def test_nested_list(self):
in_list = [[1, 2, 3, 4], [5, 6, 7, 8], [9, 10]]
flat_list = list(mh.flatten(in_list))
self.assertEqual(flat_list, range(1, 11))
in_list = [1, [[2, 3], [4, 5]], 6, 7, [8, 9, [10]]]
self.assertEqual(flat_list, range(1, 11))
def main(args):
"""Main"""
# In order to not have to load the whole file first determine the names of
# the branches that are necessary
var_names = [('costh', 'phi')]
if args.genlevel:
logging.info('Also adding generator level folding')
var_names.append(('gen_costh', 'gen_phi'))
frames = args.frames.split(',')
load_variables = ['_'.join(p) for p in product(flatten(var_names), frames)]
for infile in args.inputfiles:
logging.info('Processing file {}'.format(infile))
if not args.treename:
treename = get_treename(infile)
else:
treename = args.treename
df = get_dataframe(infile, treename, columns=load_variables)
for var_pair in var_names:
for frame in frames:
costh_f, phi_f = get_folded_frame(df, frame, *var_pair)
add_branch(costh_f, '_'.join([var_pair[0], frame, 'fold']),
infile, treename)
add_branch(phi_f, '_'.join([var_pair[1], frame, 'fold']),
infile, treename)
def shift_graphs(graphs, abs_shift, offset=1):
"""
Shift all graphs slightly along the horizontal direction.
If offset==0 the first graph will remain unchanged and the others will be
shifted symmetrically around this one, otherwise all the graphs will just be
distributed symmetrically around the 0
"""
# Just define 17 shifts and hope that this will be enough for all variations
shifts = list(flatten([0] + [[-abs_shift * i, abs_shift * i] for i in xrange(1, 8)]))
return [shift_graph_horizontal(g, shifts[i + offset]) for i, g in enumerate(graphs)]
def main(args):
"""Main"""
data = get_dataframe(args.scanfile, columns=list(flatten(args.variables)))
outfile = r.TFile.Open(args.outfile, 'recreate')
for varx, vary in args.variables:
corr_graphs = make_pair_correlation(data, varx, vary, args.physical_lambdas)
for cg in corr_graphs:
cg.Write()
outfile.Close()
def create_muon_pz_res_map(dfr, n_bins_res, n_bins_p):
"""
See above, same thing for muon pz
"""
pz_bins = get_equi_pop_bins(dfr, lambda df: df.gen_muPPz.abs(), n_bins_p)
pz_bins = np.array(sorted({b for b in flatten(pz_bins)}))
res_bins = np.linspace(-0.15, 0.15, n_bins_res)
hist_sett = (len(pz_bins) - 1, pz_bins, len(res_bins) - 1, res_bins)
pz_p_map = create_res_v_gen_map(dfr, 'muPPz', hist_sett)
pz_n_map = create_res_v_gen_map(dfr, 'muNPz', hist_sett)
muon_map = pz_p_map.Clone()
muon_map.Add(pz_n_map)
return muon_map
def get_eta_binning(effs):
"""
Get the eta binning from the efficiencies and also check that the eta range
is contiguous throughout the whole range
"""
# first get all unique bin borders than check if all adjacent pairs are
# also present in the efficiencies
# to facilitate the task sort the unique borders
uniq_bin_bord = sorted({b for b in flatten(effs)})
possible_bins = zip(uniq_bin_bord[:-1], uniq_bin_bord[1:])
for pbin in possible_bins:
if pbin not in effs:
logging.error(
'The eta binning is not contiguous! The possible bin '
'{:.1f} - {:.1f} is not present'.format(*pbin))
return None
return uniq_bin_bord
def create_muon_pxy_res_map(dfr, n_bins_res, n_bins_p):
"""
Create the muon px and py combined residual maps (see photon for binning)
"""
px_bins = get_equi_pop_bins(dfr, lambda df: df.gen_muPPx.abs(), n_bins_p)
px_bins = np.array(sorted({b for b in flatten(px_bins)}))
res_bins = np.linspace(-0.15, 0.15, n_bins_res)
hist_sett = (len(px_bins) - 1, px_bins, len(res_bins) - 1, res_bins)
px_p_map = create_res_v_gen_map(dfr, 'muPPx', hist_sett)
px_n_map = create_res_v_gen_map(dfr, 'muNPx', hist_sett)
py_p_map = create_res_v_gen_map(dfr, 'muPPy', hist_sett)
py_n_map = create_res_v_gen_map(dfr, 'muNPy', hist_sett)
muon_map = px_n_map.Clone()
for mu_map in [px_n_map, py_p_map, py_n_map]:
muon_map.Add(mu_map)
return muon_map
def compile_load_vars(selections, varx, vary):
"""Compile the list of variables that has to be loaded"""
vars_to_load = [
sf.collect_requirements(sel[0]) for sel in selections.values()
]
if any(sel[1] is not None for sel in selections.values()):
# simply load all efficiencies in case they are needed
vars_to_load.append('*eff_sm')
vars_to_load.append('costh_HX') # needed to bin in costh_HX
if varx in FUNCVARS:
vars_to_load.append(FUNCVARS[varx].requires)
else:
vars_to_load.append(varx)
if vary in FUNCVARS:
vars_to_load.append(FUNCVARS[vary].requires)
else:
vars_to_load.append(vary)
return list(set(flatten(vars_to_load))) # remove duplicates
def create_photon_res_map(dfr, n_bins_res, n_bins_p):
"""
Create the photon residuals map for a given number of resolution bins (equal
spacing) and a given number of P(x,y,z) bins (equal population)
"""
# determine the P(x,y,x) binning
px_bins = get_equi_pop_bins(dfr, lambda df: df.gen_photonPx.abs(),
n_bins_p)
# get_equi_pop_bins returns bins in an unsuitable format transform them to
# proper format
px_bins = np.array(sorted({b for b in flatten(px_bins)}))
res_bins = np.linspace(-2, 2, n_bins_res)
hist_sett = (len(px_bins) - 1, px_bins, len(res_bins) - 1, res_bins)
px_map = create_res_v_gen_map(dfr, 'photonPx', hist_sett)
py_map = create_res_v_gen_map(dfr, 'photonPy', hist_sett)
pz_map = create_res_v_gen_map(dfr, 'photonPz', hist_sett)
photon_map = px_map.Clone()
photon_map.Add(py_map)
photon_map.Add(pz_map)
return photon_map
def get_all_variations():
"""
Get all the possible variations
"""
return (list(flatten(v)) for v in product(*VARIATIONS))
def test_chunks_preserves_order(self):
in_list = range(100)
chunk_size = 20
self.assertEqual(list(mh.flatten(mh.chunks(in_list, chunk_size))),
in_list)
def test_already_flat(self):
in_list = list(xrange(10))
flat_list = list(mh.flatten(in_list))
self.assertEqual(flat_list, in_list)
