mathe 发表于 2008-12-5 11:40:07

一个极限证明

求证
$\lim_{n->infty}1/{2^n\sqrt(n)}\sum_{j=1}^n({:(n),(j):})\sqrt(j)=1/{\sqrt(2)}$
来自mymathforum.

mathe 发表于 2008-12-5 11:48:40

补充说明
$({:(n),(j):})$我们通常写成$C_n^j$,也就是组合数

northwolves 发表于 2009-2-4 22:02:20

mathe怎么证明的?

mathe 发表于 2009-2-5 09:16:08

这个应该不是很难的

无心人 发表于 2009-2-5 09:21:12

肚老大
   最近很忙啊
   在论坛发言少了
   呵呵

mathe 发表于 2009-2-5 20:01:20

**** Hidden Message *****

shangwen_ren 发表于 2009-3-4 10:39:35

So we may use squeeze thm by finding the lower and upper bound of the limit as 1/square root 2.
2^n can be written as sum of (n,j) from 0 to inf. we may assume it is from 1 to inf cause the difference is tiny when n goes inf.
also since square root of j is concave down, we can apply the jesens inequality. then the original limit will be small or equal to sqr ( sum (n,j)*j/ sum (n,j)) * 1/ sqr(n)
sum(n,j) *j is n*2^(n-1)   simplifed to 1/ sqr(2)

To find the lower bound, i am trying to adjust sqr(j) to 1/sqr(j) in order to apply the inequality again with concave up function. but there is some "re-index" problem need further clearifying. But i guess it's almost there, well, hopefully....

wayne 发表于 2009-3-24 09:35:06

回复 1# mathe 的帖子

我有一个证法: 放缩法
左边式子的$=1/2(sqrt(j)+sqrt(n-j))$,而
$sqrt(n/2)<=1/2(sqrt(j)+sqrt(n-j))<=1/2sqrt(2n)$

分别代进去,都得$sqrt(2)/2$

wayne 发表于 2009-3-24 09:44:52

左边弄错了:L

jackie 发表于 2009-3-25 10:11:43

学习ing
页: [1] 2
查看完整版本: 一个极限证明