Master NumPy: Turn Math Formulas into Python Code

Machine learning and data analysis rely heavily on mathematical formulas, yet implementing them in code can be challenging; this guide shows how to convert those formulas into Python using NumPy.

About NumPy

NumPy is the foundational package for scientific computing in Python. It provides a powerful N‑dimensional array object, broadcasting functions, tools to integrate C/C++ and Fortran code, and extensive linear algebra, Fourier transform, and random number capabilities.

Powerful N‑dimensional array object

Advanced broadcasting functions

Integration tools for C/C++ and Fortran

Robust linear algebra, FFT, and random number utilities

Because NumPy offers a concise, intuitive API, it is the most common library for scientific computing in machine‑learning workflows.

Environment Setup

Create a virtual environment (optional) and install NumPy: pip install numpy Test the installation: >> import numpy In the examples below, NumPy is imported as np:

import numpy as np

Basic Operations

Power Operations

The exponent operator is **. For example, x**2 computes the square of x, and x**3 computes the cube.

Square roots can be expressed as x**(1/2) or x**0.5, but NumPy provides the convenient function np.sqrt(x).

NumPy also offers np.square(x) for element‑wise squaring.

Absolute Value

The absolute value of a number x is |x|. NumPy computes it with np.abs(x).

Understanding Vectors and Matrices

Vector

A vector is a one‑dimensional array of numbers; geometrically, it represents a point in a coordinate system, with its direction pointing from the origin to that point.

Matrix

A matrix is a collection of vectors, i.e., a two‑dimensional array representing multiple points. Matrix operations correspond to operations on groups of vectors.

Initialization

NumPy can create vectors and matrices from Python lists: m = np.array([(1,2,3),(2,3,4),(3,4,5)]) This creates a 3×3 matrix.

Basic Arithmetic

Addition: x + 2 Subtraction: x - 2 Division:

x / 2

Matrix Power

Matrix exponentiation uses the same ** operator. For the matrix m, its square is written as m**2.

Matrix Dot Product

Multiplication of matrices of compatible dimensions is performed with the dot product function:

Code example:

m.dot(n)

Sum and Product

Summing all elements of a matrix m is done with m.sum(). Multiplying all elements uses m.prod().

Practice

Compute Mean

The mean of a vector x can be calculated as (1/x.size) * x.sum() or simply x.sum() / x.size:

(1/x.size)*x.sum()

x.sum()/x.size

Frobenius Norm

The Frobenius norm of a matrix m is computed by squaring each element, summing, and taking the square root:

np.sqrt((m**2).sum())

Sample Variance

Using the definition, variance can be expressed as:

np.sqrt(((x - (x.sum()/x.size))**2).sum()/(x.size-1))

With NumPy's mean helper, the expression simplifies to: np.sqrt(((x - np.mean(x))**2).sum()/(x.size-1)) Standard deviation is directly available via np.std(x).

Euclidean Distance

The Euclidean distance between vectors a and b is np.sqrt(((a-b)**2).sum()), which NumPy also provides as np.linalg.norm(a-b):

np.linalg.norm(a-b)

Summary

NumPy is a deep and versatile mathematical library that forms the backbone of scientific computing in Python. By translating formulas into NumPy code, you gain a clear understanding of underlying operations, which accelerates learning and practical work in data analysis, machine learning, and research.

References

https://blog.csdn.net/garfielder007/article/details/51386683

https://blog.csdn.net/robert_chen1988/article/details/102712946

https://mathtocode.com/