def testVectorizeRoundRobin(self):
env = _MultiAgentEnvToBaseEnv(lambda v: RoundRobinMultiAgent(2), [], 2)
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0}, 1: {0: 0}})
self.assertEqual(rew, {0: {0: None}, 1: {0: None}})
env.send_actions({0: {0: 0}, 1: {0: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {1: 0}, 1: {1: 0}})
env.send_actions({0: {1: 0}, 1: {1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0}, 1: {0: 0}})
def testVectorizeRoundRobin(self):
env = _MultiAgentEnvToBaseEnv(lambda v: RoundRobinMultiAgent(2), [], 2)
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0}, 1: {0: 0}})
self.assertEqual(rew, {0: {0: None}, 1: {0: None}})
env.send_actions({0: {0: 0}, 1: {0: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {1: 0}, 1: {1: 0}})
env.send_actions({0: {1: 0}, 1: {1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0}, 1: {0: 0}})
def testVectorizeBasic(self):
env = _MultiAgentEnvToBaseEnv(lambda v: BasicMultiAgent(2), [], 2)
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
self.assertEqual(rew, {0: {0: None, 1: None}, 1: {0: None, 1: None}})
self.assertEqual(
dones, {
0: {
0: False,
1: False,
"__all__": False
},
1: {
0: False,
1: False,
"__all__": False
}
})
for _ in range(24):
env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
self.assertEqual(rew, {0: {0: 1, 1: 1}, 1: {0: 1, 1: 1}})
self.assertEqual(
dones, {
0: {
0: False,
1: False,
"__all__": False
},
1: {
0: False,
1: False,
"__all__": False
}
})
env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(
dones, {
0: {
0: True,
1: True,
"__all__": True
},
1: {
0: True,
1: True,
"__all__": True
}
})
# Reset processing
self.assertRaises(
ValueError, lambda: env.send_actions({
0: {
0: 0,
1: 0
},
1: {
0: 0,
1: 0
}
}))
self.assertEqual(env.try_reset(0), {0: 0, 1: 0})
self.assertEqual(env.try_reset(1), {0: 0, 1: 0})
env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
self.assertEqual(rew, {0: {0: 1, 1: 1}, 1: {0: 1, 1: 1}})
self.assertEqual(
dones, {
0: {
0: False,
1: False,
"__all__": False
},
1: {
0: False,
1: False,
"__all__": False
}
})
def testNoResetUntilPoll(self):
env = _MultiAgentEnvToBaseEnv(lambda v: BasicMultiAgent(2), [], 1)
self.assertFalse(env.get_unwrapped()[0].resetted)
env.poll()
self.assertTrue(env.get_unwrapped()[0].resetted)
def testVectorizeBasic(self):
env = _MultiAgentEnvToBaseEnv(lambda v: BasicMultiAgent(2), [], 2)
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
self.assertEqual(rew, {0: {0: None, 1: None}, 1: {0: None, 1: None}})
self.assertEqual(
dones, {
0: {
0: False,
1: False,
"__all__": False
},
1: {
0: False,
1: False,
"__all__": False
}
})
for _ in range(24):
env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
self.assertEqual(rew, {0: {0: 1, 1: 1}, 1: {0: 1, 1: 1}})
self.assertEqual(
dones, {
0: {
0: False,
1: False,
"__all__": False
},
1: {
0: False,
1: False,
"__all__": False
}
})
env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(
dones, {
0: {
0: True,
1: True,
"__all__": True
},
1: {
0: True,
1: True,
"__all__": True
}
})
# Reset processing
self.assertRaises(
ValueError,
lambda: env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}}))
self.assertEqual(env.try_reset(0), {0: 0, 1: 0})
self.assertEqual(env.try_reset(1), {0: 0, 1: 0})
env.send_actions({0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
obs, rew, dones, _, _ = env.poll()
self.assertEqual(obs, {0: {0: 0, 1: 0}, 1: {0: 0, 1: 0}})
self.assertEqual(rew, {0: {0: 1, 1: 1}, 1: {0: 1, 1: 1}})
self.assertEqual(
dones, {
0: {
0: False,
1: False,
"__all__": False
},
1: {
0: False,
1: False,
"__all__": False
}
})
def testNoResetUntilPoll(self):
env = _MultiAgentEnvToBaseEnv(lambda v: BasicMultiAgent(2), [], 1)
self.assertFalse(env.get_unwrapped()[0].resetted)
env.poll()
self.assertTrue(env.get_unwrapped()[0].resetted)
def test_no_reset_until_poll(self):
env = _MultiAgentEnvToBaseEnv(lambda v: BasicMultiAgent(2), [], 1)
self.assertFalse(env.get_sub_environments()[0].resetted)
env.poll()
self.assertTrue(env.get_sub_environments()[0].resetted)
