vo008_numpyInterface.py File Reference

This tutorial illustrates the conversion of RVec to numpy arrays without copying the data.

The memory-adoption is achieved by the dictionary array_interface, which is added dynamically to the Python objects by PyROOT.

 
import ROOT
from sys import exit
 
try:
    import numpy as np
except:
    exit()
 
# Create a vector ROOT object and assign values
# Note that this works as well with a ROOT.std.vector
vec = ROOT.RVecF(2) # or ROOT.std.vector("float")(2)
vec[0] = 1
vec[1] = 2
print("Content of the ROOT vector object: {}".format([x for x in vec]))
 
# Interface ROOT vector with a numpy array
array = np.asarray(vec)
print("Content of the associated numpy array: {}".format([x for x in array]))
 
# The numpy array adopts the memory of the vector without copying the content.
# Note that the first entry of the numpy array changes when assigning a new
# value to the first entry of the ROOT vector.
vec[0] = 42
print(
    "Content of the numpy array after changing the first entry of the ROOT vector: {}".
    format([x for x in array]))
 
# Use numpy features on data of ROOT objects
print("Mean of the numpy array entries: {}".format(np.mean(array)))

Content of the ROOT vector object: [1.0, 2.0]
Content of the associated numpy array: [np.float32(1.0), np.float32(2.0)]
Content of the numpy array after changing the first entry of the ROOT vector: [np.float32(42.0), np.float32(2.0)]
Mean of the numpy array entries: 22.0

Date

April 2018

Author

Stefan Wunsch

Definition in file vo008_numpyInterface.py.