Skip to content

Latest commit

 

History

History
111 lines (85 loc) · 4.11 KB

README.rst

File metadata and controls

111 lines (85 loc) · 4.11 KB
Raw

Numpy Structures

Documentation Status

Simple data structures that augments the numpy library

Features

The main feature is the RaggedArray class which enables numpy-like behaviour and performance for arrays where the length of the rows differ.

RaggedArray is meant as a drop-in replacement for numpy when you have arrays with differing row lengths. As such, familiarity with numpy is assumed. The simplest way to construct a RaggedArray is from a list of lists:

>>> from npstructures import RaggedArray
>>> ra = RaggedArray([[1, 2], [4, 1, 3, 7], [9], [8, 7, 3, 4]])

A RaggedArray can be indexed much like a numpy array:

>>> ra[1]
array([4, 1, 3, 7])
>>> ra[1, 3]
7
>>> ra[1:3]
RaggedArray([[4, 1, 3, 7], [9]])
>>> ra[[0, 3]]
RaggedArray([[1, 2], [8, 7, 3, 4]])
>>> ra[0] = [0, 0]
>>> ra
RaggedArray([[0, 0], [4, 1, 3, 7], [9], [8, 7, 3, 4]])
>>> ra[1:3] = [[10], [20]]
>>> ra
RaggedArray([[0, 0], [10, 10, 10, 10], [20], [8, 7, 3, 4]])
>>> ra[[0, 2, 3]] = RaggedArray([[2, 2], [3], [5, 5, 5, 5]])
>>> ra
RaggedArray([[2, 2], [10, 10, 10, 10], [3], [5, 5, 5, 5]])

numpy ufuncs can be applied to RaggedArray objects:

>>> ra + 1
RaggedArray([[2, 3], [5, 2, 4, 8], [10], [9, 8, 4, 5]])
>>> ra*2
RaggedArray([[2, 4], [8, 2, 6, 14], [18], [16, 14, 6, 8]])
>>> ra + [[1], [10], [100], [1000]]
RaggedArray([[2, 3], [14, 11, 13, 17], [109], [1008, 1007, 1003, 1004]])
>>> ra - (ra*2)
RaggedArray([[-1, -2], [-4, -1, -3, -7], [-9], [-8, -7, -3, -4]])

Some numpy functions can be applied to RaggedArray objects:

>>> import numpy as np
>>> ra = RaggedArray([[1, 2], [4, 1, 3, 7], [9], [8, 7, 3, 4]])
>>> np.concatenate((ra, ra*10))
RaggedArray([[1, 2], [4, 1, 3, 7], [9], [8, 7, 3, 4], [10, 20], [40, 10, 30, 70], [90], [80, 70, 30, 40]])
>>> np.nonzero(ra>3)
(array([1, 1, 2, 3, 3, 3]), array([0, 3, 0, 0, 1, 3]))
>>> np.ones_like(ra)
RaggedArray([[1, 1], [1, 1, 1, 1], [1], [1, 1, 1, 1]])

In addition to this. HashTable and Counter provides simple dict-like behaviour for numpy arrays:

HashTable can be used for dict-like functionality of numpy arrays. The simplest way to construct a HashTable is from an array of keys and an array of values (note that the set of keys cannot be modified after the initialization of the object):

>>> table = HashTable([11, 113, 1191, 11199], [2, 3, 5, 7])
>>> table[11]
array([2])
>>> table[[113, 11199]]
array([3, 7])
>>> table[11]=1000
>>> table
HashTable([  113  1191    11 11199], [   3    5 1000    7])
>>> table[[113, 1191]]=2000
>>> table
HashTable([  113  1191    11 11199], [2000 2000 1000    7])
>>> table[[113, 1191, 11, 11191]] = [1, 2, 3, 4]
>>> table[[113, 1191, 11, 11199]] = [1, 2, 3, 4]
>>> table
HashTable([  113  1191    11 11199], [1 2 3 4])

Counter objects supports counting the occurances of a predefined set of keys in a set of samples. For instance, to count the occurances of 3 and 1 in the list [3, 2, 1, 3, 4, 1, 1]:

>>> from npstructures import Counter
>>> counter = Counter([3, 1])
>>> counter.count([3, 2, 1, 3, 4, 1, 1])
>>> counter
Counter([3 1], [2 3])

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.