flat.SetParameter(0, 1.)
flat.SetParLimits(0, 0.95, 1.05)
sfTruth.Fit(flat)
canvas.addObject(flat)
linear = ROOT.TF1('linear', '[0] + [1] * x', scaleFactor.GetXaxis().GetXmin(), scaleFactor.GetXaxis().GetXmax())
linear.SetParameters(1., 0.01)
linear.SetParLimits(0, 0.95, 1.05)
linear.SetParLimits(1, -0.0001, 0.0001)
sfTruth.Fit(linear)
canvas.addObject(linear)
text = 'flat = %.3f #pm %.3f' % (flat.GetParameter(0), flat.GetParError(0))
canvas.addText(text, 0.3, 0.25, 0.5, 0.2)
text = 'line = %.6f + %.6f * p_{T}' % (linear.GetParameter(0), linear.GetParameter(1))
canvas.addText(text, 0.3, 0.35, 0.5, 0.25)
canvas.xtitle = binningTitle
canvas.printWeb(outputName, 'scaleFactor_' + binningName, logy = False)
print 'Fit Results:'
for iBin, (bin, _) in enumerate(fitBins):
print '%15s [%.3f +- %.3f]' % (bin, scaleFactor.GetBinContent(iBin + 1), scaleFactor.GetBinError(iBin + 1))
print '\nTruth Results:'
result.Fit(power)
canvas.addObject(power)
outputFile.cd()
power.Write(PRODUCT + '_fit')
text = 'f = %.4f + %.3f#times(p_{T} ' % (power.GetParameter(0),
power.GetParameter(1))
if power.GetParameter(2) >= 0.:
text += ' - '
else:
text += ' + '
text += '%.2f)^{%.2f}' % (abs(
power.GetParameter(2)), power.GetParameter(3))
canvas.addText(text, 0.4, 0.15, 0.6, 0.2)
for ibin, (cont, err) in contents.items():
result.SetBinContent(ibin, cont)
result.SetBinError(ibin, err)
# Throw toys and get uncertainties
# draw one normal parameter for total systematic (assume full correlation across bins)
# draw one normal parameter per bin for statistical
# 68% percentile on both sides become the up & down variations
original = tuple(power.GetParameter(i) for i in range(power.GetNpar()))
ntoys = 200
npx = 100
