How to define histogram with variable bin width in python

[saswatinandan]

Hi,

I am trying to define histogram with variable bin width in python in following way:

import numpy
x=numpy.array([2,4,8,9])
import ROOT as r
h=r.TH1F(‘’,‘’,3,x)
h
<ROOT.TH1F object at 0x4f3bf30>

but the width of each bin is zero

h.GetXaxis().GetBinWidth(1)
0.0
h.GetXaxis().GetBinWidth(2)
0.0
h.GetNbinsX()
3

Could anyone please tell me the correct way to define the histogram with variable bin width in python?

Thanks in advance -:))

Saswati

[bellenot]

See for example:

Python macro and histogram with variable bin width

Angi, the TH1::FillN() isn’t going to work with just two arguments, but to show how to pass a list through a double* pointer:[code]from ROOT import * from array import array binLowE = [2,4,9,15,20,22] uniformTau = TH1D(“uniformTau”,"",5,array(‘d’,binLowE)) #uniformTau.FillN(5,10) uniformTau.Draw()[/code] I.e. just use an array from module array. The buffer interface does the rest. HTH, Wim

(searching on the forum is sometimes useful…)

[wlav]

In defense of TS, the code won’t work with a list per the other topic but should do with a numpy array.

There was a recent thread and bug report about incorrect array converters being selected: the selection is on element size, not element type as numpy arrays don’t carry a typecode as an attribute (but do as part of the buffer info). See:

https://sft.its.cern.ch/jira/browse/ROOT-9040
pyROOT TMultiGraph Segmentation Fault

[ksmith]

As @wlav has pointed out this is a bug about types. ROOT interprets your array as one with floats, even though it contains integers and you should have received an error. Specifying dtype when defining your array will resolve the issue until the bug is resolved.

import numpy as np
import ROOT

#Must explicit declare variable type
bin_edges = np.array([2, 4, 8, 9], dtype='float64')

h = ROOT.TH1F("","",len(bin_edges)-1, bin_edges)

for bin_num in range(len(bin_edges)-1):
    print h.GetBinWidth(bin_num)