4

【Python】运行效率研究.

 3 years ago
source link: https://www.guofei.site/2017/04/04/pythonspeed.html
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
neoserver,ios ssh client

【Python】运行效率研究.

2017年04月04日

Author: Guofei

文章归类: 趣文，文章编号: 1111

版权声明：本文作者是郭飞。转载随意，但需要标明原文链接，并通知本人
原文链接：https://www.guofei.site/2017/04/04/pythonspeed.html

Edit

1. 自加和赋值加法的效率比较

import time

n=100000
start=time.time()
i=1
for j in range(n):
    i+=1
print(time.time()-start)

start=time.time()
i=1
for j in range(n):
    i=i+1

print(time.time()-start)

0.01636791229248047 0.01322317123413086

结论：自加和每次赋值效率几乎相同，这一点与C语言有区别

2. ufunc方法

数据准备

import numpy as np
def tri_func(x,a=1,b=0):
    x=x-int(x)
    if x<0.3:
        y=x
    elif x<0.7:
        y=0.3
    else:
        y=1-x
    y=a*y+b
    return y

运行效率比较

import time

x=np.linspace(0,5,100000)
start_time=time.time()
y1=[tri_func(i) for i in x]
print(time.time()-start_time)

start_time=time.time()
func1=np.frompyfunc(tri_func,3,1)
y2=func1(x,1,0)
print(time.time()-start_time)

0.12008452415466309

0.06804776191711426

ufunc方法比[for ]效率高

反向排序效率比较

import time
n=5
count=10**n
start=time.time()
nums=[]
for i in range(count):
    nums.append(i)
nums.reverse()
print(time.time()-start)

start=time.time()
nums=[]
for i in range(count):
    nums.insert(0,i)
print(time.time()-start)

start=time.time()
import numpy as np
nums=np.array([])
for i in range(count):
    np.append(nums,i)
a=nums[:-1:]
print(time.time()-start)

start=time.time()
import numpy as np
nums=[]
for i in range(count):
    nums.append(i)
num_reverse=np.array(nums)[:-1:]
print(time.time()-start)

out: 0.02801966667175293 2.725426435470581 0.7875568866729736 0.02852034568786621

list.insert速度最慢，2.7秒才运行完 numpy虽然不是最快，但慢在了np.append环节，反转操作并不慢，由于np.append的适用范围更广，所以还是比较有用的

直接在jupyter里插入%%timeit，便可以主动运行上万次并测速，这比MATLAB效率研究方便太多了

生成随机数速度比较

import numpy as np
from scipy.stats import randint
rv=randint(low=0,high=2)
  1. numpy 
    %%timeit
a=np.random.randint(low=0,high=2,size=(50,30))

    7.15 µs ± 768 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

  2. scipy.stats 
    %%timeit
a=rv.rvs(size=(50,30))

    1.94 ms ± 58.4 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)

您的支持将鼓励我继续创作！

report

Recommend

About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK