kuniga.me > Docs > Numpy Cheatsheet

Numpy Cheatsheet

Assuming numpy is imported as:

import numpy as np

Vector (1D)

Un-initialized vector of size N:

np.empty(shape=N)

Similar versions for vectors filled with 1s: ones, ans 0s: zeros. Note that np.empty(shape=N) is different from np.empty(shape=(N, 1)). The former has x[N], while the latter x[N, 1].

To flatten a matrix m(Nx1) to a vector v(N):

v = m.reshape(shape=N)

Matrix (2D)

Initialization

Un-initialized matrix of size N x M:

m = np.empty(shape=(N, M))

Similar versions for vectors filled with 1s: ones, and 0s: zeros.

Number of rows:

len(m)

Number of columns:

len(m[0])

Initialize from nested Python lists:

m = np.array([
    [0.1, 0.3, 0.5],
    [0.9, 0.7, 0.5],
])

Transformations

Transpose:

t = m.transpose()

Einstein Summation:

This article is a great introduction:

https://ajcr.net/Basic-guide-to-einsum/