##language:zh
#pragma section-numbers off
##含有章节索引导航的 ZPyUG 文章通用模板
<<TableOfContents>>
## 默许导航,请保留
<<Include(ZPyUGnav)>>


= 求素新篇 =
##startInc
[[IntroPy|介绍Python]] 中的单行求素:

{{{
[p for p in range(2,100) if  0 not in [p%d for d in range(2,p-1)]]
}}}    

== 问 ==
{{{
Her0 <reactionism@gmail.com>
reply-to	python-cn@googlegroups.com
to	python-cn`CPyUG`华蟒用户组 <python-cn@googlegroups.com>
date	Sat, Jul 19, 2008 at 17:39
subject	[CPyUG:59701] 求0~100之间的所有素数 大家可能都做过。请对一下几个方法给出你的评价！
}}}
	
{{{
  1 #coding:utf-8
  2 '''cdays-5-exercise-3.py 求0~100之间的所有素数
  3     @note: for循环, 列表类型
  4     @see: math模块使用可参考http://docs.python.org/lib/module-math.html
  5 '''
  6
  7 from math import sqrt
  8
  9 N = 100
 10 #基本的方法
 11 result1 = []
 12 for num in range(2, N):
 13     f = True
 14     for snu in range(2, int(sqrt(num))+1):
 15         if num % snu == 0:
 16            f = False
 17            break
 18     if f:
 19         result1.append(num)
 20 print result1
 21
 22 #更好的方法
 23 result2 = [ p for p in range(2, N) if 0 not in [ p% d for d in
range(2, int(sqrt(p))+1)] ]
 24 print result2
}}}

== 新功能 ==
{{{
subowen <onenew@gmail.com>
reply-to	python-cn@googlegroups.com
to	python-cn@googlegroups.com
date	Sat, Jul 19, 2008 at 23:34
subject	[CPyUG:59739] Re: 求0~100之间的所有素数 大家可能都做过。请对一下几个方法给出你的评价！
}}}
	
	
时代总要进步的
老皇历也要改进
 * 2.5 的新函数 all 和any +  generator 不论从语义还是速度上都是巨大的进步
{{{
python -mtimeit "[p for p in range(2,100) if  0 not in [p%d for d in range(2,p-1)]]"
  1000 loops, best of 3: 1.17 msec per loop
python -mtimeit "[p for p in range(2,100) if all( p%d for d in range(2,p//2))]"
  1000 loops, best of 3: 477 usec per loop
python -mtimeit "[p for p in range(2,1000) if  0 not in [p%d for d in range(2,p-1)]]"
  10 loops, best of 3: 95 msec per loop
python -mtimeit "[p for p in range(2,1000) if all( p%d for d in range(2,p//2))]"
  100 loops, best of 3: 15.2 msec per loop
}}}
 * 2.5 的新函数 all 和any +  generator 不论从语义还是速度上都是巨大的进步
再来个邪恶的:)
{{{
python -mtimeit "reduce(lambda l,y: all(y % x for x in l[:len(l)//2])and l+[y] or l,xrange(2,100),[] )"
  1000 loops, best of 3: 386 usec per loop

python -mtimeit "reduce(lambda l,y: all(y % x for x in l[:len(l)//2])and l+[y] or l,xrange(2,1000),[] )"
  100 loops, best of 3: 5.61 msec per loop
}}}
---- --
知止而后有定，定而后能静，静而后能安，安而后能虑，虑而后能得。 

== 继续优化 ==
{{{
yuting cui <yutingcui@gmail.com>
reply-to	python-cn@googlegroups.com
to	python-cn@googlegroups.com
date	Sun, Jul 20, 2008 at 02:32
}}}
呵呵，再发个继续优化的版本

{{{#!python
def getprimes(n):
   if n <=2:
       return []
   if n == 3:
       return [2]
   u=n**0.5
   r=n//2 - 1
   l=range(3,n,2)
   c=3
   t=0
   s=3
   while c < u:
       if l[t]:
           for i in xrange(s,r,c):
               l[i]=0
       s+=(c<<1)+2
       c+=2
       t+=1
   return [2]+[e for e in l if e]
}}}

##endInc

----
'''反馈'''

创建 by -- ZoomQuiet [<<DateTime(2008-07-20T04:40:01Z)>>]