def compute_best_frequencies(ids, n_eval=10000, n_retry=5, generalized=True):
results = {}
for i in ids:
t, y, dy = data[i].T
print " - computing power for %i (%i points)" % (i, len(t))
kwargs = dict(generalized=generalized)
omega, power = search_frequencies(t, y, dy, n_eval=n_eval,
n_retry=n_retry,
LS_kwargs=kwargs)
results[i] = [omega, power]
return results
def compute_best_frequencies(ids, n_eval=10000, n_retry=5, generalized=True):
results = {}
for i in ids:
t, y, dy = data[i].T
print " - computing power for %i (%i points)" % (i, len(t))
kwargs = dict(generalized=generalized)
omega, power = search_frequencies(t, y, dy, n_eval=n_eval,
n_retry=n_retry,
LS_kwargs=kwargs)
results[i] = [omega, power]
return results
def test_search_frequencies():
t = np.arange(0, 1E1, 0.01)
f = 1
w = 2 * np.pi * np.array(f)
y = np.sin(w * t)
dy = np.random.normal(0, 0.1, size=len(y))
y = y + dy
omegas, power = search_frequencies(t, y, dy)
omax = omegas[power == max(power)]
assert_almost_equal(w, omax, decimal=3)
def test_search_frequencies():
t = np.arange(0, 1E1, 0.01)
f = 1
w = 2*np.pi*np.array(f)
y = np.sin(w*t)
dy = np.random.normal(0, 0.1, size=len(y))
y = y + dy
omegas, power = search_frequencies(t, y, dy)
omax = omegas[power == max(power)]
assert_almost_equal(w, omax, decimal=3)
def test_search_frequencies():
rng = np.random.RandomState(0)
t = np.arange(0, 1E1, 0.01)
f = 1
w = 2 * np.pi * np.array(f)
y = np.sin(w*t)
dy = 0.01
y += dy * rng.randn(len(y))
omegas, power = search_frequencies(t, y, dy)
omax = omegas[power == max(power)]
assert_almost_equal(w, omax, decimal=3)
def test_search_frequencies():
rng = np.random.RandomState(0)
t = np.arange(0, 1E1, 0.01)
f = 1
w = 2 * np.pi * np.array(f)
y = np.sin(w * t)
dy = 0.01
y += dy * rng.randn(len(y))
omegas, power = search_frequencies(t, y, dy)
omax = omegas[power == max(power)]
assert_almost_equal(w, omax, decimal=3)
def compute_best_frequencies(windows, image='A', n_eval=10000, n_retry=5, generalized=True):
results = {}
for window in windows:
t = data['t_eval'][data['window_id']==window]
y = data['sig_eval%s'%(image)][data['window_id']==window]
y = y - np.mean(y)
dy = data['sig_err%s'%(image)][data['window_id']==window]
dy = dy - np.mean(dy)
print " - computing power for window %s (%s points)" % (window, len(t))
kwargs = dict(generalized=generalized)
omega, power = search_frequencies(t, y, dy,
n_eval=n_eval,
n_retry=n_retry,
LS_kwargs=kwargs)
results[window] = [omega, power]
return results
def compute_best_frequencies(windows,
image='A',
n_eval=10000,
n_retry=5,
generalized=True):
results = {}
for window in windows:
t = data['t_eval'][data['window_id'] == window]
y = data['sig_eval%s' % (image)][data['window_id'] == window]
y = y - np.mean(y)
dy = data['sig_err%s' % (image)][data['window_id'] == window]
dy = dy - np.mean(dy)
print " - computing power for window %s (%s points)" % (window, len(t))
kwargs = dict(generalized=generalized)
omega, power = search_frequencies(t,
y,
dy,
n_eval=n_eval,
n_retry=n_retry,
LS_kwargs=kwargs)
results[window] = [omega, power]
return results
cur.execute("SELECT * from Periods WHERE id = %i" % id)
res = cur.fetchall()
if len(res) > 0:
print res[0]
else:
print ("computing period for id = %i (%i / %i)"
% (id, count + 1, len(data.ids)))
lc = data[id]
t0 = time()
omega, power = search_frequencies(lc[:, 0], lc[:, 1], lc[:, 2],
LS_func=multiterm_periodogram,
n_save=5, n_retry=5,
n_eval=10000,
LS_kwargs=dict(n_terms=5))
omega_best = omega[np.argmax(power)]
t1 = time()
print " - execution time: %.2g sec" % (t1 - t0)
# insert value and commit to disk
cur.execute("INSERT INTO Periods VALUES(%i, %f)"
% (id, omega_best))
con.commit()
con.close()
#cur.execute("SELECT * from Periods")
#print cur.fetchall()
for count, id in enumerate(data.ids):
# only compute period if it hasn't been computed before
cur.execute("SELECT * from Periods WHERE id = %i" % id)
res = cur.fetchall()
if len(res) > 0:
print(res[0])
else:
print("computing period for id = {0} ({1} / {2})"
"".format(id, count + 1, len(data.ids))))
lc = data[id]
t0 = time()
omega, power = search_frequencies(lc[:, 0], lc[:, 1], lc[:, 2],
LS_func=multiterm_periodogram,
n_save=5, n_retry=5,
n_eval=10000,
LS_kwargs=dict(n_terms=5))
omega_best = omega[np.argmax(power)]
t1 = time()
print(" - execution time: %.2g sec" % (t1 - t0))
# insert value and commit to disk
cur.execute("INSERT INTO Periods VALUES(%i, %f)"
% (id, omega_best))
con.commit()
con.close()
res = cur.fetchall()
if len(res) > 0:
print res[0]
else:
print("computing period for id = %i (%i / %i)" %
(id, count + 1, len(data.ids)))
lc = data[id]
t0 = time()
omega, power = search_frequencies(lc[:, 0],
lc[:, 1],
lc[:, 2],
LS_func=multiterm_periodogram,
n_save=5,
n_retry=5,
n_eval=10000,
LS_kwargs=dict(n_terms=5))
omega_best = omega[np.argmax(power)]
t1 = time()
print " - execution time: %.2g sec" % (t1 - t0)
# insert value and commit to disk
cur.execute("INSERT INTO Periods VALUES(%i, %f)" %
(id, omega_best))
con.commit()
con.close()
#cur.execute("SELECT * from Periods")
