Part2:比较两个渗透函数的性能

关于该部分的项目描述请见Project2 Percolation in Grids 网格渗透

 　测试的场景要求如下：

·         set n=75

·         consider values of p from 0 to 1 in increments of 0.05 (or smaller)

·         for each value of p, generate 10 random grids and record for each algorithm the average running time on the ten grids

 　　可是我在编写该部分的测试代码时，遇到了些麻烦，一开始还摸不清头脑。

 

My Problem that You should Avoid

 

　　我的思路是执行每次for循环前，也就是对于每个p值，产生一个当前时间t，然后每发生渗透，记录下时间差time_delta，并设置一个求和变量sum_time，累加每次的执行时间。代码如下，发现我错在哪了么？

def test_time(func):    #这是错误的
    p = 0
    results = []
    while p < 1:
        print 'Running for p =', p, '...'
        sum_time=0
        t = time.clock()
        for k in range(trial_count):
            g = random_grid(size, p)
            flow,perc = func(g, trace=False)
            if perc:
                time_delta = time.clock()-t
                sum_time+=time_delta

        Aver_time = float(sum_time)/trial_count
        print Aver_time        
        results.append(Aver_time)
        p += step
        
    return results

第一，错在不用为每一次运算计算运行时间，只要每执行trial_count次，跳出循环，计算一次时间差即可。

第二，算法的执行时间不是仅指发生渗透的情况下啊！ 我错了，我真的错了...

　　 正确的代码应该如下：

    p = 0
    results = []
    old_clock = time.clock()
    print 'testing', func, '...'

    while p < 1:
        print 'Running for p =', p, '...'
        perc_count = 0
        for k in range(trial_count):
            g = random_grid(size, p)
            flow,perc = func(g,trace=False)
        new_clock = time.clock()
        results.append((new_clock - old_clock)/trial_count)
        old_clock = new_clock
        p += step

    附上美丽的曲线图，经过一段曲折的纠缠，最终水波算法胜出！

 

 

测试的完整代码，贴在这里，至此Python的网格渗透项目也完全结束

# -*- coding: utf-8 -*-
"""
Created on Wed Aug 10 13:34:21 2011

@author: Nupta
"""

from percolation_wave import *
from my_percolation_recursive import *
from pylab import *
import time
import sys

step = 0.05
trial_count = 10
size = 75


def test_time(func):
    p = 0
    results = []
    old_clock = time.clock()

    print 'testing', func, '...'

    while p < 1:
        print 'Running for p =', p, '...'

        perc_count = 0
        for k in range(trial_count):
            g = random_grid(size, p)
            flow,perc = func(g,trace=False)
        new_clock = time.clock()
        results.append((new_clock - old_clock)/trial_count)
        old_clock = new_clock
        p += step

    return results

if __name__ == '__main__':
    
    sys.setrecursionlimit(250000)
    
    res1 = test_time(percolation_wave)
    res2 = test_time(percolation_recursive)

    plot(arange(0,1,step), res1, 'b', label='wave')
    plot(arange(0,1,step), res2, 'r', label='recursive')
    legend()
    xlabel('p')
    ylabel('Running Time')
    title('Running Time')
    show()