def main():
parser = argparse.ArgumentParser(description='do symbolic regression')
parser.add_argument('train_file', help='file to train')
parser.add_argument('-i', type=int, default=100, help='number of iteration')
parser.add_argument('-p', type=int, default=1000, help='population size')
parser.add_argument('-v', action='store_true', help='print detail information')
parser.add_argument('-r', action='store_true', help='recover from dump')
args = parser.parse_args()
if not os.path.isfile(args.train_file):
print('main: train file not exist')
exit(1)
with open(args.train_file, 'r') as f:
lines = f.readlines()
data, label = parse_data(lines)
load_data(data, label)
set_config({
'ITER_NUM': args.i,
'POP_SIZE': args.p,
'PRESERVE_NUM': args.i // 2,
'CONST': [0, 2, 2.5, 3, 4, 5, 10],
'PROB_OP': {'+': 2, '-': 2, '*': 2, '/': 2, '^': 0.5, '~': 3,
'abs': 0.5, 'sin': 1.5, 'cos': 1.5, 'tan': 1,
'asin': 0.5, 'acos': 0.5, 'atan': 0.5,
'sinh': 0.5, 'cosh': 0.5, 'tanh': 0.5,
'exp': 2, 'sqrt': 2, 'log': 2},
'VERBOSE': args.v
})
result = train(args.r)
print('MSE: %s' % result[1])
print('EXPR (postfix): %s' % result[0].to_postfix())
print('EXPR (infix): %s' % result[0])
def cvd(data, L, convert_stim, params, num_iters, gamma, tau_r, tau_d):
'''
Estimate latent factor activity on held-out test data, keeping static params fixed.
'''
f, s = core.load_data(data, convert_stim)
N, T = f.shape
K = s.shape[0]
alpha, beta, w, b, sigma = params
kernel = core.calcium_kernel(tau_r, tau_d, T)
init_x = np.random.rand(L, T)
latent_bounds = [(0, None)] * (L * T)
args = [f, s, kernel, N, T, K, L, gamma, sigma, alpha, beta, w, b]
t_start = time.time()
x_hat = core.estimate_latents(init_x, args, latent_bounds, num_iters)
t_end = time.time()
print('Total elapsed time: %.2fs (%im).' % (t_end - t_start,
(t_end - t_start) // 60))
return x_hat
def train(data, convert_stim, L, num_iters, iters_per_altern, gamma, tau_r,
tau_d, imrate):
'''
Train the calcium-imaging latent variable analysis model.
'''
# Load training data
f, s = core.load_data(data, convert_stim)
N, T = f.shape
K = s.shape[0]
# Estimate imaging-noise variance
sigma = core.estimate_noise_sdevs(f, N, T, imrate)
# Initialise params
kernel = core.calcium_kernel(tau_r, tau_d, T)
init_alpha = np.random.normal(1, 1e-2, N)
init_beta = np.zeros((N, ))
init_w = core.fit_regressors(f, s, kernel, N, T, K)
init_b = np.random.rand(N, L)
init_x = 1e-1 * np.random.rand(L, T)
initial_params = [init_alpha, init_beta, init_w, init_b, init_x]
args = [f, s, kernel, N, T, K, L, gamma, sigma]
# Set non-negative parameter bounds for L-BFGS-B optimisation
eps = 1e-8
static_bounds = [(eps, None)] * N # alpha
static_bounds += [(0, None)] * (N + N * K + N * L) # beta, w, b
latent_bounds = [(0, None)] * (L * T) # x
# Alternating optimisation of static parameters and latent variables
t_start = time.time()
alpha, beta, w, b, x = core.alternating_minimisation(
initial_params, args, static_bounds, latent_bounds, num_iters,
iters_per_altern)
t_end = time.time()
print('Total elapsed time: %.2fs (%im).' % (t_end - t_start,
(t_end - t_start) // 60))
# Parameter identification
param_identification_args = [N, L, s]
alpha_hat, beta_hat, w_hat, b_hat, x_hat, sigma_hat = core.identify_params(
alpha, beta, w, b, x, sigma, param_identification_args)
return [alpha_hat, beta_hat, w_hat, b_hat, x_hat, sigma_hat]
def main(model, epochs, batch_size, save_intervals, category):
my_model = core.create_model(model, category)
print(
"Python main program for generating images using {} with category {}".
format(model, category))
## preprocess data images if init_train and save the images as pickle file
final_images_stacked = core.load_data(my_model.rows)
my_model.train(data=final_images_stacked,
epochs=epochs,
batch_size=batch_size,
save_intervals=save_intervals,
sample_intervals=save_intervals,
hi_sample_intervals=save_intervals)
def select_model(folder,
data,
data_type,
return_fit='both',
convert_stim=False):
'''
Selects the model-fit with the highest posterior probability among a folder of fits.
Arguments:
folder:
Directory containing a set of model fits.
data:
Path to the associated data.
data_type:
Set to 'train' if input data is training data, 'test' for test data.
return_type:
Set to 'train' to return the fits for training data, 'test' to return the fits for test data,
or 'both' to return both fits.
convert_stim:
Set to True if stimulus must be converted from a 1d to 2d representation.
'''
model_fits = [f for f in os.listdir(folder) if not f.startswith('.')]
llhs = [None] * len(model_fits)
f, s = core.load_data(data, convert_stim)
N, T = f.shape
for findx, fit in enumerate(model_fits):
alpha, beta, w, b, x, sigma, tau_r, tau_d, gamma, L = load_fit(
folder + '/' + fit, data_type)
if L == 1:
x = np.reshape(x, [int(L), T])
b = np.reshape(b, [N, int(L)])
kernel = core.calcium_kernel(tau_r, tau_d, T)
K = s.shape[0]
llhs[findx] = -core.log_joint(f, s, kernel, N, T, K, L, gamma, sigma,
alpha, beta, w, b, x)
return load_fit(folder + '/' + model_fits[np.argmax(llhs)], return_fit)
def get_weather_statistics_view(request):
"""api view to return computed weather statistics"""
get_data = request.GET
city = get_data.get("city", "Kampala")
start_date = get_data.get("start_date", "2020-01-01")
end_date = get_data.get("end_date", "2020-01-31")
# load city data for a given period
weather_data = load_data(city, start_date, end_date)
# get weather info
weather_info = get_weather_info(weather_data)
# generate weather dataframe to be used to
# compute statistics
df = generate_df(weather_info)
# compute required statics
computed_df = compute_statistics(df)
statistics = computed_df.to_dict()
stat_data = {
"max_tempC": statistics["max_temp_C"]["max"],
"max_tempF": statistics["max_temp_F"]["max"],
"min_tempC": statistics["min_temp_C"]["min"],
"min_tempF": statistics["min_temp_C"]["min"],
"avg_tempC": statistics["avg_temp_C"]["mean"],
"avg_tempF": statistics["avg_temp_F"]["mean"],
"max_humidity": statistics["max_humidity"]["max"],
"min_humidity": statistics["min_humidity"]["min"],
"avg_humidity": statistics["avg_humidity"]["mean"]
}
return Response(data=stat_data)
import core
import re
data = core.load_data("5.txt")
seats = []
for value in data:
value = re.sub("[FL]", "0", value)
value = re.sub("[BR]", "1", value)
row = int(value[0:7], 2)
column = int(value[7:], 2)
id = row * 8 + column
seats.append(id)
seats.sort()
min = seats[0]
max = seats[-1]
print("Max is %d" % max)
for i in range(min, max):
if not i in seats:
print("Seat %d is missing" % i)
def main(m, vec):
imgs = m.generate_images(vec)
filename = args.filename.replace('{EP}', str(m.epoch))
filename = filename.replace(
'{D}', "{}{}".format(int(args.use_img), int(args.noise)))
mi, ma = np.min(imgs), np.max(imgs)
imgs = ((imgs - mi) / (ma - mi) * 255).astype(np.uint8)
for i, im in enumerate(imgs):
fn = filename.replace('{N}', str(i))
utils.save_image(im, os.path.join(args.path, fn))
if "VAE" in args.model:
if args.use_img:
data = core.load_data()
if args.shuffle: random.shuffle(data)
data = data[0:args.n]
vec = m.get_vector_representation(data)
vec = vec[2]
else:
vec = np.random.normal(0, 1, (args.n, m.latent_dim))
if args.noise:
vec = vec + np.random.normal(0, 1, (args.n, m.latent_dim))
if args.epocycle:
eps = list(m.available_epochs())
for ep in eps:
m.load_epoch(ep)
main(m, vec)
else:
return
l2 += 1
l1 += 1
def find_3(numbers):
l1 = 0
for n1 in numbers:
l2 = 0
for n2 in numbers:
l3 = 0
for n3 in numbers:
n1 = int(n1)
n2 = int(n2)
n3 = int(n3)
if (n1 + n2 + n3
== 2020) and (l1 != l2) and (l1 != l3) and (l2 != l3):
print("%d + %d + %d = 2020. %d x %d x %d = %d" %
(n1, n2, n3, n1, n2, n3, n1 * n2 * n3))
return
l3 += 1
l2 += 1
l1 += 1
numbers = core.load_data("1.txt")
find_2(numbers)
find_3(numbers)
import core
def check(trees, dx=3, dy=1):
total = 0
x = dx
width = len(trees[0])
for y in range(dy, len(trees), dy):
if trees[y][x] == "#":
total += 1
x += dx
x %= width
return total
trees = core.load_data("3.txt")
total1 = check(trees, 1, 1)
total2 = check(trees, 3, 1)
print("%d for %d, %d" % (total2, 3, 1))
total3 = check(trees, 5, 1)
total4 = check(trees, 7, 1)
total5 = check(trees, 1, 2)
total = total1 * total2 * total3 * total4 * total5
print("%d trees in all\n" % total)
while l < max:
matches = re.match('^(nop|acc|jmp) \+*(\-*\d+)$', data[l])
done.append(l)
if matches.group(1) == "acc":
a += int(matches.group(2))
l += 1
elif matches.group(1) == "jmp":
if l == j:
l += 1
else:
l += int(matches.group(2))
elif matches.group(1) == "nop":
l += 1
if l in done:
if alter:
l = 999999
else:
print("Line #%d has already run. Accumulator is %d" %
(l, a))
return
if l < 999999:
print("Code complete. Accumulator is %d" % a)
return
data = core.load_data("8.txt")
test(data, False)
test(data, True)
def check_1(data):
total = 0
for group in data:
combined = "".join(group)
cntr = Counter(combined)
total += len(cntr.items())
print(total)
def check_2(data):
total = 0
for group in data:
cntr = Counter(group[0])
shared = cntr.items()
for answers in group:
cntr = Counter(answers)
shared = list(set(shared) & set(cntr.items()))
total += len(shared)
print(total)
data = core.load_data("6.txt")
data = get_groups(data)
check_1(data)
check_2(data)
def total_bags(bag, bags):
total = 0
for child in bag:
if child in bags:
number = bag[child]
total += number + (number * total_bags(bags[child], bags))
return total
def check_1(bags):
total = 0
for colour in bags:
can = can_contain(bags[colour], bags)
if can:
total += 1
print("%d matching" % total)
def check_2(bags):
total = total_bags(bags["shiny gold"], bags)
print("%d in total" % total)
data = core.load_data("7.txt")
bags = get_bags(data)
check_1(bags)
check_2(bags)
if matches == None:
continue
else:
count = matches.group(4).count(matches.group(3))
valid = (count >= int(matches.group(1))) and (count <= int(matches.group(2)))
if valid:
total += 1
print("%s are valid" % total)
def validate_2(numbers):
total = 0
for line in numbers:
matches = re.search('^(\d+)\-(\d+) (\w): (\w+)$', line)
if matches == None:
continue
else:
char1 = matches.group(4)[int(matches.group(1)) - 1]
char2 = matches.group(4)[int(matches.group(2)) - 1]
valid1 = char1 == matches.group(3)
valid2 = char2 == matches.group(3)
valid = valid1 ^ valid2
if valid:
total += 1
print("%s are valid" % total)
passwords = core.load_data("2.txt")
validate_1(passwords);
validate_2(passwords);
m = re.match('^(\d+)(cm|in)$', match[1])
if (m == None) or (int(
m.group(1)) < hgt[m.group(2)][0]) or (int(
m.group(1)) > hgt[m.group(2)][1]):
valid = False
elif match[0] == "hcl":
m = re.match('^#([\da-f]{6})$', match[1])
if m == None:
valid = False
elif match[0] == "ecl":
if not match[1] in [
'amb', 'blu', 'brn', 'gry', 'grn', 'hzl', 'oth'
]:
valid = False
elif match[0] == "pid":
m = re.match('^\d{9}$', match[1])
if m == None:
valid = False
if valid:
total += 1
print("%d valid" % total)
data = core.load_data("4.txt")
data = get_records(data)
validate_1(data)
validate_2(data)
import sys
import core
league = sys.argv[1]
name, started, data = core.load_data(league)
print 'Enter results (Return or 0 for exit).'
while True:
team_games_wins = raw_input('Enter team (p1/p2-games-wins): ')
if team_games_wins in ['', '0']:
break
# quite painful way to seperate user input to relevant fields
# maybe there is a better way
team, games, wins = team_games_wins.split('-')
games = int(games)
wins = int(wins)
assert(wins <= games), 'Can not have more wins than games.'
# here we sort the players by alphabet to avoid duplicating entries
# with first and second player interchanged
first_player, second_player = sorted(team.upper().split('/'))
# check whether this combination exists
data = core.add_entry_to_data(data, first_player, second_player, games, wins)
core.store_data(name, started, data)
