def getQuantiles(limits, median, onesigma, twosigma):
nit = len(limits)
if nit==0:
return
# sort the list with limits
limits = sorted(limits)
limits_array = array('d',limits)
# median for the expected limit
median[0] = TMath.Median(nit, limits_array)
# quantiles for the expected limit bands
prob = array('d',[]) # array with quantile boundaries
prob.append(0.021)
prob.append(0.159)
prob.append(0.841)
prob.append(0.979)
# array for the results
quantiles = array('d',[0., 0., 0., 0.])
TMath.Quantiles(nit, 4, limits_array, quantiles, prob) # evaluate quantiles
onesigma[0] = quantiles[1]
onesigma[1] = quantiles[2]
twosigma[0] = quantiles[0]
twosigma[1] = quantiles[3]
return
def getMedian(self, h):
#compute the median for 1-d histogram h1
nbins = h.GetXaxis().GetNbins()
xList = []
yList = []
for i in range(0, nbins):
xList.append(h.GetBinCenter(i + 1))
yList.append(h.GetBinContent(i + 1))
return TMath.Median(len(xList), np.array(xList, 'd'),
np.array(yList, 'd'))
def getMedian(h):
#compute the median for 1-d histogram h1
nbins = h.GetXaxis().GetNbins()
xList = []
yList = []
for i in range(0, nbins):
xList.append(h.GetBinCenter(i + 1))
yList.append(h.GetBinContent(i + 1))
# For update to ROOT 6, explicit type is needed. Seems to be in connection with
# only templated method available in ROOT 6. Write:
#
# return TMath.Median('double')(len(xList),np.array(xList,'d'),np.array(yList,'d'))
#
return TMath.Median(len(xList), np.array(xList, 'd'), np.array(yList, 'd'))
def add_expected(self, cat, filename, legend, scale = 1.0,
line_color = 1, line_style = 1, line_width = 1,
marker_color = 1, marker_size = 1.2, marker_style=8,
value_type = 'median',
error_type = 'rms',
fill_style = 1002,
fill_color = ROOT.kYellow,
fill2_style = 1002,
fill2_color = ROOT.kRed,
extra_scale_name = 'unit',
smooth = None # 'single' or 'comb' or None: special tricks for smoothing expected bands
):
self.type[cat] = 'expected'
self.cat.append(cat)
self.file = filename
self.tlegend[cat] = legend
self.line_color[cat] = line_color
self.line_style[cat] = line_style
self.line_width[cat] = line_width
self.marker_color[cat] = marker_color
self.marker_size[cat] = marker_size
self.marker_style[cat] = marker_style
self.fill_style[cat] = fill_style
self.fill2_style[cat] = fill2_style
self.fill_color[cat] = fill_color
self.fill2_color[cat] = fill2_color
_x = array('d')
_y = array('d')
_exh = array('d')
_exl = array('d')
_eyh = array('d')
_eyl = array('d')
_exh2 = array('d')
_exl2 = array('d')
_eyh2 = array('d')
_eyl2 = array('d')
_d = [] # list of tuples, one tuple per mass point
with open(filename, 'r') as _file:
_mass = None
_arr = array('d')
_prev_limit = 0.0
for line in _file:
#print self.legend, line
_buf = line.split()
aMass = float(_buf[0])
aLim = float(_buf[1])*scale*get_scale(extra_scale_name, aMass)
if _mass == None:
_mass = aMass
if abs(_mass-aMass) > 0.1:
_d.append((_mass, _arr))
#print self.legend, cat, 'mass =', _mass, 'PEs per point:', len(_arr)
_mass = aMass
_arr=array('d')
#if abs(aLim - _prev_limit) < 0.000001:
# #print self.legend, 'WARNING: duplicate expected limit, ignoring...'
# continue
#_arr.append(float(_buf[1])*scale)
_arr.append(aLim)
_prev_limit = aLim
_d.append((_mass, _arr))
#print self.legend, _d
for t in _d:
_x.append(t[0])
_mean = TMath.Mean(len(t[1]), t[1])
_median = TMath.Median(len(t[1]), t[1])
_rms = TMath.RMS(len(t[1]), t[1])
_nrms = _rms/len(t[1])
#print self.legend, t[0], _mean, _rms, _nrms
if value_type == 'mean':
_value = _mean
_y.append(_mean)
elif value_type == 'median':
_value = _median
_y.append(_median)
_exh.append(0)
_exl.append(0)
_exh2.append(0)
_exl2.append(0)
if error_type == 'rms':
_eyh.append(_rms/2)
_eyl.append(_rms/2)
_eyh2.append(_rms)
_eyl2.append(_rms)
elif error_type == 'quantile':
_prob = array('d')
_prob.append(0.021)
_prob.append(0.159)
_prob.append(0.841)
_prob.append(0.979)
_nprob = 4
_quantiles = array('d')
_quantiles.append(0)
_quantiles.append(0)
_quantiles.append(0)
_quantiles.append(0)
TMath.Quantiles(len(t[1]), _nprob, t[1], _quantiles, _prob)
#print self.legend, 'Quantiles:', _quantiles
_eyh.append(abs(_value-_quantiles[2]))
_eyl.append(abs(_value-_quantiles[1]))
_eyh2.append(abs(_value-_quantiles[3]))
_eyl2.append(abs(_value-_quantiles[0]))
self.x[cat] = _x
self.y[cat] = _y
self.exh[cat] = _exh
self.exl[cat] = _exl
self.exh2[cat] = _exh2
self.exl2[cat] = _exl2
# non-smoothed bands
#self.eyh[cat] = _eyh
#self.eyl[cat] = _eyl
#self.eyh2[cat] = _eyh2
#self.eyl2[cat] = _eyl2
# smoothed bands
# ee, mumu
#self.eyh[cat] = self.smooth_band(_x,_eyh,_y,[1],300,600,'high')
#self.eyh2[cat] = self.smooth_band(_x,_eyh2,_y,[1],300,600,'high')
#self.eyl[cat] = self.smooth_band(_x,_eyl,_y,[],300,600)
#self.eyl2[cat] = self.smooth_band(_x,_eyl2,_y,[1],300,600)
# comb
self.eyh[cat] = self.smooth_band(_x,_eyh,_y,[],300,600,'high')
self.eyh2[cat] = self.smooth_band(_x,_eyh2,_y,[1],300,600,'high')
self.eyl[cat] = self.smooth_band(_x,_eyl,_y,[1],300,600)
self.eyl2[cat] = self.smooth_band(_x,_eyl2,_y,[1],300,600)
def add_median(self, cat, filename, legend, scale = 1.0,
line_color = 1, line_style = 1, line_width = 1,
marker_color = 1, marker_size = 1.2, marker_style=8,
dofit=False, fit_min = 100, fit_max = 3000,
var_scale = None,
fill_style = 1002,
fill_color = None,
fill2_style = 1002,
fill2_color = ROOT.kRed,
extra_scale_name = 'unit',
value_type = 'median',
error_type = 'quantile'):
#
# same as add() but for multiple values
#
if fill_color == None:
fill_color = line_color
self.type[cat] = 'observed'
self.dofit[cat] = dofit
self.fit_min[cat] = fit_min
self.fit_max[cat] = fit_max
self.cat.append(cat)
self.file = filename
self.tlegend[cat] = legend
self.line_color[cat] = line_color
self.line_style[cat] = line_style
self.line_width[cat] = line_width
self.marker_color[cat] = marker_color
self.marker_size[cat] = marker_size
self.marker_style[cat] = marker_style
self.fill_style[cat] = fill_style
self.fill2_style[cat] = fill2_style
self.fill_color[cat] = fill_color
self.fill2_color[cat] = fill2_color
_x = array('d')
_y = array('d')
_exh = array('d')
_exl = array('d')
_eyh = array('d')
_eyl = array('d')
_d = [] # list of tuples, one tuple per mass point
with open(filename, 'r') as _file:
_mass = None
_arr = array('d')
_prev_limit = 0.0
for line in _file:
#print self.legend, line
_buf = line.split()
if _buf[0][0] == '#':
continue
#_mass = float(_buf[0])
aMass = float(_buf[0])
_scale = get_scale(var_scale, aMass)
_scale_err = get_scale_err(var_scale, aMass)
if _scale != None:
_scale = _scale*scale
else:
_scale = scale
#_xval = float(_buf[0])
#_x.append(_xval)
#_yval = float(_buf[1])*_scale*get_scale(extra_scale_name, _xval)
aLim = float(_buf[1])*_scale*get_scale(extra_scale_name, aMass)
#_y.append(_yval)
if _mass == None:
_mass = aMass
if abs(_mass-aMass) > 0.1:
_d.append((_mass, _arr))
#print self.legend, cat, 'mass =', _mass, 'PEs per point:', len(_arr)
_mass = aMass
_arr=array('d')
#if abs(aLim - _prev_limit) < 0.000001:
# #print self.legend, 'WARNING: duplicate expected limit, ignoring...'
# continue
#_arr.append(float(_buf[1])*scale)
_arr.append(aLim)
_prev_limit = aLim
_d.append((_mass, _arr))
for t in _d:
_x.append(t[0])
_mean = TMath.Mean(len(t[1]), t[1])
_median = TMath.Median(len(t[1]), t[1])
_rms = TMath.RMS(len(t[1]), t[1])
_nrms = _rms/len(t[1])
#print self.legend, t[0], _mean, _rms, _nrms
if value_type == 'mean':
_value = _mean
_y.append(_mean)
elif value_type == 'median':
_value = _median
_y.append(_median)
_exh.append(0)
_exl.append(0)
if error_type == 'rms':
_eyh.append(_rms/2)
_eyl.append(_rms/2)
elif error_type == 'quantile':
_prob = array('d')
_prob.append(0.021)
_prob.append(0.159)
_prob.append(0.841)
_prob.append(0.979)
_nprob = 4
_quantiles = array('d')
_quantiles.append(0)
_quantiles.append(0)
_quantiles.append(0)
_quantiles.append(0)
TMath.Quantiles(len(t[1]), _nprob, t[1], _quantiles, _prob)
#print self.legend, 'Quantiles:', _quantiles
#_eyh.append(abs(_value-_quantiles[2]))
#_eyl.append(abs(_value-_quantiles[1]))
_eyh.append(0)
_eyl.append(0)
self.x[cat] = _x
self.y[cat] = _y
self.exh[cat] = _exh
self.exl[cat] = _exl
self.eyh[cat] = _eyh
self.eyl[cat] = _eyl
