def plot_dist(yvals_f, var, bin, label, outpath, sample):
hists_gen = get_distribution(yvals_f, "gen", bin, var)
# hists_cand = get_distribution(yvals_f, "cand", bin, var)
hists_pred = get_distribution(yvals_f, "pred", bin, var)
plt.figure()
ax = plt.axes()
v1 = mplhep.histplot(
[h[bh.rebin(2)] for h in hists_gen],
stack=True,
label=[class_names[k] for k in [13, 11, 22, 1, 2, 130, 211]],
lw=1)
mplhep.histplot(
[h[bh.rebin(2)] for h in hists_pred],
stack=True,
color=[x.stairs.get_edgecolor() for x in v1],
lw=2,
histtype="errorbar",
)
legend1 = plt.legend(v1, [x.legend_artist.get_label() for x in v1],
loc=(0.60, 0.44),
title="true")
# legend2 = plt.legend(v2, [x.legend_artist.get_label() for x in v1], loc=(0.8, 0.44), title="pred")
plt.gca().add_artist(legend1)
plt.ylabel("Total number of particles / bin")
cms_label(ax)
sample_label(sample, ax)
plt.yscale("log")
plt.ylim(top=1e9)
plt.xlabel(f"PFCandidate {label} [GeV]")
plt.savefig(f"{outpath}/pfcand_{var}.pdf", bbox_inches="tight")
plt.close()
def test_rebin_1d():
h = bh.Histogram(bh.axis.Regular(20, 1, 5))
h.fill(1.1)
hs = h[{0: slice(None, None, bh.rebin(4))}]
assert_array_equal(hs.view(), [1, 0, 0, 0, 0])
hs = h[{0: bh.rebin(4)}]
assert_array_equal(hs.view(), [1, 0, 0, 0, 0])
def test_rebin_nd():
h = bh.Histogram(
bh.axis.Regular(20, 1, 3), bh.axis.Regular(30, 1, 3), bh.axis.Regular(40, 1, 3)
)
s = bh.tag.Slicer()
assert h[{0: s[:: bh.rebin(2)]}].axes.size == (10, 30, 40)
assert h[{1: s[:: bh.rebin(2)]}].axes.size == (20, 15, 40)
assert h[{2: s[:: bh.rebin(2)]}].axes.size == (20, 30, 20)
assert h[{0: s[:: bh.rebin(2)], 2: s[:: bh.rebin(2)]}].axes.size == (10, 30, 20)
assert h[{1: s[:: bh.sum]}].axes.size == (20, 40)
def test_mix_value_with_slice_2():
h = bh.Histogram(bh.axis.Regular(10, 0, 10), bh.axis.Regular(10, 0, 10),
bh.axis.Integer(0, 2))
vals = np.arange(100).reshape(10, 10)
h[:, :, True] = vals
assert h[0, 1, True] == 1
assert h[1, 0, True] == 10
assert h[1, 1, True] == 11
assert h[3, 4, False] == 0
assert_array_equal(h[:, :, True].view(), vals)
assert_array_equal(h[:, :, False].view(), 0)
h2 = h[bh.rebin(2), bh.rebin(5), :]
assert_array_equal(h2.shape, (5, 2, 2))
def _step_shortcut(x: Any) -> Any:
"""
Convert some specific indices to step.
"""
if not isinstance(x, complex):
return x
if x.real != 0:
raise ValueError("The step should not have real part")
if x.imag % 1 != 0:
raise ValueError("The imaginary part should be an integer")
return bh.rebin(int(x.imag))
def test_get_1D_slice():
h1 = bh.Histogram(bh.axis.Regular(10, 0, 1))
h2 = bh.Histogram(bh.axis.Regular(5, 0, 0.5))
h1.metadata = {"that": 3}
h1.fill([0.25, 0.25, 0.25, 0.15])
h2.fill([0.25, 0.25, 0.25, 0.15])
assert h1 != h2
assert h1[:5] == h2
assert h1[:bh.loc(0.5)] == h2
assert h1[2:4] == h2[2:4]
assert h1[bh.loc(0.2):bh.loc(0.4)] == h2[bh.loc(0.2):bh.loc(0.4)]
assert len(h1[2:4].view()) == 2
assert len(h1[2:4:bh.rebin(2)].view()) == 1
assert len(h1[::bh.rebin(2)].view()) == 5
# Shortcut
assert len(h1[bh.rebin(2)].view()) == 5
assert h1[2:4].metadata == {"that": 3}
def test_repr():
assert repr(bh.loc(2)) == "loc(2)"
assert repr(bh.loc(3) + 1) == "loc(3) + 1"
assert repr(bh.loc(1) - 2) == "loc(1) - 2"
assert repr(bh.underflow) == "underflow"
assert repr(bh.underflow + 1) == "underflow + 1"
assert repr(bh.underflow - 1) == "underflow - 1"
assert repr(bh.overflow) == "overflow"
assert repr(bh.overflow + 1) == "overflow + 1"
assert repr(bh.overflow - 1) == "overflow - 1"
assert repr(bh.rebin(2)) == "rebin(2)"
def test_shrink_rebin_1d():
h = bh.Histogram(bh.axis.Regular(20, 0, 4))
h.fill(1.1)
hs = h[{0: slice(bh.loc(1), bh.loc(3), bh.rebin(2))}]
assert_array_equal(hs.view(), [1, 0, 0, 0, 0])
def __init__(self,
files,
observable="nCleaned_Cands",
name="QCD",
channel="",
kfactor=1.0,
ptype="background",
luminosity=1.0,
rebin=1,
rebin_piecewise=[],
normalise=True,
xsections=None,
mergecat=True,
binrange=None):
self._files = files
self.observable = observable
self.name = name
self.ptype = ptype
self.lumi = luminosity
self.xsec = xsections
self.outfile = None
self.channel = channel
self.nominal = {}
self.systvar = set()
self.rebin = rebin
self.rebin_piecewise = np.array(rebin_piecewise).astype(np.float)
self.binrange = binrange # dropping bins the same way as droping elements in numpy arrays a[1:3]
for fn in self._files:
_proc = os.path.basename(fn).replace(".root", "")
_file = uproot.open(fn)
if not _file:
raise ValueError("%s is not a valid rootfile" % self.name)
histograms = None
_scale = 1
if ptype.lower() != "data":
_scale = self.lumi * 1000.0 #
for name in _file.keys():
name = name.replace(";1", "")
if name != self.observable: continue
roothist = _file[name]
newhist = roothist.to_boost() * _scale
#### merge bins
if self.rebin >= 1 and newhist.values(
).ndim == 1: #written only for 1D right now
newhist = newhist[::bh.rebin(self.rebin)]
####merge bins to specified array
if len(self.rebin_piecewise) != 0 and newhist.values(
).ndim == 1: #written only for 1D right now
current_bins, current_edges = newhist.to_numpy()
#current_bins = newhist.axes.edges #Confused Chad. This returns a 'boost_histogram.axis.ArrayTuple' which cannot be cast into np,array?
new_freq = self.rebin_piecewise_constant(
current_edges, newhist.values(), self.rebin_piecewise)
new_variances = self.rebin_piecewise_constant(
current_bins, newhist.variances(),
self.rebin_piecewise)
newhist = bh.Histogram(bh.axis.Variable(
self.rebin_piecewise),
storage=bh.storage.Weight())
newhist[:] = np.stack([new_freq, new_variances], axis=-1)
newhist.name = name
if name in self.nominal.keys():
self.nominal[name] += newhist
else:
self.nominal[name] = newhist
try:
self.systvar.add(re.search("sys_[\w.]+", name).group())
except:
pass
self.merged = {}
self.merged = {i: (i, c) for i, c in self.nominal.items()}
histname='h1_phoPt'
Entries = hist_dict[histname].sum(flow=True)
Lumi=150000
## --Normalize histogram
print("Number of events: ",Entries)
weight = Lumi / Entries # In this case, the hist already weighted by xsec
#weight = 1
## --Rebining or Slicing
rebin=10
print(hist_dict[histname][::bh.rebin(rebin)])
hist_obj = hist_dict[histname][::bh.rebin(rebin)]
bin_width = hist_obj.axes[0].widths[0]
## --Draw hist
mplhep.histplot(hist_obj*weight,label='%s %d' % ("Entries: ",Entries))
plt.title(histname)
# Log scale
plt.yscale('log')
# Label name
xlabel_name = histname.split('_')[1] + '[GeV]'
ylabel_name = "Number of events / %3.1f GeV" %(bin_width)
plt.xlabel(xlabel_name)
plt.ylabel(ylabel_name)
