A//B, np.

floor_divide(A,B)
A%B, [Link](A,B) # Fancy indexing
NumPy Cheatsheet A**B,
A==B,
[Link](A,B)
[Link](A,B)
A[:, [0,1,0,0,1]]
A[[0,1,0],[2,3,3]]
# use indices on axis
# 1D array A[0,2],A[1,3],A[0,3]
©2022 [Link] A!=B, np.not_equal(A,B)
A>B, [Link](A,B) # Pick elements from A if cond. is True, B otherwise
A>=B, np.greater_equal(A,B) [Link](A>B, A, B)
import numpy as np
# print help for a function A<B, [Link](A,B)
A<=B, np.less_equal(A,B) # Select based on multiple conditions with default
print([Link]([Link])) # All three below are equivalent
# Power and log [Link](A, 0.1, 0.9)
Generating array [Link](A), [Link](2, A), [Link](A, 3) [Link]([A<0.1, A>0.9], [0.1, 0.9], A)
[Link]([1, 1.5, 2]) [Link](A), np.log10(A), np.log2(A) [Link](A>0.1, [Link](A<0.9, A, 0.9), 0.1)
[Link]([1, 1.5, 2]) # Square, square root, cube root
[Link](A), [Link](A), [Link](A) Transforming arrays
[Link](0,1,5) # Trigonometric functions Z = [Link](A) # clone array
# »> array([0., 0.25, 0.5, 0.75, 1.]) [Link](A), [Link](A), [Link](A) Z = [Link](A) # 1D array of unique elements
[Link](0,1,0.25) [Link](A), [Link](A), [Link](A)
# »> array([0., 0.25, 0.5, 0.75]) # absolute value, sign Z = [Link](A) # sort on each row
[Link](A), [Link](A) Z = [Link](A, axis=0) # sort on each columns
# Create array of shape (3,4) # NaN and infinity [Link](axis=0) # in-place sort
[Link]((3,4)) # all elements are 0 [Link](A) Z = [Link](axis=0) # return indices that sorts
[Link]((3,4)) # all elements are 1 [Link](A), [Link](A), [Link](A)
[Link]((3,4), 2) # all elements are 2 # ceiling, floor, and rounding to 2 d.p. # Transpose matrix: all below are equivalent
[Link](A), [Link](A), [Link](A, 2) A.T, [Link](0,1), [Link](A)
[Link](3) # 3x3 identity matrix # clip to between lower and upper bound # flatten array to 1D
[Link](A, 0.1, 0.9) [Link](), [Link](-1)
# make random arrays # use elements from A to fill an array of shape (3,4)
[Link]((3,4)) # between 0 and 1 Aggregate operations [Link](A, (3,4))
[Link](10, size=(3,4)) # between 0 and 9 # All can take argument ”axis=0” to aggregate along # add a new dimension and becomes axis=2
[Link](3,4) # normal distrib. # an axis. Otherwise aggregate on flattened array np.expand_dims(A, 2)
[Link]((3,4)) # uninitialized # padding for width=2 in each dimension
# Mean, standard deviation, median [Link](A, 2)
a = [Link]([1,2]) [Link](), [Link](), [Link](A)
b = [Link]([3,4,5]) [Link](axis=0), [Link](axis=0), [Link](A, axis=0) # concatenate arrays vertically
[Link](a,b) # Correlation matrix of two vector, or rows of a matrix [Link]((A,B)), np.r_[A,B]
# » [array([[1, 2],[1, 2],[1, 2]]), [Link](u, v), [Link](A) [Link](A, B, axis=0)
# array([[3, 3],[4, 4],[5, 5]])] [Link]((A,B), axis=0)
# Aggregate to boolean value, non-zero is True # concatenate horizontally
Inspecting array [Link](), [Link]() [Link]((A,B)), np.c_[A,B]
A = [Link](3,4) # Numerical sum, product, min, max
[Link] # shape = (3,4) [Link](), [Link]() # insert value to/overwrite/delete column 2
[Link] # num of elements = 12 [Link](), [Link]() [Link](A, 2, 0.2, axis=1)
len(A) # axis 0 length = 3 # argmin/argmax: Return index of flattened array, A[:, 2] = 0.25
[Link] # num dimension = 2 # or indices on the axis [Link](A, 2, axis=1)
[Link] # dtype('float64') [Link](), [Link]()
[Link](axis=0), [Link](axis=0) # split array into 2 equal-shape subarray
[Link](int) # cast array to integers # cumulative sum/product [Link](A, 2), [Link](A, 2)
[Link](), [Link]()
Vector dot-product/Matrix multiplication Linear algebra
Slicing # pseudo-inverse, inverse, determinant, rank, norm
A @ B # all three are identical
[Link](A,B) A[2] # select one ”row” [Link](A), [Link](M)
[Link](A,B) A[2:3] # subarray on axis 0 [Link](M), [Link].matrix_rank(A)
A[2:3, 1:3] # subarray on multiple axes [Link](A), [Link](A, ord=1)
A[2:3, ..., 1:3] # subarray on first & last axes
Element-wise operations A[2, 1:3], A[2][1:3] # these two are equivalent L = [Link](M) # lower triangular
A+B, [Link](A,B) U, diag, V = [Link](A) # compact SVD
A-B, [Link](A,B) # Boolean indexing w, V = [Link](M) # eigenvector=columns in V
A*B, [Link](A,B) A[(B>0.1) & (B<0.9)] # return 1D array
A/B, [Link](A,B) A[:, [Link](axis=1)>1] # select ”columns” v1.0.20220405