def perform_update(cx, suite, prefix, fetch):
# Fetch the actual data.
metadata = load_metadata(prefix)
earliest_run_id = metadata['earliest_run_id']
sys.stdout.write('Querying ' + prefix + '... ')
sys.stdout.flush()
with Profiler() as p:
rows = fetch(earliest_run_id)
diff = p.time()
new_rows = len(rows)
print('found ' + str(new_rows) + ' new rows in ' + diff)
# Break everything into months.
year = 0
month = 0
current = []
months = []
for row in rows:
stamp = int(row[1]) - timezone_offset()
t = time.gmtime(stamp)
if t.tm_year != year or t.tm_mon != month:
if year and len(current):
months.append(((year, month), current))
current = []
year = t.tm_year
month = t.tm_mon
current.append(row)
if len(current):
months.append(((year, month), current))
for when, data in months:
name = prefix + '-' + str(when[0]) + '-' + str(when[1])
sys.stdout.write('Updating cache for ' + name + '...')
sys.stdout.flush()
with Profiler() as p:
update_cache(cx, suite, name, when, data)
diff = p.time()
print('took ' + diff)
if len(rows):
metadata['earliest_run_id'] = rows[-1][0]
save_metadata(prefix, metadata)
return new_rows
def run_train_test(ds_name, metric, params, obj):
path = _DATA_PATH + ds_name
with Profiler('initial feature selection'):
x_initial_raw, y_initial, _ = load_data(f'{path}/train.csv',
mode='train',
sample=_SAMPLE)
x_initial, ini_params = initial_processing(x_initial_raw, mode='train')
tf = CatTransformer(ini_params['cat_cols'])
# tf.fit(x_initial)
x_initial_tf = tf.fit_transform(x_initial)
selected_features, feat_list = ols_selection(x_initial_tf, y_initial,
obj)
hp_params = hyperopt_lgb(x_initial_tf[feat_list], y_initial, params,
obj)
print('selected features=', len(selected_features))
x_train_raw, y_train, _ = load_data(f'{path}/train.csv',
mode='train',
sample=_SAMPLE,
used_cols=selected_features)
x_test_raw, _, _ = load_data(f'{path}/test.csv', mode='test')
y_test = load_test_label(f'{path}/test-target.csv')
x_train, train_params = initial_processing(x_train_raw, mode='train')
x_test, test_params = initial_processing(x_test_raw, mode='test')
with Profiler('fit transform cat columns'):
x_test_rein = x_test.reindex(columns=train_params['used_cols'])
tf = CatTransformer(train_params['cat_cols'])
tf.fit(x_train)
x_train_tf = tf.transform(x_train)
x_test_tf = tf.transform(x_test_rein)
with Profiler('run train'):
model = lgb.train(hp_params, lgb.Dataset(x_train_tf, label=y_train),
600)
with Profiler('predict'):
y_train_out = model.predict(x_train_tf)
y_test_out = model.predict(x_test_tf)
train_err = metric(y_train, y_train_out)
test_err = metric(y_test, y_test_out)
return train_err, test_err
def test_class_profiler_should_calculate_total_acceleration_correctly_balanced_forces(
):
# Testing to see acceleration = 0 when F_gravity == F_buoyancy
parameters = {
'body_d': 0.50465,
'piston_d': 0.0,
'piston_l': 0.0,
'density': 1025,
'depth': 0.0,
'velocity': 0.0,
'mass': 50.0
}
# f_b - f_g = 0
# f_g = m * g
# f_b = V * r * g
# f_b = m * acc_b
# m * acc_b - m * g = 0
# acc_b - g = 0
f_g = parameters['mass'] * g
volume = parameters['mass'] / parameters['density']
area = ((parameters['body_d'] / 2)**2) * np.pi
length = volume / area
parameters['body_l'] = length
p = Profiler(**parameters)
assert 0 == p.acceleration
def train():
# Initialize torch.distributed
init_distributed()
print_rank_0('AutoMP: training GPT2...')
# Use fake train data
batch_size = args.batch_size
sequence_length = args.sequence_length
hidden_size = args.hidden_size
vocab_size = args.vocab_size
dropout_prob = args.hidden_dropout
input_indices = torch.randint(low=0,
high=vocab_size,
size=(batch_size, sequence_length))
input_indices = input_indices.to(torch.cuda.current_device())
position_indices = torch.tile(torch.arange(start=0, end=sequence_length),
(batch_size, 1))
position_indices = position_indices.to(torch.cuda.current_device())
print_rank_0(f'AutoMP: input_indices shape = {input_indices.size()}')
print_rank_0(f'AutoMP: position_indices shape = {position_indices.size()}')
def init_method_normal(tensor):
return torch.nn.init.normal_(tensor, mean=0.0, std=1.0)
embedding = Embedding(hidden_size=hidden_size,
vocab_size=vocab_size,
max_sequence_length=sequence_length,
embedding_dropout_prob=dropout_prob,
init_method=init_method_normal)
optimizer = torch.optim.SGD(embedding.parameters(), lr=0.01)
profiler = Profiler(os.path.join('benchmark', args.exp_name))
num_epochs = 5
tot_time = 0
nproc = torch.distributed.get_world_size()
for epoch in range(num_epochs):
overall_name = f'emb_np-{nproc}_vs-{vocab_size}'
profiler.start(overall_name)
# Forward pass
profiler.start(f'emb_forward_np-{nproc}_vs-{vocab_size}')
embedding_output = embedding.forward(input_indices, position_indices)
train_loss = torch.mean(embedding_output)
torch.cuda.synchronize()
profiler.stop(f'emb_forward_np-{nproc}_vs-{vocab_size}')
# Backward pass
profiler.start(f'emb_backward_np-{nproc}_vs-{vocab_size}')
optimizer.zero_grad()
train_loss.backward()
optimizer.step()
torch.cuda.synchronize()
profiler.stop(f'emb_backward_np-{nproc}_vs-{vocab_size}')
profiler.stop(overall_name)
def test_class_profiler_should_initialize_with_given_parameters():
parameters = {
'body_d': 0.4,
'body_l': 1.0,
'piston_d': 0.1,
'piston_l': 0.3,
'density': 1023.2,
'depth': 0.0,
'velocity': 0.0,
'mass': 12.2
}
p = Profiler(**parameters)
expected_body_volume = 0.12566
expected_piston_volume = 0.002356
assert np.isclose(expected_body_volume,
p.body.volume,
rtol=1e-3,
atol=1e-4)
assert np.isclose(expected_piston_volume,
p.piston.volume,
rtol=1e-3,
atol=1e-4)
assert parameters['density'] == p.water.density
assert parameters['depth'] == p.water.depth
assert parameters['velocity'] == p.velocity
assert parameters['mass'] == p._mass
def __init__(self, players):
pf = Profiler()
pf.printer.indent()
pf.printer.silent = not PERFORMANCE_LOGGING
self.players = players
# Set up the deck
deck = INITIAL_DECK[:]
pf.measure("Set up deck")
shuffle(deck)
pf.measure("Shuffle deck")
self.deck = deck
""" Current draw deck. Excludes Snitches. """
self.discard_deck = []
""" Deck discarded (played) cards. Excludes Snitches. """
pf.measure("Other set up")
# Give each player the initial number of Snitches and deal the rest.
for player in self.players:
number_of_character_cards_to_deal = INITIAL_CARDS_PER_PLAYER - INITIAL_SNITCHES_PER_PLAYER
character_hand = self.deck[:number_of_character_cards_to_deal]
del self.deck[:number_of_character_cards_to_deal]
player.set_up(
hand=INITIAL_SNITCH_HAND + character_hand,
coins=INITIAL_COINS_PER_PLAYER)
player.prepare()
pf.measure("Give players cards")
def renew_cache(cx, machine, suite, prefix, when, fetch):
delete_cache(prefix + '-' + str(when[0]) + '-' + str(when[1]));
# Delete corresponding condensed graph
before, after = prefix.split("raw", 1)
delete_cache(before + "condensed" + after + '-' + str(when[0]) + '-' + str(when[1]));
dt = datetime.datetime(year=when[0], month=when[1], day=1)
start_stamp = int(time.mktime(dt.timetuple()))
next_month = when[1] + 1
next_year = when[0]
if next_month == 13:
next_month = 1
next_year += 1
dt = datetime.datetime(year=next_year, month=next_month, day=1)
stop_stamp = int(time.mktime(dt.timetuple())) - 1
name = prefix + '-' + str(when[0]) + '-' + str(when[1])
# Querying all information from this month.
sys.stdout.write('Fetching monthly info ' + name + '... ')
sys.stdout.flush()
with Profiler() as p:
rows = fetch(machine, approx_stamp=(start_stamp,stop_stamp))
diff = p.time()
new_rows = len(rows)
print('found ' + str(new_rows) + ' rows in ' + diff)
update_cache(cx, suite, name, when, rows)
def main(rank, world_size):
global tb_logger, profiler, dataset, log_dir, trainloader, testloader
if not args.no_tensorboard:
log_dir = os.path.join('log',
args.expid,
datetime.now().isoformat())
tb_logger = SummaryWriter(log_dir=log_dir)
logger = Logger()
if not args.no_profiler:
profiler = Profiler(logger, tb_logger, freq=args.profile_freq)
profiler.log(log_network=args.profile_networkio)
tb_logger.add_text('params/batch_size', str(args.batch_size/world_size), 1)
sync_batch = args.batch_size / world_size
trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train)
trainset.train_data = np.split(trainset.train_data, args.split_by)[args.split_to_use]
trainloader = torch.utils.data.DataLoader(trainset, batch_size=sync_batch, shuffle=False, num_workers=2)
testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transform_test)
testloader = torch.utils.data.DataLoader(testset, batch_size=100, shuffle=False, num_workers=2)
for epoch in range(args.epochs):
train(epoch)
test(epoch)
def renew_cache(cx, machine, suite, prefix, when, last_stamp, fetch):
delete_cache(prefix + '-' + str(when[0]) + '-' + str(when[1]))
# Delete corresponding condensed graph
if prefix[0:3] == "raw":
delete_cache("condensed" + prefix[3:] + '-' + str(when[0]) + '-' +
str(when[1]))
else:
delete_cache("bk-condensed" + prefix[6:] + '-' + str(when[0]) + '-' +
str(when[1]))
dt = datetime.datetime(year=when[0], month=when[1], day=1)
start_stamp = int(time.mktime(dt.timetuple()))
next_month = when[1] + 1
next_year = when[0]
if next_month == 13:
next_month = 1
next_year += 1
dt = datetime.datetime(year=next_year, month=next_month, day=1)
stop_stamp = int(time.mktime(dt.timetuple())) - 1
if last_stamp < stop_stamp:
stop_stamp = last_stamp
# Querying all information from this month.
sys.stdout.write('Querying ' + prefix + '... ')
sys.stdout.flush()
with Profiler() as p:
rows = fetch(machine, test_stamp=(start_stamp, stop_stamp))
diff = p.time()
new_rows = len(rows)
print('found ' + str(new_rows) + ' rows in ' + diff)
name = prefix + '-' + str(when[0]) + '-' + str(when[1])
update_cache(cx, suite, name, when, rows)
def test_create_opinion_button(self):
"""
Test create opinion button
"""
print 'run create_opinion_button...'
opinion_button = self.profiler.create_opinion_button()
print 'opinion_button:'
print opinion_button
self.assertEqual(type(opinion_button), dict)
self.assertTrue(opinion_button['saved'])
self.assertEqual(opinion_button['object'].id, self.position_opinion.id)
print '\n' + '.' * 60 + '\n'
print 'test false...'
print '\n' + '.' * 60 + '\n'
print 'run create_opinion_button...'
self.profiler = Profiler(self.position_set, date='2015-04-09')
opinion_button = self.profiler.create_opinion_button()
self.assertEqual(type(opinion_button), dict)
self.assertFalse(opinion_button['saved'])
print 'opinion_button:'
print opinion_button
def setUp(self):
TestSetUpDB.setUp(self)
# set date
self.date = '2015-04-28'
# create underlying
self.underlying = Underlying(symbol='AAPL', company='APPLE INC')
# create position set
self.position_set = PositionSet()
self.position_set.underlying = self.underlying
self.position_set.name = 'EQUITY'
self.position_set.spread = 'LONG_STOCK'
self.position_set.start_date = datetime.strptime(
'2015-04-13', '%Y-%m-%d').date()
self.position_set.stop_date = datetime.strptime(
'2015-04-28', '%Y-%m-%d').date()
self.position_set.save()
# create position opinion
self.position_opinion = self.create_opinion(
date=Profiler.move_bday(self.date, 1),
position_set=self.position_set,
direction='BULL')
# create profiler now
self.profiler = Profiler(position_set=self.position_set,
date=self.date)
def test_class_profiler_should_calculate_total_acceleration_with_drag_included(
):
parameters = {
'body_d': 0.50465,
'body_l': 1.0,
'piston_d': 0.0,
'piston_l': 0.0,
'density': 1025,
'depth': 0.0,
'velocity': 1.0,
'mass': 50.0
}
expected_f_drag = 84.05
f_buoy = 2010.55
f_grav = 490.33
f_total = f_buoy - (expected_f_drag + f_grav)
p = Profiler(**parameters)
print(p.acceleration)
assert np.isclose(expected_f_drag,
p.drag.drag,
rtol=1e-4,
atol=1e-6,
equal_nan=False)
def __init__(self):
self.conf_space = ConfSpace()
self.profiler = Profiler()
self.predictor = PerfPredict()
self.exploit_times = 0
self.exploit_max = 5
# how many configurations to predict with performance predictor before really runs the benchmark
self.predict_max = 1000
def setUp(self):
TestUnitSetUp.setUp(self)
self.position_set = PositionSet.objects.get(id=65)
self.date = self.position_set.filledorder_set.order_by('trade_summary__date') \
.last().trade_summary.date.strftime('%Y-%m-%d')
self.profiler = Profiler(position_set=self.position_set,
date=self.date)
def __init__(self, socket):
log.error(DEBUG, msg='worker __init__')
self._socket = socket
self._epoll = select.epoll()
self._profiler = Profiler()
self._activity_iterators = {}
self._child_pids = []
self._client_conns = []
self._socket.setblocking(False)
def main(argv):
sys.stdout.write('Computing master properties... ')
sys.stdout.flush()
with Profiler() as p:
cx = data.Context()
diff = p.time()
print('took ' + diff)
update_all(cx)
condenser.condense_all(cx)
export_master(cx)
def condense(cx, suite, prefix, name):
with Profiler() as p:
sys.stdout.write('Importing all datapoints for ' + name + '... ')
sys.stdout.flush()
files = find_all_months(cx, prefix, name)
diff = p.time()
print('took ' + diff)
if not len(files):
return False
change = False
for when, raw_file in files:
condensed_name = prefix + 'condensed-' + name + '-' + str(
when[0]) + '-' + str(when[1])
condensed_file = condensed_name + '.json'
# Only update the graph when condensed file is older.
if os.path.exists(os.path.join(
awfy.path, condensed_file)) and file_is_newer(
os.path.join(awfy.path, condensed_file),
os.path.join(awfy.path, raw_file)):
continue
# There was a datapoint added to one of the condensed files.
change = True
with Profiler() as p:
sys.stdout.write('Condensing ' + condensed_name + '... ')
sys.stdout.flush()
graph = retrieve_graph(cx, raw_file)
condense_month(cx, suite, graph, prefix, condensed_name)
diff = p.time()
print(' took ' + diff)
return change
def main():
model = Network()
checkpoint = torch.load('/mnt/nvme/adas/code/custom_layer/checkpoint_8.pt')
model.load_state_dict(checkpoint['model_state_dict'])
P = Profiler(model, (4, 3, 512, 256))
total_ops, total_params, total_mult, total_add = P.profile()
print("=====================================================")
print("Mult Ops: %f GOps" % (total_mult / 1e9))
print("Add Ops: %f GOps" % (total_add / 1e9))
print("Total Ops: %f GOps" % (total_ops / 1e9))
print("Total Parameters: %f * 10^6 " % (total_params / 1e6))
def profiler(self, job_name):
db_meta = self._load_db_metadata()
if isinstance(job_name, basestring):
job_id = None
for job in db_meta.jobs:
if job.name == job_name:
job_id = job.id
break
if job_id is None:
raise ScannerException('Job name {} does not exist'.format(job_name))
else:
job_id = job_name
return Profiler(self, job_id)
def fetch_test_scores(machine_id,
suite_id,
name,
finish_stamp=(0, "UNIX_TIMESTAMP()"),
approx_stamp=(0, "UNIX_TIMESTAMP()")):
c = awfy.db.cursor()
query = "SELECT id FROM awfy_suite_test \
WHERE name = %s"
c.execute(query, [name])
suite_ids = ['0']
for row in c.fetchall():
suite_ids.append(str(row[0]))
with Profiler() as p:
query = "SELECT id \
FROM awfy_run \
WHERE status > 0 \
AND machine = %s \
AND approx_stamp >= " + str(approx_stamp[0]) + " \
AND approx_stamp <= " + str(approx_stamp[1]) + " \
AND finish_stamp >= " + str(finish_stamp[0]) + " \
AND finish_stamp <= " + str(finish_stamp[1]) + " \
"
c.execute(query, [machine_id])
run_ids = ['0']
for row in c.fetchall():
run_ids.append(str(row[0]))
diff = p.time()
print('found ' + str(len(run_ids)) + ' rows in ' + diff)
query = "SELECT r.id, r.approx_stamp, bu.cset, s.score, bu.mode_id, v.id, s.id \
FROM awfy_suite_version v \
JOIN awfy_suite_test t ON v.id = t.suite_version_id \
JOIN awfy_breakdown s ON s.suite_test_id = t.id \
JOIN awfy_score s1 ON s.score_id = s1.id \
JOIN awfy_build bu ON s1.build_id = bu.id \
JOIN awfy_run r ON r.id = bu.run_id \
WHERE v.suite_id = %s \
AND t.id in (" + (
",".join(suite_ids)) + ") \
AND r.id in (" + (
",".join(run_ids)) + ") \
ORDER BY r.sort_order ASC \
"
c.execute(query, [suite_id])
return c.fetchall()
def train(self):
rnn_type = LSTM
rnn_dropout = 0.0
rnn_units = 128
rnn_timesteps = 128
rnn_features = 32
output_size = 1
batch_size = 512
epochs = 10000000
input_shape = (rnn_timesteps, rnn_features)
if os.path.isfile(self.model_path):
profiler = Profiler()
model = load_model(self.model_path,
custom_objects={
'root_mean_squared_error':
root_mean_squared_error
})
profiler.stop(f'Loaded model from "{self.model_path}".')
else:
model = Sequential()
model.add(
rnn_type(rnn_units,
dropout=rnn_dropout,
return_sequences=False,
input_shape=input_shape))
model.add(Dense(output_size))
model.add(Activation('tanh'))
optimizer = Adam(lr=0.01)
model.compile(optimizer=optimizer, loss=root_mean_squared_error)
training_generator = SlidingWindowGenerator(self.x_training_wav,
self.y_training_wav,
input_shape, output_size,
batch_size)
validation_generator = SlidingWindowGenerator(self.x_validation_wav,
self.y_validation_wav,
input_shape, output_size,
batch_size)
save_callback = SaveCallback(self.model_path)
history = model.fit_generator(generator=training_generator,
epochs=epochs,
verbose=1,
validation_data=validation_generator,
callbacks=[save_callback])
def main(size=50000, sort=quicksort):
lyst = []
p = Profiler()
for count in range(size):
lyst.append(random.randint(1, size + 1))
print(lyst)
p.test(sort,
lyst,
size,
unique=True,
comp=True,
exch=True,
trace=False,
mod=False)
def condense(cx, suite, prefix, name):
sys.stdout.write('Importing all datapoints for ' + name + '... ')
sys.stdout.flush()
with Profiler() as p:
graphs = find_all_months(cx, prefix, name)
diff = p.time()
print('took ' + diff)
if not len(graphs):
return
for when, graph in graphs:
new_name = prefix + 'condensed-' + name + '-' + str(
when[0]) + '-' + str(when[1])
# Don't condense if it already exists...
if not should_export(new_name + '.json', when):
continue
sys.stdout.write('Condensing ' + new_name + '... ')
sys.stdout.flush()
with Profiler() as p:
condense_month(cx, suite, graph, prefix, new_name)
diff = p.time()
print(' took ' + diff)
# Combine everything.
sys.stdout.write('Aggregating ' + name + '... ')
sys.stdout.flush()
with Profiler() as p:
combined = combine([graph for when, graph in graphs])
summary = aggregate(combined)
diff = p.time()
print('took ' + diff)
return summary
def __init__(self, init_state: tuple):
self.init_state = init_state
# saves states tuples, uses a double ended queue;
# can behave both as a FIFO queue and as a LIFO stack
self.fringe = deque()
# saves states tuples; uses a set for speed when checking if a state
# has been explored
self.explored = set()
# saves nodes details for every node state:
# nodeState -> (nodeParent, action, cost, depth)
self.node_db = dict()
self.profiler = Profiler()
self.set_goal()
def __init__(self):
self.conf_space = ConfSpace()
self.semantics = self.conf_space.hadoop_semantics
self.profiler = Profiler()
# self.predictor = PerfPredict()
self.profile_num = 0 # initial profiling index
# self.performance_track = []
# self.K_iters_for_convergence_test = 5 # check the last 5 iterations
# self.perf_improvement_threshold = 0.02 # 2% of the improvement on the last cycle
self.best_conf = None
self.global_improvement = 0.5
self.max_iter = 6000
self.initial_perf = sys.maxsize
self.type_checker = HadoopConfChecker()
self.type_checker.set_all_param_value(self.conf_space.param_values)
self.invalid_confs = []
self.curr_genconf_folder = cfg.gen_confs + os.sep + 'conf'
def main():
old_stdout = sys.stdout
with open('data_dictionary/profiles/backup.nquads', "w+") as backup:
# backup database
sys.stdout = backup
Backup()
with open('data_dictionary/jupiter_ontology.md', "w+") as ontology:
# serialize a new jupiter ontology file
sys.stdout = ontology
owlDocument().generate()
sys.stdout = old_stdout
for ptype in [
"collection", "community", "generic", "thesis", "oai_pmh",
"oai_etdms"
]:
# serialize a profile for each object type
Profiler(ptype)
excelGen()
def findBestCase():
'''
Function to compile a table of the time used for sorting for:
different list sizes and threshold values.
'''
p = Profiler()
masterList = [] #list to contain the lists of different sizes
resultNoMod = []
print(" " * 2, end="")
for exponent in range(
1, 5
): #this creates the master list using random numbers for different list sizes
size = 5 * 10**exponent
lyst = []
for count in range(size):
lyst.append(random.randint(1, size + 1))
masterList.append(lyst)
print(f"{size:<13}",
end="") #this prints the table headers (list size)
print()
#this creates the results for non-modified quicksort
for i in range(0, len(masterList)):
tempLyst = masterList[
i][:] #create list so we don't change the master list
resultNoMod.append(repr(p.test(quicksort, tempLyst, mod=False)))
print(resultNoMod, "Not Modified")
#this creates the results for non-modified quicksort, with different thresholds
for multiplier in range(1, 5):
resultMod = []
thresholdValue = multiplier * 10
for i in range(0, len(masterList)):
tempLyst = masterList[
i][:] #create list so we don't change the master list
resultMod.append(
repr(
p.test(quicksort,
tempLyst,
threshold=thresholdValue,
mod=True)))
print(resultMod, "Threshold =", thresholdValue) #prints threshold
def test_class_profiler_should_calculate_total_volume_correctly():
parameters = {
'body_d': 0.4,
'body_l': 1.0,
'piston_d': 0.1,
'piston_l': 0.3,
'density': 1023.2,
'depth': 0.0,
'velocity': 0.0,
'mass': 12.2
}
p = Profiler(**parameters)
expected_body_volume = 0.12566
expected_piston_volume = 0.002356
expected_total_volume = expected_body_volume + expected_piston_volume
assert np.isclose(expected_total_volume, p.volume, rtol=1e-4, atol=1e-6)
def testPlanner(planner,numTrials,maxTime,filename):
print "Testing planner for %d trials, %f seconds"%(numTrials,maxTime)
print "Saving to",filename
f = open(filename,'w')
f.write("trial,plan iters,plan time,best cost\n")
for trial in range(numTrials):
print
print "Trial",trial+1
planner.reset()
curCost = float('inf')
t0 = time.time()
numupdates = 0
iters = 0
hadException = False
while time.time()-t0 < maxTime:
try:
planner.planMore(10)
except Exception as e:
if hadException:
print "Warning, planner raise two exceptions in a row. Quitting"
break
else:
print "Warning, planner raised an exception... soldiering on"
print e
hadException = True
continue
iters += 10
if planner.bestPathCost != None and planner.bestPathCost != curCost:
numupdates += 1
curCost = planner.bestPathCost
t1 = time.time()
f.write(str(trial)+","+str(iters)+","+str(t1-t0)+","+str(curCost)+'\n')
if hasattr(planner,'stats'):
print
temp = Profiler()
temp.items["Stats:"] = planner.stats
temp.pretty_print()
print
print "Final cost:",curCost
print
f.write(str(trial)+","+str(iters)+","+str(maxTime)+","+str(curCost)+'\n')
f.close()
def modifiedQuicksort(lyst, threshold):
p = Profiler()
if len(
lyst
) <= threshold: #I took the instructions to mean less than or equal to because the testing phase is with
# data of 50 to 5000... use variable to make the threshold editable from the testing side.
p.test(insertionSort,
lyst=lyst,
size=len(lyst),
comp=False,
exch=False,
trace=False)
else:
p.test(qs.quicksort,
lyst=lyst,
size=len(lyst),
comp=False,
exch=False,
trace=False)
