def test_slicing(self, dhdl):
'''Test if extract_u_nk assign the attr correctly'''
dataset = load_benzene()
u_nk = extract_u_nk(dataset['data']['Coulomb'][0], 310)
new_u_nk = slicing(u_nk)
assert new_u_nk.attrs['temperature'] == 310
assert new_u_nk.attrs['energy_unit'] == 'kT'
def test_slicing_inefficiency_equivalence(
self, dataloader, lower, upper, conservative, request
):
"""
Test that first slicing the data frame, then subsampling is equivalent to
subsampling with lower / upper bounds set
"""
# Load data
data = request.getfixturevalue(dataloader)
# Check that the input data is appropriate for the test
_check_data_is_outside_bounds(data, lower, upper)
# Slice dataframe, then subsample it based on the sum of its components
sliced_data = slicing(data, lower=lower, upper=upper)
subsampled_sliced_data = self.slicer(sliced_data,
series=sliced_data.sum(axis=1),
conservative=conservative)
# Subsample the dataframe based on the sum of its components while
# also specifying the slicing range
subsampled_data = self.slicer(data,
series=data.sum(axis=1),
lower=lower,
upper=upper,
conservative=conservative)
assert (subsampled_sliced_data == subsampled_data).all(axis=None)
def get_dHdl(sim, lower=None, upper=None, step=None):
try:
s = time.time()
dHdl = sim.data.retrieve('dHdl')
e = time.time()
if dHdl is None:
dHdl =
delayed(slicing_delayed)(delayed(pd.concat)([delayed(get_dHdl_XVG_delayed)(xvg)
for xvg in sim['WORK/dhdl/'].glob('*.xvg')]), lower=lower,
upper=upper, step=step)
else:
logging.info("get_dHdl,hdf5_read,{},{},{},{}".format(sim.name, e - s, s, e))
dHdl = slicing(dHdl.sort_index(0),
lower=lower,
upper=upper,
step=step)
except:
# THIS WILL NOT STORE THE VALUE FOR LATER USE SO YOU SHOULD REALLY
# CONTINUOUSLY UPDATE THE dHdl DATA IN THE SIMS
dHdl =
delayed(slicing_delayed)(delayed(pd.concat)([delayed(get_dHdl_XVG_delayed)(xvg)
for xvg in sim['WORK/dhdl/'].glob('*.xvg')]), lower=lower,
upper=upper, step=step)
return dHdl
def test_no_series(self, data):
"""Check that we get the same result as simple slicing with no Series.
"""
df_sub = self.slicer(data, lower=200, upper=5000, step=2)
df_sliced = slicing(data, lower=200, upper=5000, step=2)
assert np.all((df_sub == df_sliced))
def slicer(self, *args, **kwargs):
return slicing(*args, **kwargs)
def slicing_delayed(dhdls, lower=None, upper=None, step=None):
return slicing(dhdls.sort_index(0), lower=lower, upper=upper, step=step)