TTTT2/tau_studies.py

#!/usr/bin/env python
import numpy as np
import matplotlib.pyplot as plt

from filval.result_set import ResultSet
from filval.histogram import hist, hist2d, hist_add, hist_sub, hist_div, hist_norm, hist_scale, hist2d_norm
from filval.plotting import (decl_plot, render_plots, hist_plot, hist_plot_stack, hist2d_plot,
                             Plot, generate_dashboard, hists_to_table)


def set_xticks():
    x_ticks = [f'SR{i}' for i in range(1, 17)]
    plt.xticks([x for x in range(1, 17)], x_ticks, rotation=60)


def get_sr(rs, tau_category, tm=False):
    if tm:
        if tau_category == 0:
            return hist(rs.SRs_0tmtau)
        if tau_category == 1:
            return hist(rs.SRs_1tmtau)
        elif tau_category == 2:
            return hist(rs.SRs_2tmtau)
    else:
        if tau_category == -1:
            return hist(rs.ignore_tau_SRs)
        elif tau_category == 0:
            return hist(rs.SRs_0tau)
        elif tau_category == 1:
            return hist(rs.SRs_1tau)
        elif tau_category == 2:
            return hist(rs.SRs_2tau)


@decl_plot
def plot_yield_grid(tau_category=-1):
    r"""## Event Yield

    The event yield for the eight signal regions defined in AN-17-115. Data is normalized
    to the integrated luminosity of $35.9\textrm{fb}^{-1}$.

    Ignoring taus means both that there is no requirement on number of good taus *and*
    the taus, if present, are not considered for the SS pair.

    If taus are not ignored, any good tau with $p_T$>20Gev is considered in constructing the SS lepton pair. The yields
    are then further broken down by the number of good taus in the event.

    A "good" tau in the above means any tau candidate passing the `byTightIsolationMVArun2v1DBoldDMwLT` ID w/ pt>20GeV.
    It is also required to pass tau-lepton cross cleaning where it must not match any electron or muon within
    $\delta R < 0.4$.

    Truth-matched taus are those that match within $\delta R < 0.3$ with gen-level taus that pass the flag `fromHardProcessDecayed`.
    """

    _, ((ax_tttt, ax_ttw), (ax_ttz, ax_tth)) = plt.subplots(2, 2)
    tttt, ttw, ttz, tth = [get_sr(rs, tau_category) for rs in rss]
    tm_tttt, tm_ttw, tm_ttz, tm_tth = [get_sr(rs, tau_category, True) for rs in rss]

    plt.sca(ax_tttt)
    plt.title('TTTT')
    hist_plot(tttt, stats=False, label='Mock Analysis', include_errors=True)
    if tau_category >= 0:
        hist_plot(tm_tttt, stats=False, label='Truth-Matched Taus', include_errors=True)
    plt.ylim((0, None))
    set_xticks()

    plt.sca(ax_ttw)
    plt.title('TTW')
    hist_plot(ttw, stats=False, label='Mock Analysis', include_errors=True)
    if tau_category >= 0:
        hist_plot(tm_ttw, stats=False, label='Truth-Matched Taus', include_errors=True)
    plt.ylim((0, None))
    set_xticks()
    plt.legend()

    plt.sca(ax_ttz)
    plt.title('TTZ')
    hist_plot(ttz, stats=False, label='Mock Analysis', include_errors=True)
    if tau_category >= 0:
        hist_plot(tm_ttz, stats=False, label='Truth-Matched Taus', include_errors=True)
    plt.ylim((0, None))
    set_xticks()
    plt.xlabel('Signal Region')

    plt.sca(ax_tth)
    plt.title('TTH')
    hist_plot(tth, stats=False, label='Mock Analysis', include_errors=True)
    if tau_category >= 0:
        hist_plot(tm_tth, stats=False, label='Truth-Matched Taus', include_errors=True)
    plt.ylim((0, None))
    set_xticks()
    plt.xlabel('Signal Region')

    def to_table(hists):
        return hists_to_table(hists, row_labels=['TTTT', 'TTW', 'TTZ', 'TTH'],
                              column_labels=[f'SR{n}' for n in range(1, len(tttt[0])+1)])

    tables = '<h2>Mock</h2>'
    tables += to_table([tttt, ttw, ttz, tth])
    if tau_category >= 0:
        tables += '<h2>TM Taus</h2>'
        tables += to_table([tm_tttt, tm_ttw, tm_ttz, tm_tth])
    return tables


@decl_plot
def plot_yield_v_gen(tau_category=-1):
    r"""## Event Yield Vs. # of Generated Taus

    """
    def get_sr(rs):
        h = None
        if tau_category == 0:
            h = rs.SRs_0tmtau_diff_nGenTau
        if tau_category == 1:
            h = rs.SRs_1tmtau_diff_nGenTau
        elif tau_category == 2:
            h = rs.SRs_2tmtau_diff_nGenTau
        return hist2d_norm(hist2d(h), norm=1, axis=0)

    _, ((ax_tttt, ax_ttw), (ax_ttz, ax_tth)) = plt.subplots(2, 2)
    tttt, ttw, ttz, tth = [get_sr(rs) for rs in rss]

    def do_plot(h, title):
        hist2d_plot(h, title=title, txt_format='{:.2f}')

    plt.sca(ax_tttt)
    do_plot(tttt, 'TTTT')
    plt.ylabel("\# Gen Taus")

    plt.sca(ax_ttw)
    do_plot(ttw, 'TTW')
    plt.legend()

    plt.sca(ax_ttz)
    do_plot(ttz, 'TTZ')
    plt.xlabel('Signal Region')
    plt.ylabel("\# Gen Taus")

    plt.sca(ax_tth)
    do_plot(tth, 'TTH')
    plt.xlabel('Signal Region')


@decl_plot
def plot_nGen_v_nSel():
    _, ((ax_tttt, ax_ttw), (ax_ttz, ax_tth)) = plt.subplots(2, 2)
    tttt, ttw, ttz, tth = [hist2d(rs.nGen_v_RecoTaus_in_SR) for rs in rss]

    def do_plot(h, title):
        hist2d_plot(h, title=title, txt_format='{:.2f}')

    plt.sca(ax_tttt)
    do_plot(tttt, 'TTTT')
    plt.ylabel("\# Selected Taus")

    plt.sca(ax_ttw)
    do_plot(ttw, 'TTW')
    plt.legend()

    plt.sca(ax_ttz)
    do_plot(ttz, 'TTZ')
    plt.xlabel('\# Gen Taus')
    plt.ylabel("\# Selected Taus")

    plt.sca(ax_tth)
    do_plot(tth, 'TTH')
    plt.xlabel('\# Gen Taus')


@decl_plot
def plot_yield_stack():
    r"""## Event Yield - Stacked

    The event yield for the eight signal regions defined in AN-17-115. Data is normalized
    to the Moriond 2018 integrated luminosity ($35.9\textrm{fb}^{-1}$). Code for the histogram generation is
    here: <https://github.com/cfangmeier/FTAnalysis/blob/master/studies/tau/Yield.C>
    """
    tttt, ttw, ttz, tth = map(lambda rs: hist(rs.SRs), rss)
    hist_plot_stack([ttw, ttz, tth], labels=['TTW', 'TTZ', 'TTH'])
    tttt = tttt[0]*10, tttt[1], tttt[2]
    hist_plot(tttt, label='TTTT (x10)', stats=False, color='k')

    plt.ylim((0, 60))
    plt.xlabel('Signal Region')
    plt.legend()


@decl_plot
def plot_lep_multi(dataset):
    _, (ax_els, ax_mus, ax_taus) = plt.subplots(3, 1)
    els = list(map(lambda rs: hist_norm(hist(rs.nEls)), rss))
    mus = list(map(lambda rs: hist_norm(hist(rs.nMus)), rss))
    taus = list(map(lambda rs: hist_norm(hist(rs.nTaus)), rss))

    def _plot(ax, procs):
        plt.sca(ax)
        tttt, ttw, ttz, tth = procs
        h = {'TTTT': tttt,
             'TTW': ttw,
             'TTZ': ttz,
             'TTH': tth}[dataset]
        hist_plot(h, stats=False, label=dataset)

    _plot(ax_els, els)
    plt.xlabel('\\# Good Electrons')
    plt.legend()
    _plot(ax_mus, mus)
    plt.xlabel('\\# Good Muons')
    _plot(ax_taus, taus)
    plt.xlabel('\\# Good Taus')


@decl_plot
def plot_sig_strength(tau_category, tm):
    r""" The signal strength of the TTTT signal defined as

    $\frac{S}{\sqrt{S+B+\sigma_B^2}}$

    """
    tttt, ttw, ttz, tth = map(lambda rs: get_sr(rs, tau_category, tm), rss)
    bg = hist_add(ttw, ttz, tth)
    strength = tttt[0] / np.sqrt(tttt[0] + bg[0] + bg[1]**2)
    hist_plot((strength, tttt[1], tttt[2]), stats=False)


@decl_plot
def plot_event_obs(dataset, in_sr=True):
    r"""The distribution of $N_{jet}$, $N_{Bjet}$, MET, and $H_T$ in either all events
    or only signal region (igoring taus) events.

    """
    _, ((ax_njet, ax_nbjet), (ax_ht, ax_met)) = plt.subplots(2, 2)
    # tttt, ttw, ttz, tth = map(lambda rs: hist(rs.SRs), rss)
    tttt, ttw, ttz, tth = rss
    rs = {'TTTT': tttt,
          'TTW': ttw,
          'TTZ': ttz,
          'TTH': tth}[dataset]

    def _plot(ax, obs):
        plt.sca(ax)
        if in_sr:
            h = {'MET': rs.met_in_SR,
                 'HT': rs.ht_in_SR,
                 'NJET': rs.njet_in_SR,
                 'NBJET': rs.nbjet_in_SR}[obs]
        else:
            h = {'MET': rs.met,
                 'HT': rs.ht,
                 'NJET': rs.njet,
                 'NBJET': rs.nbjet}[obs]
        hist_plot(hist(h), stats=False, label=dataset)
        plt.xlabel(obs)

    _plot(ax_njet, 'NJET')
    _plot(ax_nbjet, 'NBJET')
    _plot(ax_ht, 'HT')
    _plot(ax_met, 'MET')


@decl_plot
def plot_event_obs_stack(in_signal_region=True):
    r"""

    """
    _, ((ax_njet, ax_nbjet), (ax_ht, ax_met)) = plt.subplots(2, 2)

    def _plot(ax, obs):
        plt.sca(ax)
        if in_signal_region:
            attr = {'MET': 'met_in_SR',
                    'HT': 'ht_in_SR',
                    'NJET': 'njet_in_SR',
                    'NBJET': 'nbjet_in_SR'}[obs]
        else:
            attr = {'MET': 'met',
                    'HT': 'ht',
                    'NJET': 'njet',
                    'NBJET': 'nbjet'}[obs]
        tttt, ttw, ttz, tth = map(lambda rs: hist(getattr(rs, attr)), rss)
        hist_plot_stack([ttw, ttz, tth], labels=["TTW", "TTZ", "TTH"])
        hist_plot(hist_scale(tttt, 5), label="TTTT (x5)", color='k')
        plt.xlabel(obs)

    _plot(ax_njet, 'NJET')
    _plot(ax_nbjet, 'NBJET')
    plt.legend()
    _plot(ax_ht, 'HT')
    _plot(ax_met, 'MET')


@decl_plot
def plot_s_over_b(tau_category):

    tttt, ttw, ttz, tth = [get_sr(rs, tau_category) for rs in rss]
    sig = get_sr(rss[0], tau_category, tm=True)
    fake_tttt = hist_sub(tttt, sig)

    total_bg = hist_add(fake_tttt, ttw, ttz, tth)

    plt.subplot(3, 1, 1)
    hist_plot(total_bg, include_errors=True, label="TTV+TTH+Fake TTTT")
    hist_plot(sig, include_errors=True, label="Truth Matched TTTT")
    plt.ylabel('\\# Events')
    plt.ylim((0, 5.5))
    set_xticks()
    plt.legend()

    plt.subplot(3, 1, 2)
    hist_plot(hist_div(sig, total_bg), label="ratio", color='k')
    plt.ylabel('Ratio')
    plt.ylim((0, 2.7))
    set_xticks()

    plt.subplot(3, 1, 3)
    bg_vals, bg_errs, _ = total_bg
    sig_vals, _, _ = sig
    den = np.sqrt(sig_vals + bg_vals + bg_errs**2)
    sig = sig_vals / den, *sig[1:]

    hist_plot(sig, color='k')
    plt.ylabel(r'$S/\sqrt{S+B+\sigma_B^2}$')
    plt.ylim((0, 0.5))
    set_xticks()

    # tables = '<h2>Mock</h2>'
    # tables += to_table([total_bg, ttw, ttz, tth])
    # if tau_category >= 0:
    #     tables += '<h2>TM Taus</h2>'
    #     tables += to_table([tm_tttt, tm_ttw, tm_ttz, tm_tth])
    # return tables


@decl_plot
def plot_tau_efficiency():
    _, ((ax_tttt, ax_ttw), (ax_ttz, ax_tth)) = plt.subplots(2, 2)
    tttt, ttw, ttz, tth = list(map(lambda rs: hist(rs.tau_efficiency_v_pt), rss))

    def _plot(ax, dataset):
        plt.sca(ax)
        h = {'TTTT': tttt,
             'TTW': ttw,
             'TTZ': ttz,
             'TTH': tth}[dataset]
        hist_plot(h, stats=False, label=dataset, include_errors=True)
        plt.text(200, 0.05, dataset)
        plt.xlabel(r"$P_T$(GeV)")
        plt.ylim((0, 1.1))

    _plot(ax_tttt, 'TTTT')
    _plot(ax_ttw, 'TTW')
    _plot(ax_ttz, 'TTZ')
    _plot(ax_tth, 'TTH')


@decl_plot
def plot_tau_purity():
    _, ((ax_tttt, ax_ttw), (ax_ttz, ax_tth)) = plt.subplots(2, 2)
    tttt, ttw, ttz, tth = list(map(lambda rs: hist(rs.tau_purity_v_pt), rss))

    def _plot(ax, dataset):
        plt.sca(ax)
        h = {'TTTT': tttt,
             'TTW': ttw,
             'TTZ': ttz,
             'TTH': tth}[dataset]
        hist_plot(h, stats=False, label=dataset, include_errors=True)
        plt.text(200, 0.05, dataset)
        plt.xlabel(r"$P_T$(GeV)")
        plt.ylim((0, 1.1))

    _plot(ax_tttt, 'TTTT')
    _plot(ax_ttw, 'TTW')
    _plot(ax_ttz, 'TTZ')
    _plot(ax_tth, 'TTH')


if __name__ == '__main__':
    ext = ""
    # ext = "_dr03"
    data_path = f'data/TauStudies/output{ext}/'
    save_plots = False

    rss = (ResultSet("tttt", data_path+'yield_tttt.root'),
           ResultSet("ttw", data_path+'yield_ttw.root'),
           ResultSet("ttz", data_path+'yield_ttz.root'),
           ResultSet("tth", data_path+'yield_tth.root'))

    tttt_event_obs_in_sr = plot_event_obs, ('TTTT',), {'in_sr': True}
    ttw_event_obs_in_sr = plot_event_obs, ('TTW',), {'in_sr': True}
    ttz_event_obs_in_sr = plot_event_obs, ('TTZ',), {'in_sr': True}
    tth_event_obs_in_sr = plot_event_obs, ('TTH',), {'in_sr': True}

    tttt_event_obs = plot_event_obs, ('TTTT',), {'in_sr': False}
    ttw_event_obs = plot_event_obs, ('TTW',), {'in_sr': False}
    ttz_event_obs = plot_event_obs, ('TTZ',), {'in_sr': False}
    tth_event_obs = plot_event_obs, ('TTH',), {'in_sr': False}

    tttt_lep_multi = plot_lep_multi, ('TTTT',)
    ttw_lep_multi = plot_lep_multi, ('TTW',)
    ttz_lep_multi = plot_lep_multi, ('TTZ',)
    tth_lep_multi = plot_lep_multi, ('TTH',)

    yield_v_gen_0tm = plot_yield_v_gen, (0,)
    yield_v_gen_1tm = plot_yield_v_gen, (1,)
    yield_v_gen_2tm = plot_yield_v_gen, (2,)
    nGen_v_nSel = plot_nGen_v_nSel

    # Now assemble the plots into figures.
    plots = [

        Plot((plot_yield_grid, (-1,)),
             'Yield Ignoring Taus'),
        Plot((plot_yield_grid, (0,)), 'Yield For events with 0 Tau'),
        Plot((plot_s_over_b, (0,)), 'Real TTTT and Total BG - 0 Tau'),
        Plot((plot_yield_grid, (1,)), 'Yield For events with 1 Tau'),
        Plot((plot_s_over_b, (1,)), 'Real TTTT and Total BG - 1 Tau'),
        Plot((plot_yield_grid, (2,)), 'Yield For events with 2 or more Tau'),
        Plot((plot_s_over_b, (2,)), 'Real TTTT and Total BG - 2 or more Tau'),
        Plot(nGen_v_nSel,
             r'#Generated tau vs. #Selected tau'),
        Plot(tttt_lep_multi,
             'Lepton Multiplicity - TTTT'),
        Plot(ttw_lep_multi,
             'Lepton Multiplicity - TTW'),
        Plot(ttz_lep_multi,
             'Lepton Multiplicity - TTZ'),
        Plot(tth_lep_multi,
             'Lepton Multiplicity - TTH'),
        Plot(tttt_event_obs_in_sr,
             'TTTT - Event Observables (In SR)'),
        Plot(ttw_event_obs_in_sr,
             'TTW - Event Observables (In SR)'),
        Plot(ttz_event_obs_in_sr,
             'TTZ - Event Observables (In SR)'),
        Plot(tth_event_obs_in_sr,
             'TTH - Event Observables (In SR)'),

        Plot(plot_tau_efficiency,
             'Tau Efficiency'),
        Plot(plot_tau_purity,
             'Tau Purity'),
    ]

    # Finally, render and save the plots and generate the html+bootstrap
    # dashboard to view them
    render_plots(plots, to_disk=save_plots)
    if not save_plots:
        with open(data_path+'config.yaml') as f:
            config_txt = f.read()
        generate_dashboard(plots, 'Tau Studies',
                           output=f'tau_studies{ext}.html',
                           source=__file__,
                           config=config_txt
                           )