def get_weather_by_station(sno):
par = data_parser.data_parser()
par.update_current_data()
bw_df = par.generate_bike_weather_data()
bw_df_sno = bw_df[bw_df['sno'] == sno]
bw_df_sno = bw_df_sno[[
'WDIR', 'WDSE', 'TEMP', 'HUMD', 'PRES', 'H_24R', 'WSGust', 'WDGust',
'UVI', 'PrecpHour', 'Visb'
]]
return bw_df_sno
def track_particles(datapath, dataFormat, minLength, search_range):
types = os.listdir(os.path.join(datapath, dataFormat))
# Import and parse data
untracked = [data_parser(os.path.join(datapath, dataFormat, type_), dataFormat) for i, type_ in enumerate(types)]
print('Particles imported')
# Track the particles
tracked_particle = [tp.link_df(particles, search_range, memory=5).sort_values(by=['particle','frame'], ignore_index=True)
for particles in untracked]
nFiles = len(tracked_particle[0].frame.unique())
del untracked
# Remove small tracks and repair small holes (linear interpolation) and fill large holes with NaN
tracked_repaired = [track_repair(track, 10, nFiles, minLength)
for track in tracked_particle]
del tracked_particle
# Save sorted, renumbered, and complete tracks
tracked_complete = [[] for track in tracked_repaired]
nParticles = [len(track.particle.unique()) for track in tracked_repaired]
for i, track in enumerate(tracked_repaired):
tracked_complete[i] = renumber_particles(track, 'particle', int(np.sum(nParticles[:i])))
tracked_complete = pd.concat(tracked_repaired, ignore_index=True)
tracked_complete = tracked_complete.sort_values(by=['frame', 'particle'], ignore_index=True)
tracked_complete.to_pickle(os.path.join(datapath, r'tracked.pkl'))
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
#tf.enable_eager_execution()
with tf.Session() as sess:
parser = argparse.ArgumentParser(description='')
parser.add_argument(
'--dataset_dir',
dest='dataset_dir',
default='/notebooks/dataset/GANs/Persian_miniature',
help='path of the dataset')
parser.add_argument('--max_depth',
dest='max_depth',
type=float,
default=1,
help='maximum depth')
args = parser.parse_args()
ds = data_parser(args=args)
ds_train = ds.create_dataset()
ds = ds_train.shuffle(buffer_size=1000)
ds = ds.repeat()
#ds = ds.batch(1)
ds = ds.apply(tf.contrib.data.batch_and_drop_remainder(1))
ds = ds.prefetch(buffer_size=AUTOTUNE)
# creating the training iterator
iter = ds.make_one_shot_iterator()
rgb = iter.get_next()
image_mean, image_var = tf.nn.moments(rgb, axes=[0, 1, 2])
print("Nodes in the Ray cluster:")
print(ray.nodes())
from RCTMA_parallel import rctma
else:
from RCTMA_sequential import rctma
IN_FILE = "InputDataC{}.txt".format(str(test_num))
time1 = time.time()
lbd, n0, particles = parameter_parser(nmax, IN_FILE)
os.system("module load cmake")
os.system("cmake external/CMakeLists.txt")
os.system("(cd external && make)")
os.system("./external/external_hij")
data_parser(num, nmax)
time2 = time.time()
print("Time for preparing Hij is:", time2 - time1)
T = rctma(num, nmax, lbd, n0, particles)
time3 = time.time()
print("Time for computing vscatter and rcmta matrices is:", time3 - time2)
J = pickle.load(open("external/data/J.npy", "rb"))
Cext = oaec(lbd, n0, T, J)
time4 = time.time()
print("Cext:", Cext)
print("Time for reading Jij and computing OAEC is :", time4 - time3)
def read_data(self):
dp = data_parser(args=self.args)
self.ds_train = dp.create_dataset()