def bubble(numbers):
ret = 0
for i in range(0, len(numbers) - 1):
for j in range(0, len(numbers) - i - 1):
if numbers[i] > numbers[i + 1]:
numbers[i], numbers[i + 1] = numbers[i + 1], numbers[i]
ret += 1
print_res("Bubble sort:", ret)
def selection(numbers):
ret = 0
for x in range(0, len(numbers) - 1):
min = x
for y in range(x + 1, len(numbers)):
if numbers[y] < numbers[min]:
min = y
ret += 1
numbers[x], numbers[min] = numbers[min], numbers[x]
print_res("Selection sort:", ret)
def main(args):
numbers = get_values(args[0])
if (len(numbers) <= 0):
sys.exit(84)
print(len(numbers), " element", sep="", end="")
print("s" if len(numbers) > 1 else "")
selection(numbers[::])
insertion(numbers[::])
bubble(numbers[::])
print_res("Quicksort:", 0 if (len(numbers) <= 1) else quicksort(numbers[::])[1])
print_res("Merge sort:", merge(numbers[::])[1])
def insertion(numbers):
ret = 0
for x in range(1, len(numbers)):
y = x
while (y > 0):
ret += 1
if (numbers[y - 1] <= numbers[y]):
numbers[y], numbers[y - 1] = numbers[y - 1], numbers[y]
y -= 1
else:
break
print_res("Insertion sort:", ret)
def process_ms(msname, outfn):
station_map = utils.get_station_map(msname)
station_names = sorted(station_map.keys())
stations2 = sorted(map(station_map.get, station_names))
ref_station_name = '1' # ANTENNA table has ID, Name and 'Station' fields'; here we want the last of these
ref_station2 = station_map[ref_station_name]
ref_s_ind2 = stations2.index(ref_station2)
swid, pol_id, polind, = (0, 0, 0)
for timeq in timeqs:
delays, phases, rates, sigs = [], [], [], []
for pol_id in [0,3]:
dels, phs, rs, sig = fringer.fit_fringe_lm(msname, stations2, ref_s_ind2,
swid, polind, pol_id, timeq, solint=300)
delays.append(dels)
phases.append(phs)
rates.append(rs)
sigs.append(sig)
outf = file(outfn, 'w')
print >>outf, "#\n#{}\n#".format(timeq)
utils.print_res(station_map, stations2, phases, delays, rates, sigs=sigs, outf=outf)
def main():
testset, test_sampler = get_test_sampler(args)
loss = get_val_loss_fn(args)
# getting number of classes in the training set
if 'imagenet' in args['model.dataset']:
ncls = 351 if 'tiered' in args['model.dataset'] else 64
else:
ncls = 100
# defining the model
model = get_model(ncls, args, ensemble=False)
model.eval()
checkpointer = CheckPointer('singles', args, model)
ckpt_path = checkpointer.best_ckpt
if os.path.isfile(ckpt_path):
start_epoch, best_val_loss, best_val_acc, waiting_for =\
checkpointer.restore_model(ckpt='best', strict=False)
else:
print('Model ckpt file was not found at {}'.format(ckpt_path))
exit(1)
loss = get_val_loss_fn(args)
if args['test.determ']:
set_determ(args['test.seed'])
losses, accs, _ = [], [], []
for sample in tqdm(test_sampler):
with torch.no_grad():
_, stats_dict, pred_dict = loss(model, sample,
args['test.distance'])
losses.append(stats_dict['loss'])
accs.append(stats_dict['acc'] * 100)
model_path = '/'.join(ckpt_path.split('/')[:-1])
print_res(losses, 'loss', os.path.join(model_path, 'stats.txt'))
print_res(accs, 'accuracy', os.path.join(model_path, 'stats.txt'), prec=2)
def main():
testset, test_sampler = get_test_sampler(args)
loss = get_val_loss_fn(args)
# getting number of classes in the training set
if 'imagenet' in args['model.dataset']:
ncls = 351 if 'tiered' in args['model.dataset'] else 64
else:
ncls = 100
# defining the model
model = get_model(ncls, args, ensemble=True)
model.eval()
checkpointer = CheckPointer('ensembles', args, model)
ckpt_path = checkpointer.best_ckpt
if os.path.isfile(ckpt_path):
start_epoch, best_val_loss, best_val_acc, waiting_for =\
checkpointer.restore_model(ckpt='best', strict=False)
elif 'copy__' in ckpt_path:
checkpointer.restore_from_singles()
else:
print('Model ckpt file was not found at {}'.format(ckpt_path))
exit(1)
# to ensure deterministic behavior
if args['test.determ']:
set_determ(args['test.seed'])
# to collect testing results and report statistics later
losses, probsum_accs, voted_accs, all_accs, agreements = [], [], [], [], []
for sample in tqdm(test_sampler):
with torch.no_grad():
_, stats_dict, pred_dict = loss(model, sample)
losses.append(stats_dict['loss'])
voted_accs.append(stats_dict['voted_acc'] * 100)
probsum_accs.append(stats_dict['probsum_acc'] * 100)
all_accs.append(stats_dict['accs'] * 100)
agreements.append(stats_dict['agreement'])
model_path = '/'.join(ckpt_path.split('/')[:-1])
stats_path = os.path.join(model_path, 'stats.txt')
if 'copy__' in model_path or args['test.noprint']:
stats_path = None
# processing and printing out statistics
agreement_mat = np.array(agreements).mean(0)
np.set_printoptions(precision=2)
print(agreement_mat)
accs_mat = np.stack(all_accs)
accs = accs_mat.mean(0)
min_ind = accs.argmin(0)
max_ind = accs.argmax(0)
print('All accs: ', accs)
print_res(losses, 'loss', stats_path)
print_res(accs_mat.mean(1), 'mean_accuracy', stats_path, prec=2)
print_res(accs_mat[:, min_ind], 'min_accuracy', stats_path, prec=2)
print_res(accs_mat[:, max_ind], 'max_accuracy', stats_path, prec=2)
print_res(voted_accs, 'voted_accuracy', stats_path, prec=2)
print_res(probsum_accs, 'probsum_accuracy', stats_path, prec=2)
stations2 = [0, 1, 2, 3, 4, 5, 6, 7, 9, 10]
ref_station_name = myrefant # '1' is Effelsberg
ref_station2 = station_map[ref_station_name]
ref_s_ind2 = stations2.index(ref_station2)
swid, polind, = (0, 0)
#stations2 = [0,1,2,3,4,5,6,7,9,10]
stations2 = [0, 1, 2, 3, 4, 5, 6, 7, 9, 10]
for timeq, swid in itertools.product(timeqs, range(4)):
timeq2 = ffd.actual_timerangeq(msname, timeq)
delays, phases, rates, sigs = [], [], [], []
for pol_id in [0, 3]:
print "# Doing a thing", swid, pol_id
dels, phs, rs, sig = fringer.fit_fringe_lm(msname,
stations2,
ref_s_ind2,
swid,
polind,
pol_id,
timeq2,
solint=mysolint,
threshold=1000)
delays.append(dels)
phases.append(phs)
rates.append(rs)
sigs.append(sig)
print "#\n#{}\n# IF {}\n#".format(timeq, swid + 1)
utils.print_res(station_map, stations2, phases, delays, rates)
