Title: Sorting and Ranking of Self-Delimiting Numbers with Applications to Tree Isomorphism

Abstract: Assume that an $N$-bit sequence $S$ of $k$ self-delimiting numbers is given as input. We present space-efficient algorithms for sorting, dense ranking and (competitive) ranking $S$ on the word RAM model with word size $\Omega(\log N)$ bits. Our algorithms run in $O(k + \frac{N}{\log N})$ time and use $O(N)$ bits. The sorting algorithm returns the given numbers in sorted order, stored within a bit-vector of $N$ bits, whereas our ranking algorithms construct data structures that allow us subsequently to return the (dense) rank of each number $x$ in $S$ in constant time if the position of $x$ in $S$ is given together with $x$.
As an application of our algorithms we give an algorithm for tree isomorphism, which runs in $O(n)$ time and uses $O(n)$ bits on $n$-node trees. The previous best linear-time algorithm for tree isomorphism uses $\Theta(n \log n)$ bits.