There are $n$ elements in a hash table of size $m \geq 2n$ which uses open addressing to avoid collisions.

The hash function was chosen randomly among a set of uniform functions. A set $H$ of hash-functions $h:U\to\{0,\dots,m-1\}$ is called uniform, if for every tuple of different keys $x,y \in U$ the number of hash-functions $h \in H$ with $h(x) = h(y)$ is $\frac{|H|}{m}$ at most.

Show that the propability that for $i = 1, 2, \dots,n$ the $i$-th insert operation needs more than $k$ attempts, is $2^{-k}$ at most.

This is an assignment, which I got as homework. What I already worked out:

The propability $p_1$ for a collision is 0 of course for an empty table.

The propability $p_i$ for a collision after k attempts should be $\frac{i - 1}{2n}\cdot k$ assuming that the table is filled with $i-1$ elements to this point and the tables size is $2n$ as worst case.

So I have $$ p_i= \frac{i-1}{2n} \cdot k \leq 2^{-k}, $$

but I don't know where to go from here.

The method of open hashing used here simply iterates over different hash-functions until a free place is found (for example $h(x) = (x \bmod j) \bmod n$ with increasing prime numbers for $j$.

  • $\begingroup$ There are several different variants of open hashing. Can you give pseudocode for your version? $\endgroup$ Oct 7, 2013 at 4:39

1 Answer 1


Hint: Show that at a given value of $i$, the probability of failure of the $l$th attempt is at most $i/m \leq 1/2$ (perhaps using a union bound), and the events are independent for different attempts.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.