|
|
一开始我一直顺着原文的叙述试图理解概率为何为1/(k+1), 很困惑。谢谢数值分析坛友的提醒,终于想明白了。下面试着用同一思路但不同的语言叙述一下,作为总结。
% m7 t' Z' _8 G; X& N2 o# J" o7 @% Y. i; _8 B' F& g
Let S be the set of the n elements in which there are k and only k elements that have value x. For each element w, let I be the indicator if w is examined or not, that is, I(w) = 1 if w is examined and 0 if w is not examined. X, the number of elements being examined, will be the sum of I(w) for all w in S. Accordingly, E[X] will be the sum of E[I(w)]=P{I(w)=1}. ( Y* Q* G( U' |+ [. R0 B4 B
6 X( I8 ]& U6 w! U+ V
For w that has a value x, the chance of w being examined is the chance that w is at the first position of a permutation of k x-valued elements. Therefore it's 1/k.
5 g+ R4 j; K, _( y1 d
. k* ~( c3 f, H; K: O) z4 G! |For w that has a value not being x, the chance of x being examined is the chance that w is at the first position of a permutation of all k x-valued elements plus w. Therefore it's 1/(k+1).
6 i* w! q' J2 \& @) d3 \
2 b9 z% N/ a' o% ]There are k elements that have value x and n-k elements that are not equal to x, so the sum of all these probabilities will be k*(1/k) + (n-k)*(1/(k+1)) = (n+1)/(k+1).
1 R8 v2 G4 h% r# G# a% @' ]* X3 L' K
. F0 o8 S4 O4 k' j6 ?/ c& v理解上述解法的一个关键点是对于所有不等于x的element,它能不能有机会被查验取决于而且只取决于它与k个值为x的elements的相对位置。 |
评分
-
查看全部评分
|