) == int and states == -1):
running = False
training_loop()