def test_ConnectivityConverter(self):
converter = ConnectivityConverter(self.env.action_space)
converter.init_converter()
converter.seed(0)
assert np.all(converter.subs_ids == np.array([
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3,
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4,
4, 4, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 8, 8, 8, 8, 8,
8, 8, 8, 8, 8, 12, 12, 12, 12, 12, 12
]))
assert len(converter.obj_type) == converter.n
assert len(set(converter.obj_type)) == converter.n
assert converter.pos_topo.shape[0] == converter.n
assert len(set([tuple(sorted(el))
for el in converter.pos_topo])) == converter.n
coded_act = np.random.rand(converter.n)
pred = converter._compute_disagreement(coded_act, np.ones(converter.n))
assert np.abs((converter.n - coded_act.sum()) / converter.n -
pred) <= self.tol_one
pred = converter._compute_disagreement(coded_act,
np.arange(converter.n))
assert np.abs(coded_act.sum() / converter.n - pred) <= self.tol_one
# and not test i can produce an action that can be implemented
act = converter.convert_act(encoded_act=coded_act)
obs, reward, done, info = self.env.step(act)
# test sample
obs = self.env.reset()
act = converter.sample()
obs, reward, done, info = self.env.step(act)
def test_max_sub_changed(self):
for ms_sub in [1, 2, 3]:
converter = ConnectivityConverter(self.env.action_space)
converter.init_converter(max_sub_changed=ms_sub)
converter.seed(0)
coded_act = np.random.rand(converter.n)
# and not test i can produce an action that can be implemented
act = converter.convert_act(encoded_act=coded_act)
lines_impacted, subs_impacted = act.get_topological_impact()
assert np.sum(
subs_impacted
) == ms_sub, "wrong number of substations affected. It should be {}".format(
ms_sub)
obs, reward, done, info = self.env.step(act)
# test sample
obs = self.env.reset()
act = converter.sample()
lines_impacted, subs_impacted = act.get_topological_impact()
assert np.sum(
subs_impacted
) == ms_sub, "wrong number of substations affected. It should be {}".format(
ms_sub)
obs, reward, done, info = self.env.step(act)
def test_ConnectivityConverter(self):
converter = ConnectivityConverter(self.env.action_space)
converter.seed(0)
converter.init_converter()
res = np.array([
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3,
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4,
4, 4, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5,
5, 5, 5, 5, 5, 5, 5, 5, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 12, 12, 12,
12, 12, 12
])
assert np.array_equal(converter.subs_ids, res)
assert len(converter.obj_type) == converter.n
assert len(set(converter.obj_type)) == converter.n
assert converter.pos_topo.shape[0] == converter.n
assert len(set([tuple(sorted(el))
for el in converter.pos_topo])) == converter.n
n_trial = 100
preds = np.zeros(n_trial)
for i in range(n_trial):
coded_act = np.random.rand(converter.n) * 2. - 1.
preds[i] = converter._compute_disagreement(coded_act,
np.ones(converter.n))
# check the formula is properly implemented in case of "everything connected together"
assert np.abs(preds[i] - 0.5 *
(1 - np.mean(coded_act))) <= self.tol_one
# check that on average over "a lot" of stuff the distance to the "everything connected" is 0.5
assert np.abs(np.mean(preds) - 0.5) <= 1 / np.sqrt(n_trial)
# and not test i can produce an action that can be implemented
act = converter.convert_act(encoded_act=coded_act)
obs, reward, done, info = self.env.step(act)
# test sample
obs = self.env.reset()
act = converter.sample()
obs, reward, done, info = self.env.step(act)
def test_bug_in_doc(self):
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
env = make("rte_case14_realistic", test=True)
converter = ConnectivityConverter(env.action_space)
# it's a good practice to seed the element that can be, for reproducibility
converter.seed(0)
# to avoid creating illegal actions affecting more than the allowed number of parameters
converter.init_converter(
max_sub_changed=env.parameters.MAX_SUB_CHANGED)
encoded_act = np.zeros(converter.n)
encoded_act[
0] = 1 # i want to connect "line_ex id 0" and the "line_or id 2"
encoded_act[
1] = -1 # i don't want to connect "line_ex id 0" and the "line_or id 3"
encoded_act[
2] = -1 # i don't want to connect "line_ex id 0" and the "line_or id 4"
encoded_act[
3] = -1 # i don't want to connect "line_ex id 0" and the "gen id 0"
encoded_act[
4] = 1 # i want to connect "line_ex id 0" and the "load id 0"
# and now retrieve the corresponding grid2op action:
grid2op_act = converter.convert_act(encoded_act)
assert converter.last_disagreement == 0.
assert np.array_equal(grid2op_act.set_bus[3:9], [1, 1, 2, 2, 2, 1])
# second way to express the same action
encoded_act2 = np.zeros(converter.n)
encoded_act2[
0] = 1 # i want to connect "line_ex id 0" and the "line_or id 2"
encoded_act2[
4] = 1 # i want to connect "line_ex id 0" and the "load id 0"
encoded_act2[
9] = 1 # i want to connect "line_or id 3" and the "line_or id 4"
encoded_act2[
10] = 1 # i want to connect "line_or id 3" and the "gen id 0"
encoded_act2[
14] = -1 # i don't want to connect "gen id 0" and the "load id 0"
# and now retrieve the corresponding grid2op action:
grid2op_act2 = converter.convert_act(encoded_act2)
assert converter.last_disagreement == 0.
assert np.array_equal(grid2op_act2.set_bus[3:9], [1, 1, 2, 2, 2, 1])
# trick it: i don't specified enough constraints (used to be infinite loop)
encoded_act3 = np.zeros(converter.n)
encoded_act3[
0] = 1 # i want to connect "line_ex id 0" and the "line_or id 2"
encoded_act3[
4] = 1 # i want to connect "line_ex id 0" and the "load id 0"
encoded_act3[
9] = 1 # i want to connect "line_or id 3" and the "line_or id 4"
encoded_act3[
10] = 1 # i want to connect "line_or id 3" and the "gen id 0"
grid2op_act3 = converter.convert_act(encoded_act3)
assert converter.last_disagreement == 0.
assert np.array_equal(grid2op_act3.set_bus[3:9], [1, 1, 1, 1, 1, 1])
size_ = converter.n
# trick (the compute_disagreement function) in another way: not all components are set
missing = np.ones(converter.n)
missing[3] = 0 # encodes for line_ex id O
disag0 = converter._compute_disagreement(encoded_act3, missing)
# 2 constraints not met because the line_ex id O is not set in the action
assert abs(disag0 - 2. / size_) <= self.tol_one
missing2 = np.ones(converter.n)
missing2[8] = 0 # encodes for load id 0
disag2 = converter._compute_disagreement(encoded_act3, missing2)
# 1 constraints not met because the load id 0 O is not set in the action
assert abs(disag2 - 1. / size_) <= self.tol_one
missing3 = np.ones(converter.n)
missing3[3:9] = 0 # all constraints not met
disag3 = converter._compute_disagreement(encoded_act3, missing3)
# one component not set in the "action candidate" among 4 constraints
assert abs(disag3 - 4. / size_) <= self.tol_one
def test_can_make_action(self):
converter = ConnectivityConverter(self.env.action_space)
converter.init_converter(
max_sub_changed=self.env.parameters.MAX_SUB_CHANGED)
converter.seed(0)
with self.assertRaises(Exception):
# one too shot
tmp = np.zeros(converter.n - 1)
converter.convert_act(tmp)
with self.assertRaises(Exception):
# one too long
tmp = np.zeros(converter.n + 1)
converter.convert_act(tmp)
with self.assertRaises(Exception):
# one too low
tmp = np.zeros(converter.n)
tmp[3] = -1.1
converter.convert_act(tmp)
with self.assertRaises(Exception):
# one too high
tmp = np.zeros(converter.n)
tmp[3] = 1.1
converter.convert_act(tmp)
size_ = converter.n
# test do nothing gives do nothing indeed
dn_enc = converter.do_nothing_encoded_act()
glop_dn_act = converter.convert_act(dn_enc)
assert glop_dn_act == self.env.action_space()
# encode the topology [1,1 ,2,2] at sub 12, meaning: line_ex 9 / line_ex 13 together,
# and line or 14 / load 9 together
complex_act = 1.0 * dn_enc
complex_act[84] = 1. # line ex 9 and line ex 13 together
complex_act[85] = -1. # line ex 9 and line or 14 not together
complex_act[86] = -1. # line ex 9 and load 9 not together
glop_act = converter.convert_act(complex_act)
aff_line, aff_sub = glop_act.get_topological_impact()
assert np.sum(aff_line) == 0
assert np.sum(aff_sub) == 1
assert aff_sub[12]
assert glop_act.line_ex_set_bus[9] != 0
assert glop_act.line_ex_set_bus[13] != 0
assert glop_act.line_or_set_bus[14] != 0
assert glop_act.load_set_bus[9] != 0
assert glop_act.line_ex_set_bus[9] == glop_act.line_ex_set_bus[13]
assert glop_act.line_ex_set_bus[9] != glop_act.line_or_set_bus[14]
assert glop_act.line_ex_set_bus[9] != glop_act.load_set_bus[9]
assert glop_act.line_or_set_bus[14] == glop_act.load_set_bus[9]
# encode the topology [1,1 ,2,2] at sub 12, meaning: line_ex 9 / line_ex 13 together,
# and line or 14 / load 9 together but in a "soft" manner
complex_act = 1.0 * dn_enc
complex_act[84] = 0.8 # line ex 9 and line ex 13 together
complex_act[85] = -0.9 # line ex 9 and line or 14 not together
complex_act[86] = -0.9 # line ex 9 and load 9 not together
glop_act2 = converter.convert_act(complex_act)
assert abs(converter.last_disagreement - 0.5 *
(0.2 + 0.1 + 0.1) / size_) <= self.tol_one
assert glop_act == glop_act2
# now tricky stuff, such that the greedy do not work and i need to explore a bit
# line_ex 9 / line_ex 13 together and line or 14 / load 9 together
complex_act = 1.0 * dn_enc
complex_act[84] = 0.6 # line ex 9 and line ex 13 together
complex_act[85] = -0.9 # line ex 9 and line or 14 not together
complex_act[86] = -0.9 # line ex 9 and load 9 not together
complex_act[
87] = 0.61 # "line_ex id 13" and the "line_or id 14" together
complex_act[88] = -0.2 # "line_ex id 13" and the "load id 9" together
complex_act[
89] = 0. # "line_or id 14" and the "load id 9" no preferences
glop_act3 = converter.convert_act(complex_act)
# this gives : [ 1 2 2 2 ], which is sub optimal
# Lex9 Lex13 lor14 loa9
frst_disag = converter.last_disagreement
assert abs(frst_disag - 0.5 *
(1.6 + 0.1 + 0.1 + 0.39 + 1.2) / size_) <= self.tol_one
assert glop_act != glop_act3 # but glop_act should be better !
glop_act4 = converter.convert_act(complex_act, explore=3)
this_disag = converter.last_disagreement
assert this_disag < frst_disag, "this disagreement should always be lower !"
assert converter.indx_sel != 0, "the first has been selected, it should not !"
assert glop_act == glop_act4, "the same action first action should be optimal"
assert abs(this_disag - 0.5 *
(0.4 + 0.1 + 0.1 + 1.61 + 0.8) / size_) <= self.tol_one