dmtcs:2776 
Discrete Mathematics & Theoretical Computer Science,
January 1, 2010,
DMTCS Proceedings vol. AM, 21st International Meeting on Probabilistic, Combinatorial, and Asymptotic Methods in the Analysis of Algorithms (AofA'10)

https://doi.org/10.46298/dmtcs.2776
On unary nodes in tries
Authors: Stephan Wagner ^{1}
NULL
Stephan Wagner
1 Department of Mathematical Sciences [Matieland, Stellenbosch Uni.]
The difference between ordinary tries and Patricia tries lies in the fact that all unary nodes are removed in the latter. Their average number is thus easily determined from earlier results on the size of tries/Patricia tries. In a wellknown contention resolution algorithm, whose probabilistic model is essentially equivalent to tries, unary nodes correspond to repetitions, i.e., steps in the algorithm that do not resolve anything at all. In this paper, we take an individual's view on such repetitions: we consider the distribution of the number of repetitions a certain contender encounters in the course of the algorithm―which is equivalent to the number of unary nodes on the path from the root to a random string in a trie. We encounter an example of a sequence of distributions that does not actually converge to a limit distribution, but rather oscillates around a (discrete) limit distribution.
Volume: DMTCS Proceedings vol. AM, 21st International Meeting on Probabilistic, Combinatorial, and Asymptotic Methods in the Analysis of Algorithms (AofA'10)