# Find: The expected number of urns that are empty

A total of $n$ balls, numbered $1$ through $n$, are put into $n$ urns, also numbered $1$ through $n$ in such a way that ball $i$ is equally likely to go into any of the urns $1, 2, . . . , i$. Find the expected number of urns that are empty.

Can somebody help me? I don´t understand.

Thanks very much.

-
Do you mean that ball $i$ is equally likely to into any of the urns $1,2, \dots i$ or $1,2,\dots, n$? That is, is it impossible that, for example, ball 3 goes into urn 4? Or is any ball equally likely to go into any urn? –  sol May 20 '13 at 1:19
I think that it must be the latter. 'Cause if Ball #1 can't go in Urn #2 or up, then it goes in Urn #1. Then Ball #2 can't go in Urn #3 or above, and Urn #1 is occupied, so... The whole thing falls apart. –  User58220 May 20 '13 at 2:42
If it helps anyone formulate a proof, the expected value of empty urns for $n=1,2,3,4$ is $0,1/2,1,3/2$ respectively. From this, we might expect $(n-1)/2$ empty urns, given $n$. –  A Walker May 20 '13 at 2:57
@User58220 It never says that the urns can't hold multiple balls. Thus it may be that "the prior" holds. –  A Walker May 20 '13 at 2:59

## 2 Answers

Let $X$ be the number of empty urns. Then $X=X_1+X_2+\dots+X_n$ where $X_i$ is the indicator variable whose value is $1$ if the $i$th urn is empty, $0$ otherwise. Then $E(X_i)$ is the probability that the $i$th urn is empty, i.e., $E(X_i)=\frac{i-1}{i}\frac{i}{i+1}\frac{i+1}{i+2}\dots\frac{n-1}{n}=\frac{i-1}{n}$ (telescoping product). By linearity of expectation, $$E(X)=E(X_1)+\dots+E(X_n)=\frac{0}{n}+\frac{1}{n}+\frac{2}{n}+\dots+\frac{n-1}{n}=\frac{{n\choose 2}}{n}=\frac{n-1}{2}.$$

-

We proceed by induction. For the case $n=1$, it is clear that the expected value of empty urns is $0$. Now, suppose that with $n$ urns our expected value, $E(n)$, is given by $(n-1)/2$.

For the case $n+1$, we can think of our procedure in two steps:

1. Place $n$ balls (according to the rule for $n$ balls) into the first $n$ urns.
2. Place one ball in any urn.

After step 1, there are on average $E(n)+1$ empty urns. As the last ball will end up in any urn with equal probability, there is a probability of $(E(n)+1)/(n+1)$ that our final ball will land in an empty urn. Subtracting, there remains a $(n-E(n))/(n+1)$ chance that the final ball will land in a previously-occupied urn. That is, our expected value on empty urns is given by

$$E(n+1)=E(n) \left(\frac{E(n)+1}{n+1}\right)+(E(n)+1)\left(\frac{n-E(n)}{n+1}\right)=\frac{n-1}{2}\cdot \frac{1}{2} + \frac{n+1}{2}\cdot \frac{1}{2},$$ in which we've used the inductive hypothesis $E(n)=(n-1)/2$. The right-hand side of the expression above is $n/2$, and this proves the $n+1$ case (so that we conclude by induction).

I'd like to point out that I would not have known how to formulate this induction without doing some cases by hand. I noticed that this pattern held for $n \leq 4$ before attempting the proof you see above.

-
@A Walker Thanks you very much ,You helped me a lot :D –  user63192 May 20 '13 at 3:47