def test_histograms_to_pandas_dataframe(self):
data = [1.0, 1.1, 1.2, 1.3, 2.0, 1.4, 2.3, 2.5, 3.1, 3.5]
# This length needs to be larger than 10 to see a difference between
# str ordering and int ordering
hists = [Histogram(n_bins=5) for i in range(11)]
for hist in hists:
_ = hist.histogram(data)
df = histograms_to_pandas_dataframe(hists)
# sort like is done in analysis
df = df.sort_index(axis=1)
# This breaks if the sorting is done based on strings as that will
# return [0, 1, 10 ...] instead of [0, 1, 2, ...]
for i, c in enumerate(df.columns):
assert str(c) == str(i)
def total_crossing_probability(self,
steps=None,
method="wham",
force=False):
"""Return the total crossing probability using `method`
Parameters
----------
steps : iterable of :class:`.MCStep`
cycles to be analyzed
method : "wham" (later: or "mbar" or "tram")
approach to use to combine the histograms
force : bool (False)
if true, cached results are overwritten
"""
if method == "wham":
run_ensembles = False
for ens in self.ensembles:
try:
hist = self.histograms['max_lambda'][ens]
except KeyError:
run_ensembles = True
if run_ensembles or force:
if steps is None:
raise RuntimeError(
"Unable to build histograms without steps source")
self.all_statistics(steps, force=True)
df = histograms_to_pandas_dataframe(
self.histograms['max_lambda'].values(),
fcn="reverse_cumulative").sort_index(axis=1)
# if lambdas not set, returns None and WHAM uses fallback
lambdas = self.interfaces.lambdas
wham = WHAM(interfaces=lambdas)
# wham.load_from_dataframe(df)
# wham.clean_leading_ones()
tcp = wham.wham_bam_histogram(df).to_dict()
elif method == "mbar":
pass
else:
raise ValueError(
"Only supported methods are 'wham' and 'mbar'. " +
"Whereas 'mbar' is not yet implemented!")
self.tcp = LookupFunction(tcp.keys(), tcp.values())
return self.tcp
def total_crossing_probability(self, steps=None, method="wham", force=False):
"""Return the total crossing probability using `method`
Parameters
----------
steps : iterable of :class:`.MCStep`
cycles to be analyzed
method : "wham" (later: or "mbar" or "tram")
approach to use to combine the histograms
force : bool (False)
if true, cached results are overwritten
"""
if method == "wham":
run_ensembles = False
for ens in self.ensembles:
try:
hist = self.histograms['max_lambda'][ens]
except KeyError:
run_ensembles = True
if run_ensembles or force:
if steps is None:
raise RuntimeError("Unable to build histograms without steps source")
self.all_statistics(steps, force=True)
df = histograms_to_pandas_dataframe(
self.histograms['max_lambda'].values(),
fcn="reverse_cumulative"
).sort_index(axis=1)
# if lambdas not set, returns None and WHAM uses fallback
lambdas = self.interfaces.lambdas
wham = WHAM(interfaces=lambdas)
# wham.load_from_dataframe(df)
# wham.clean_leading_ones()
tcp = wham.wham_bam_histogram(df).to_dict()
elif method == "mbar":
pass
else:
raise ValueError("Only supported methods are 'wham' and 'mbar'. "
+ "Whereas 'mbar' is not yet implemented!")
self.tcp = LookupFunction(tcp.keys(), tcp.values())
return self.tcp