/ 写在前面 – 我热爱技术、热爱开源。我也相信开源能使技术变得更好、共享能使知识传播得更远。但是开源并不意味着某些商业机构/个人可以为了自身的利益而一味地索取，甚至直接剽窃大家曾为之辛勤付出的知识成果，所以本文未经允许，不得转载，谢谢。/

The concept of axis is essential for understanding NumPy.

NumPy’s main object is the homogeneous multidimensional array. It is a table of elements (usually numbers), all of the same type, indexed by a tuple of non-negative integers. In NumPy dimensions are called axes.

NumPy’s array class is called ndarray . It is also known by the alias array . Note that numpy.array is not the same as the Standard Python Library class array.array , which only handles one-dimensional arrays and offers less functionality.

Important attributes of an ndarray object are:

  • ndarray.ndim
  • ndarray.shape
  • ndarray.size
  • ndarray.dtype
  • ndarray.itemsize
  • ndarray.data
  • ndarray.flat
  • ndarray.T

Commonly used methods:

  • ravel() , flatten() - returns the array, flattened
  • reshape()
  • resize()
  • dot()
  • sum()
  • cumsum()
  • min()
  • max()

Commonly used functions:

  • np.zeros()
  • np.ones()
  • np.eye()
  • np.empty()
  • np.arange() - similar but different with built-in range
  • np.linspace()
  • np.fromfunction()

Below are key points of NumPy to understand.

Creation of array - tuple or list
_
Keyword arguments - dtype=complex , etc.

How does NumPy organize axes? Which axis is the last axis?

About upcasting, how is it applied in operations? Like += , *= , etc.

What’s the dafault dtype of the created array? - float64

Arithmetic operators on arrays apply elementwise.

Product operator * operates elementwise in NumPy arrays.

The matrix product can be performed using the @ operator (in python >=3.5) or the dot function or method.

Creation of the instance of default random number generator:

  • rg = np.random.default_rng(1)
  • a = rg.random((2,3))

Unary operations with specifying the keyword argument axis

“universal functions”( ufunc ) - elementwise:

  • np.sin()
  • np.cos()
  • np.ext()
  • np.sqrt()
  • np.add()
  • …

Understand indexing, slicing and iterating:

  • How to reverse an one-dimensional array?
  • Access multidimensional arrays in multiple ways - complete indices, fewer indeces and dots ...
  • Iterator - flat attribute

Commands that can return different shapes of an array:

  • ndarray.ravel() - ndarray.flatten()
  • ndarray.T
  • ndarray.reshape()

Difference between reshape() and resize()

Meaning of passing -1 to a reshaping operation

Stacking together different arrays:

  • np.vstack()
  • np.hstack()
  • np.column_stack() - different with np.hstack()
  • np.row_stack() - an alias for np.vstack()
  • np.concatenate()
  • np.r_ and np.c_ - difficult to understand…
  • …

Splitting an array:

  • np.hsplit()
  • np.vsplit()
  • np.array_split()

Is it a copy of an array?

  • id()
  • is
  • .view()
  • .base

Reference

  1. NumPy quickstart — NumPy v1.20 Manual
  2. NumPy: the absolute basics for beginners — NumPy v1.20 Manual