rootpy.plotting.views: Directory “Views”

Folder “View” Classes

These classes wrap Directories and perform automatic actions to Histograms retrieved from them. The different views can be composited and layered.

Summary of views:

  • ScaleView: scale histogram normalization
  • NormalizeView: normalize histograms
  • SumView: sum histograms from different folders together
  • StyleView: apply a style to histograms
  • StackView: build THStacks using histograms from different folders
  • TitleView: change the title of histograms
  • FunctorView: apply a arbitrary transformation function to the histograms
  • MultiFunctorView: apply a arbitrary transformation function to a collection of histograms
  • SubdirectoryView: A view of a subdirectory, which maintains the same view as the base.

Example use case

One has a ROOT file with the following content:

zjets/mutau_mass
zz/mutau_mass
wz/mutau_mass
data_2010/mutau_mass
data_2011/mutau_mass

and wants to do the following:

  1. Merge the two data taking periods together
  2. Scale the Z, WZ, and ZZ simulated results to the appropriate int. lumi.
  3. Combine WZ and ZZ into a single diboson sample
  4. Apply different colors to the MC samples
  5. Make a Stack of the expected yields from different simulated processes

This example can be tested by running:

python -m rootpy.plotting.views
>>> # Mock up the example test case
>>> import rootpy.io as io
>>> # We have to keep these, to make sure PyROOT doesn't garbage collect them
>>> keep = []
>>> zjets_dir = io.Directory('zjets', 'Zjets directory')
>>> zz_dir = io.Directory('zz', 'ZZ directory')
>>> wz_dir = io.Directory('wz', 'WZ directory')
>>> data2010_dir = io.Directory('data2010', 'data2010 directory')
>>> data2011_dir = io.Directory('data2011', 'data2011 directory')
>>> # Make the Zjets case
>>> _ = zjets_dir.cd()
>>> zjets_hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> zjets_hist.FillRandom('gaus', 5000)
>>> keep.append(zjets_hist)
>>> # Make the ZZ case
>>> _ = zz_dir.cd()
>>> zz_hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> zz_hist.FillRandom('gaus', 5000)
>>> keep.append(zz_hist)
>>> # Make the WZ case
>>> _ = wz_dir.cd()
>>> wz_hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> wz_hist.FillRandom('gaus', 5000)
>>> keep.append(wz_hist)
>>> # Make the 2010 data case
>>> _ = data2010_dir.cd()
>>> data2010_hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> data2010_hist.FillRandom('gaus', 30)
>>> keep.append(data2010_hist)
>>> # Make the 2011 data case
>>> _ = data2011_dir.cd()
>>> data2011_hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> data2011_hist.FillRandom('gaus', 51)
>>> keep.append(data2011_hist)

SumView

We can merge the two data periods into a single case using a SumView.

>>> data = SumView(data2010_dir, data2011_dir)
>>> data_hist = data.Get("mutau_mass")
>>> data_hist.Integral()
81.0
>>> data_hist.Integral() == data2010_hist.Integral() + data2011_hist.Integral()
True

ScaleView

The simulated results (Z & diboson) can be scaled to the expected integrated luminosity using ScaleViews.

>>> zjets = ScaleView(zjets_dir, 0.01)
>>> zjets_hist = zjets.Get("mutau_mass")
>>> abs(zjets_hist.Integral() - 50.0) < 1e-5
True
>>> # Scale the diboson contribution
>>> zz = ScaleView(zz_dir, 0.001)
>>> wz = ScaleView(wz_dir, 0.003)

Combining views

The dibosons individually are tiny, let’s put them together using a SumView. Note that this operation nests two ScaleViews into a SumView.

>>> dibosons = SumView(zz, wz)
>>> # We expect 5000*0.001 + 5000*0.003 = 20 events
>>> dibosons_hist = dibosons.Get("mutau_mass")
>>> abs(dibosons_hist.Integral() - 20) < 1e-4
True

StyleView

A style view automatically applies a style to retrieved Plottable objects. The style is specified using the same arguments as the Plottable.decorate. Let’s make the Z background red and the diboson background blue.

>>> zjets = StyleView(zjets, fillcolor=ROOT.EColor.kRed)
>>> dibosons = StyleView(dibosons, fillcolor=ROOT.EColor.kBlue)
>>> zjets_hist = zjets.Get("mutau_mass")
>>> zjets_hist.GetFillColor() == ROOT.EColor.kRed
True
>>> dibosons_hist = dibosons.Get("mutau_mass")
>>> dibosons_hist.GetFillColor() == ROOT.EColor.kBlue
True

StackView

The StackView combines multiple items into a HistStack. In our example we stack the SM backgrounds to compare to the data.

>>> sm_bkg = StackView(zjets, dibosons)
>>> sm_bkg_stack = sm_bkg.Get("mutau_mass")
>>> '%0.0f' % sm_bkg_stack.Integral()
'70'

Looks like we have an excess of 11 events - must be the Higgs.

Other Examples

NormalizeView

The normalization view renormalizes histograms to a given value (default 1.0). Here is an example of using the NormalizeView to compare the Z and diboson shapes.

>>> z_shape = NormalizeView(zjets)
>>> z_shape_hist = z_shape.Get("mutau_mass")
>>> abs(1 - z_shape_hist.Integral()) < 1e-5
True
>>> # Let's compare the shapes using a HistStack, using the "nostack" option.
>>> diboson_shape = NormalizeView(dibosons)
>>> shape_comparison = StackView(z_shape, diboson_shape)
>>> # To draw the comparison:
>>> # shape_comparison.Get("mutau_mass").Draw('nostack')

FunctorView

FunctorView allows you to apply an arbitrary transformation to the object. Here we show how you can change the axis range for all histograms in a directory.

>>> rebin = lambda x: x.Rebin(2)
>>> zjets_rebinned = FunctorView(zjets, rebin)
>>> zjets.Get("mutau_mass").GetNbinsX()
100
>>> zjets_rebinned.Get("mutau_mass").GetNbinsX()
50

The functor doesn’t have to return a histogram.

>>> mean_getter = lambda x: x.GetMean()
>>> mean = zjets.Get("mutau_mass").GetMean()
>>> zjets_mean = FunctorView(zjets, mean_getter)
>>> zjets_mean.Get("mutau_mass") == mean
True

MultiFunctorView

MultiFunctorView is similar except that it operates on a group of histograms. The functor should take one argument, a generator of the sub-objects.

Here’s an example to get the integral of the biggest histogram in a set:

>>> biggest_histo = lambda objects: max(y.Integral() for y in objects)
>>> biggest = MultiFunctorView(biggest_histo, zjets, dibosons)
>>> biggest.Get("mutau_mass") == zjets.Get("mutau_mass").Integral()
True

SubdirectoryView

If you’d like to “cd” into a lower subdirectory, while still maintaining the same view, use a SubdirectoryView.

>>> basedir = io.Directory('base', 'base directory')
>>> _ = basedir.cd()
>>> subdir1 = io.Directory('subdir1', 'subdir directory in 1')
>>> _ = subdir1.cd()
>>> hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> hist.FillRandom('gaus', 2000)
>>> keep.append(hist)
>>> _ = basedir.cd()
>>> subdir2 = io.Directory('subdir2', 'subdir directory 2')
>>> _ = subdir2.cd()
>>> hist = ROOT.TH1F("mutau_mass", "Mu-Tau mass", 100, 0, 100)
>>> hist.FillRandom('gaus', 5000)
>>> keep.append(hist)
The directory structure is now::
base/subdir1/hist base/subdir2/hist

Subdirectory views work on top of other views.

>>> baseview = ScaleView(basedir, 0.1)
>>> subdir1view = SubdirectoryView(baseview, 'subdir1')
>>> subdir2view = SubdirectoryView(baseview, 'subdir2')
>>> histo1 = subdir1view.Get('mutau_mass')
>>> histo2 = subdir2view.Get('mutau_mass')
>>> exp_histo1 = baseview.Get("subdir1/mutau_mass")
>>> exp_histo2 = baseview.Get("subdir2/mutau_mass")
>>> def equivalent(h1, h2):
...     return (abs(h1.GetMean() - h2.GetMean()) < 1e-4 and
...             abs(h1.GetRMS() - h2.GetRMS()) < 1e-4 and
...             abs(h1.Integral() - h2.Integral()) < 1e-4)
>>> equivalent(exp_histo1, histo1)
True
>>> equivalent(exp_histo2, histo2)
True
>>> equivalent(histo1, histo2)
False

Classes

plotting.views.ScaleView(directory, scale_factor) View of a folder which applies a scaling factor to histograms.
plotting.views.NormalizeView(directory[, …]) Normalize histograms to a constant value
plotting.views.StyleView(directory, **kwargs) View of a folder which applies a style to Plottable objects.
plotting.views.TitleView(directory, title) Override the title of gotten histograms
plotting.views.SumView(*directories) Add a collection of histograms together
plotting.views.StackView(*directories, **kwargs) Build a HistStack from the input histograms
plotting.views.FunctorView(directory, function) Apply an arbitrary function to the output histogram.
plotting.views.MultiFunctorView(f, *directories) Apply an arbitrary function to the output histograms.
plotting.views.PathModifierView(dir, …) Does some magic to the path
plotting.views.SubdirectoryView(dir, subdirpath) Add some base directories to the path of Get()