def test_getNewVals_for_newuncs_dim():
"""
The output dimension of newRVs should be the same as the
input dimension of newtimes
"""
newtimes = np.arange(10)
times = np.random.uniform(0, 10, 100)
rvs = np.random.uniform(-5, 5, 100)
uncs = np.random.normal(loc=1., scale=0.5, size=100)
newRVs, newUncs = wm.getNewVals(newtimes, times, rvs, uncs, timebin=1.)
assert len(newtimes) == len(newUncs)
def test_getNewVals_unc_magnitude():
"""
The median single measurement precision should be lower
for the binned data. This function ensures that to be the case.
"""
newtimes = np.arange(10)
times = np.random.uniform(0, 10, 100)
rvs = np.random.normal(loc=0, scale=5, size=100)
uncs = np.random.normal(loc=1., scale=0.5, size=100)
newRVs, newUncs = wm.getNewVals(newtimes, times, rvs, uncs, timebin=1.)
assert np.median(newUncs) < np.median(uncs)
def test_getNewVals_rv_scatter():
"""
The RV scatter (standard deviation from normally distributed points
about the mean should be reduced when binning observations down. This
routine checks that.
"""
newtimes = np.arange(10)
times = np.random.uniform(0, 10, 100)
rvs = np.random.normal(loc=0, scale=5, size=100)
uncs = np.random.normal(loc=1., scale=0.5, size=100)
newRVs, newUncs = wm.getNewVals(newtimes, times, rvs, uncs, timebin=1.)
assert np.std(newRVs) < np.std(rvs)
def test_big_gaps_getNewVals():
"""
Ensure getNewVals routine can handle big gaps in times
"""
timebin = 1.
times = np.concatenate((np.random.uniform(0, 10, 50),
np.random.uniform(30, 40, 50)))
newtimes = wm.getNewTimes(times, timebin)
rvs = np.random.normal(loc=0, scale=5, size=100)
uncs = np.random.normal(loc=1., scale=0.5, size=100)
newRVs, newUncs = wm.getNewVals(newtimes, times, rvs,
uncs, timebin=timebin)
fins = np.where(np.isfinite(newUncs))
newRVs = newRVs[fins]
newUncs = newUncs[fins]
newtimes = newtimes[fins]
assert np.median(newUncs) < np.median(uncs)
