def draw_syst(self, save_dir, results_dir, root_dir=None):
"""Draw all fits one-by-one
Args:
save_dir: directory where to save plots
results_dir: where to find intermediate results of the multi trial
root_dir: TDirectory where to save summary plots
"""
gStyle.SetOptStat(0)
gStyle.SetOptStat(0000)
gStyle.SetPalette(1)
gStyle.SetCanvasColor(0)
gStyle.SetFrameFillColor(0)
bins1_ranges = self.pars_factory.bins1_edges_low.copy()
bins1_ranges.append(self.pars_factory.bins1_edges_up[-1])
for (ibin1, ibin2), fit in self.syst_fits.items():
if not fit:
self.logger.warning(
"No systematic fit for bins (%i, %i). Skip...", ibin1,
ibin2)
continue
bin_id_match = self.pars_factory.bin_matching[ibin1]
# Some variables set for drawing
title = f"{self.pars_factory.bins1_edges_low[ibin1]:.1f} < #it{{p}}_{{T}} < " \
f"{self.pars_factory.bins1_edges_up[ibin1]:.1f}" \
f"(prob > {self.pars_factory.prob_cut_fin[bin_id_match]:.2f})"
suffix_write = self.pars_factory.make_suffix(ibin1, ibin2)
fit.results_path = os.path.join(
results_dir, f"multi_trial_bin1_{ibin1}_bin2_{ibin2}.root")
# Central fits
canvas = TCanvas("fit_canvas", suffix_write, 1400, 800)
fit.draw(canvas, title=title)
if self.pars_factory.apply_weights is False:
canvas.SaveAs(
make_file_path(save_dir, "multi_trial", "eps", None,
suffix_write))
else:
canvas.SaveAs(
make_file_path(save_dir, "multi_trial_weights", "eps",
None, suffix_write))
if root_dir:
root_dir.cd()
canvas.Write(f"multi_trial_{suffix_write}")
canvas.Close()
def draw_fits(self, save_dir, root_dir=None):
"""
Draw all fits one-by-one
Args:
save_dir: directory where to save plots
root_dir: TDirectory where to save summary plots
"""
gStyle.SetOptStat(0)
gStyle.SetOptStat(0000)
gStyle.SetPalette(1)
gStyle.SetCanvasColor(0)
gStyle.SetFrameFillColor(0)
bins2 = self.get_bins2()
bins1_ranges = self.pars_factory.bins1_edges_low.copy()
bins1_ranges.append(self.pars_factory.bins1_edges_up[-1])
n_bins1 = len(bins1_ranges) - 1
# Summarize in mult histograms in pT bins
yieldshistos = {ibin2: TH1F("hyields%d" % (ibin2), "", \
n_bins1, array("d", bins1_ranges)) for ibin2 in bins2}
means_histos = {ibin2:TH1F("hmeanss%d" % (ibin2), "", \
n_bins1, array("d", bins1_ranges)) for ibin2 in bins2}
sigmas_histos = {ibin2: TH1F("hsigmas%d" % (ibin2), "", \
n_bins1, array("d", bins1_ranges)) for ibin2 in bins2}
have_summary_pt_bins = []
means_init_mc_histos = TH1F("hmeans_init_mc", "", n_bins1,
array("d", bins1_ranges))
sigmas_init_mc_histos = TH1F("hsigmas_init_mc", "", n_bins1,
array("d", bins1_ranges))
means_init_data_histos = TH1F("hmeans_init_data", "", n_bins1,
array("d", bins1_ranges))
sigmas_init_data_histos = TH1F("hsigmas_init_data", "", n_bins1,
array("d", bins1_ranges))
nx = 4
ny = 2
canvy = 533
if n_bins1 > 12:
nx = 5
ny = 4
canvy = 1200
elif n_bins1 > 8:
nx = 4
ny = 3
canvy = 800
canvas_init_mc = TCanvas("canvas_init_mc", "MC", 1000, canvy)
canvas_init_data = TCanvas("canvas_init_data", "Data", 1000, canvy)
canvas_data = {ibin2: TCanvas("canvas_data%d" % (ibin2), "Data", 1000, canvy) \
for ibin2 in bins2}
canvas_init_mc.Divide(nx, ny)
canvas_init_data.Divide(nx, ny)
for c in canvas_data.values():
c.Divide(nx, ny)
# Need to cache some object for which the canvas is only written after the loop...
for (ibin1, ibin2), fit in self.central_fits.items():
bin_id_match = self.pars_factory.bin_matching[ibin1]
# Some variables set for drawing
title = f"{self.pars_factory.bins1_edges_low[ibin1]:.1f} < #it{{p}}_{{T}} < " \
f"{self.pars_factory.bins1_edges_up[ibin1]:.1f}" \
f"(prob > {self.pars_factory.prob_cut_fin[bin_id_match]:.2f})"
x_axis_label = "#it{M}_{inv} (GeV/#it{c}^{2})"
n_sigma_signal = self.pars_factory.n_sigma_signal
suffix_write = self.pars_factory.make_suffix(ibin1, ibin2)
kernel = fit.kernel
histo = fit.histo
# Central fits
y_axis_label = \
f"Entries/({histo.GetBinWidth(1) * 1000:.0f} MeV/#it{{c}}^{{2}})"
canvas = TCanvas("fit_canvas", suffix_write, 700, 700)
fit.draw(canvas,
sigma_signal=n_sigma_signal,
x_axis_label=x_axis_label,
y_axis_label=y_axis_label,
title=title)
if self.pars_factory.apply_weights is False:
canvas.SaveAs(
make_file_path(save_dir, "fittedplot", "eps", None,
suffix_write))
else:
canvas.SaveAs(
make_file_path(save_dir, "fittedplotweights", "eps", None,
suffix_write))
canvas.Close()
fit.draw(canvas_data[ibin2].cd(ibin1 + 1),
sigma_signal=n_sigma_signal,
x_axis_label=x_axis_label,
y_axis_label=y_axis_label,
title=title)
if fit.success:
# Only fill these summary plots in case of success
yieldshistos[ibin2].SetBinContent(ibin1 + 1,
kernel.GetRawYield())
yieldshistos[ibin2].SetBinError(ibin1 + 1,
kernel.GetRawYieldError())
means_histos[ibin2].SetBinContent(ibin1 + 1, kernel.GetMean())
means_histos[ibin2].SetBinError(ibin1 + 1,
kernel.GetMeanUncertainty())
sigmas_histos[ibin2].SetBinContent(ibin1 + 1,
kernel.GetSigma())
sigmas_histos[ibin2].SetBinError(ibin1 + 1,
kernel.GetSigmaUncertainty())
# Residual plot
c_res = TCanvas('cRes', 'The Fit Canvas', 800, 800)
c_res.cd()
h_pulls = histo.Clone(f"{histo.GetName()}_pull")
h_residual_trend = histo.Clone(
f"{histo.GetName()}_residual_trend")
h_pulls_trend = histo.Clone(f"{histo.GetName()}_pulls_trend")
_ = kernel.GetOverBackgroundResidualsAndPulls( \
h_pulls, h_residual_trend, h_pulls_trend, \
self.pars_factory.fit_range_low[ibin1], \
self.pars_factory.fit_range_up[ibin1])
h_residual_trend.Draw()
c_res.SaveAs(
make_file_path(save_dir, "residual", "eps", None,
suffix_write))
c_res.Close()
# Summary plots to be done only once per pT bin
if ibin1 in have_summary_pt_bins:
continue
have_summary_pt_bins.append(ibin1)
# Pre-fit MC
suffix_write = self.pars_factory.make_suffix(
ibin1, self.pars_factory.bins2_int_bin)
pre_fit_mc = self.pre_fits_mc[ibin1]
kernel = pre_fit_mc.kernel
histo = pre_fit_mc.histo
y_axis_label = \
f"Entries/({histo.GetBinWidth(1) * 1000:.0f} MeV/#it{{c}}^{{2}})"
canvas = TCanvas("fit_canvas_mc_init", suffix_write, 700, 700)
pre_fit_mc.draw(canvas,
x_axis_label=x_axis_label,
y_axis_label=y_axis_label,
title=title)
canvas.SaveAs(
make_file_path(save_dir, "fittedplot_integrated_mc", "eps",
None, suffix_write))
canvas.Close()
pre_fit_mc.draw(canvas_init_mc.cd(ibin1 + 1),
x_axis_label=x_axis_label,
y_axis_label=y_axis_label,
title=title)
if pre_fit_mc.success:
# Only fill these summary plots in case of success
means_init_mc_histos.SetBinContent(ibin1 + 1,
kernel.GetParameter(1))
means_init_mc_histos.SetBinError(ibin1 + 1,
kernel.GetParError(1))
sigmas_init_mc_histos.SetBinContent(ibin1 + 1,
kernel.GetParameter(2))
sigmas_init_mc_histos.SetBinError(ibin1 + 1,
kernel.GetParError(2))
pre_fit_data = self.pre_fits_data[ibin1]
kernel = pre_fit_data.kernel
histo = pre_fit_data.histo
# Pre-fit data
y_axis_label = \
f"Entries/({histo.GetBinWidth(1) * 1000:.0f} MeV/#it{{c}}^{{2}})"
canvas = TCanvas("fit_canvas_data_init", suffix_write, 700, 700)
pre_fit_data.draw(canvas,
sigma_signal=n_sigma_signal,
x_axis_label=x_axis_label,
y_axis_label=y_axis_label,
title=title)
canvas.SaveAs(
make_file_path(save_dir, "fittedplot_integrated", "eps", None,
suffix_write))
canvas.Close()
pre_fit_data.draw(canvas_init_data.cd(ibin1 + 1),
sigma_signal=n_sigma_signal,
x_axis_label=x_axis_label,
y_axis_label=y_axis_label,
title=title)
if pre_fit_data.success:
# Only fill these summary plots in case of success
means_init_data_histos.SetBinContent(ibin1 + 1,
kernel.GetMean())
means_init_data_histos.SetBinError(ibin1 + 1,
kernel.GetMeanUncertainty())
sigmas_init_data_histos.SetBinContent(ibin1 + 1,
kernel.GetSigma())
sigmas_init_data_histos.SetBinError(
ibin1 + 1, kernel.GetSigmaUncertainty())
canvas_init_mc.SaveAs(make_file_path(save_dir, "canvas_InitMC", "eps"))
canvas_init_mc.Close()
canvas_init_data.SaveAs(
make_file_path(save_dir, "canvas_InitData", "eps"))
canvas_init_data.Close()
for ibin2 in bins2:
suffix2 = f"ibin2_{ibin2}"
canvas_data[ibin2].SaveAs(
make_file_path(save_dir, "canvas_FinalData", "eps", None,
suffix2))
if root_dir:
root_dir.cd()
yieldshistos[ibin2].Write()
means_histos[ibin2].Write()
sigmas_histos[ibin2].Write()
#canvas_data[ibin2].Close()
latex_bin2_var = self.ana_config["latexbin2var"]
latex_hadron_name = self.ana_config["latexnamehadron"]
# Plot some summary historgrams
leg_strings = [f"{self.pars_factory.bins2_edges_low[ibin2]} #leq {latex_bin2_var} < " \
f"{self.pars_factory.bins2_edges_up[ibin2]}" for ibin2 in bins2]
save_name = make_file_path(save_dir, "Yields", "eps", None,
[self.case, self.ana_type])
# Yields summary plot
plot_histograms(
[yieldshistos[ibin2] for ibin2 in bins2], True, True, leg_strings,
"uncorrected yields", "#it{p}_{T} (GeV/#it{c})",
f"Uncorrected yields {latex_hadron_name} {self.ana_type}",
"mult. / int.", save_name)
save_name = make_file_path(save_dir, "Means", "eps", None,
[self.case, self.ana_type])
# Means summary plot
plot_histograms([means_histos[ibin2] for ibin2 in bins2], False, True,
leg_strings, "Means", "#it{p}_{T} (GeV/#it{c})",
"#mu_{fit} " + f"{latex_hadron_name} {self.ana_type}",
"mult. / int.", save_name)
save_name = make_file_path(save_dir, "Sigmas", "eps", None,
[self.case, self.ana_type])
#Sigmas summary plot
plot_histograms([sigmas_histos[ibin2]
for ibin2 in bins2], False, True, leg_strings,
"Sigmas", "#it{p}_{T} (GeV/#it{c})", "#sigma_{fit} " +
f"{latex_hadron_name} {self.ana_type}", "mult. / int.",
save_name)
# Plot the initialized means and sigma for MC and data
save_name = make_file_path(save_dir, "Means_mult_int", "eps", None,
[self.case, self.ana_type])
plot_histograms([means_init_mc_histos, means_init_data_histos], False,
False, ["MC", "data"], "Means of int. mult.",
"#it{p}_{T} (GeV/#it{c})",
"#mu_{fit} " + f"{latex_hadron_name} {self.ana_type}",
"", save_name)
save_name = make_file_path(save_dir, "Sigmas_mult_int", "eps", None,
[self.case, self.ana_type])
plot_histograms([sigmas_init_mc_histos, sigmas_init_data_histos],
False, False, ["MC", "data"], "Sigmas of int. mult.",
"#it{p}_{T} (GeV/#it{c})", "#sigma_{fit} " +
f"{latex_hadron_name} {self.ana_type}", "", save_name)