How to improve python running speed

Speaking of the most popular language now, we have to mention python. However, although python is easy to use, its speed is a bit impressive. How to use a simple method to accelerate python to a speed that is almost comparable to C?

Today let’s talk about the baby numba. You read that right, it’s either numpy or numba. (Recommended learning: Python video tutorial)

numba is a just-in-time compiler for Python, which is best suited for code that uses NumPy arrays and functions as well as loops. The most common way to use Numba is through its collection of decorators, which can be applied to your functions to instruct Numba to compile them. When a Numba decorated function is called, it is compiled to machine code for "just-in-time" execution, and all or part of your code can then run at native machine code speed!

When faced with a computing project, the easiest thing we can think of is to code directly and finally write a very long program. As a result, once something goes wrong, it often takes a lot of time to locate the problem.

There is a simple way to solve this problem, which is to define various functions and break the task into many small parts. Because each function is not particularly complex and can be checked at any time when written, it is easy to locate and solve problems once problems arise in the concise main program. The idea of ​​object-oriented programming is based on functions.

After writing the function, you can also use decorator to make it more powerful. The decorator itself is a function, but it is a function of functions. The purpose is to increase the function of the function. For example, first define a function that outputs the current time, and then define a function that specifies the time format. Applying the latter function to the previous function is a decorator, which is used to output the current time in a specific format.

>The advantages of Numba

1. Simple, often only one line of code can bring surprises;

2. It has miraculous effects on loops, and is often used in science What limits the speed of python in calculation is loop;

3. Compatible with commonly used scientific computing packages, such as numpy, cmath, etc.;

4. Can create ufunc;

5. It will automatically adjust the accuracy to ensure accuracy.

How to use numba

Let me introduce the advantages of numba mentioned above one by one. First import numba

import numba as nb

It only takes one line of code to speed up, and it has a miraculous effect on loops

Because the built-in function of numba is a decorator, you only need to add it in front of the function you have defined. Just @nb.jit(), it’s easy to get started. Let's take a summation function as an example

# 用numba加速的求和函数@nb.jit()def nb_sum(a):
    Sum = 0
    for i in range(len(a)):
        Sum += a[i]
    return Sum# 没用numba加速的求和函数def py_sum(a):
    Sum = 0
    for i in range(len(a)):
        Sum += a[i]
    return Sum

to test the speed

import numpy as np
a = np.linspace(0,100,100) # 创建一个长度为100的数组

%timeit np.sum(a) # numpy自带的求和函数
%timeit sum(a) # python自带的求和函数
%timeit nb_sum(a) # numba加速的求和函数
%timeit py_sum(a) # 没加速的求和函数

For more Python-related technical articles, please visit the Python Tutorial column to learn!

The above is the detailed content of How to improve python running speed. For more information, please follow other related articles on the PHP Chinese website!