2

A random function $rand()$ return a integer between $1$ and $k$ with the probability $\frac{1}{k}$. After $n$ times we obtain a sequence $\{b_i\}_{i=1}^n$, where $1\leq b_i\leq k$. Set $\mathbb{M}=\{b_1\}\cup\{b_2\}\cdots \cup\{b_n\}$. I want to known the probability $\mathbb{M}\neq \{1, 2\cdots, k\}$.

Yuansheng liu
  • 65
  • @Arnaud There is any argue in this problem discription? $\mathbb{M}\neq {1,2,\cdots,k}$. Because the random function cann't return a integer $>k$. – Yuansheng liu Apr 21 '13 at 09:59
  • There are two possible questions, depending on whether $n>k$ or $n<k$ is interesting. The question is about the $n>k$ case. – not all wrong Apr 21 '13 at 10:10

3 Answers3

1

The probability that $j$ of the $k$ numbers is excluded is $(1-j/k)^n$ and there are $\binom{k}{j}$ ways to choose those $j$ to be excluded. The inclusion-exclusion principle says that the probability that at least one of the $k$ numbers is excluded is $$ \sum_{j=1}^k(-1)^{j-1}\binom{k}{j}\left(1-\frac{j}{k}\right)^n $$

Community
  • 1
robjohn
  • 345,667
0

Hint: Obviously $n<k$ is trivial. Thereafter, the question becomes equivalent to solving

What fraction of $n$-tuples with the digits $1,\ldots,k$ are in fact $n$-tuples formed from a strict subset of these numbers?

or a surjection-counting problem.

You can find a recursive solution by letting $p_{n,k}$ be the probability that a set of $n$ digits up to $k$ contains all distinct digits, and then considering the last digit of a sequence leading to $(n+1)$-tuples.

Edit: I should make clear that a recursive 'solution' essentially is the best you can do, which is why I called it that! The numbers don't have a closed form. (See e.g. https://mathoverflow.net/questions/27071/counting-sequences-a-recurrence-relation for a discussion, once you've worked out the recursive form yourself.)

Community
  • 1
not all wrong
  • 16,178
  • 2
  • 35
  • 57
0

Principle of inclusion and exclusion? That will do, I think, because probability that you didn't get $j$, $1\leq j \leq k $, $\quad$ is $\quad$ $( \frac{k-1}{k} )^n$ and so on ...

Antoine
  • 3,439
  • To clarify - as I note in my answer, there isn't a closed form for the resulting sum. You certainly can write a sum down using PIE though. – not all wrong Apr 21 '13 at 10:37