What is the probability that a ball lands in a bin?

481 Views Asked by At

Consider the classic problem of throwing balls to bins. As usual, balls are thrown uniformly at random, and independently from one another, to the bins.

Let $N,B$ denote the number of balls and bins respectively.

Contrarily to the classic problem, however, balls can be discarded. Each ball is discarded with probability $\alpha$. Alike for the throws, balls are discarded independently from each other.

For each bin $b_{i}$, let $S_{i}$ denote a random variable corresponding to the number of balls landing in $b_{i}$.

What is the value of $P\left\{S_{i} = 0\right\}$?

(Note: if all balls were tossed, $P\left\{S_{i} = 0\right\} = \left(1 - \frac{1}{B}\right)^{N}$)

1

There are 1 best solutions below

3
On BEST ANSWER

Consider a simpler problem where N = 1 (i.e. there is only a single ball), and a fixed bin $b_{i}$.

The probability that the ball is discarded is $\alpha$.

If the ball is not discarded, the probability that it lands in $b_{i}$ is $\frac{1}{B}$. On the other hand, if the ball is tossed, the probability that it does not land in $b_{i}$ is $1 - \frac{1}{B}$.

Now, we can compute the probability that the ball does not land in $b_{i}$:

  • The ball can either be discarded (with probability $\alpha$).
  • Or it was tossed but simply did not hit the bin (with probability $\left(1 - \alpha\right) . \left(1 - \frac{1}{B}\right)$).

Summing up, the probability that the ball does not target $b_{i}$ is $\alpha + \left(1 - \alpha \right) . \left(1 - \frac{1}{B}\right) = \left(1 - \frac{1 - \alpha}{B}\right)$.

Now, we only have to generalize this result to an arbitrary number of balls, denoted by $N$. Since throws are independent from each other, the probability that none (out of $N$) hits $b_{i}$ is: $$P\left\{S_{i} = 0\right\} = \left( 1 - \frac{1 - \alpha}{B}\right)^{N}$$