[go: up one dir, main page]
More Web Proxy on the site http://driver.im/ skip to main content
On traversal sequences, exploration sequences and completeness of kolmogorov random strings
Publisher:
  • Rutgers University
  • Dept. of Computer Science Laboratory for Computer Sci. Research Hill Center, Busch Campus New Brunswick, NJ
  • United States
Order Number:AAI3092958
Pages:
145
Reflects downloads up to 21 Dec 2024Bibliometrics
Skip Abstract Section
Abstract

Traversal sequences were defined by Aleliunas et al. (1979) as a tool for the study of undirected s - t -connectivity. In the first part of this thesis we study traversal sequences. We improve on previous results and we present a simple construction of polynomial length universal traversal sequences for 2-regular graphs. These universal traversal sequences are log-space (even NC 1 ) constructible and are of length O ( n 4.03 ).

Further, we introduce a new notion of traversal sequences that we call exploration sequences . Exploration sequences share many properties with the original traversal sequences but they also exhibit some new properties. For instance, they have an ability to backtrack, and their random properties are robust under choice of the probability distribution on labels. We present simple constructions of polynomial length universal exploration sequences for several classes of graphs, e.g., 2-regular graphs, cliques, expanders, trees. These constructions do not obey previously known lower bounds on the length of universal traversal sequences thus, they highlight another difference between exploration and traversal sequences. We also show that universal traversal sequences can be efficiently converted into universal exploration sequences. Further, we show certain self-correcting properties of traversal and exploration sequences and we propose a candidate universal exploration sequence.

In the second part of this thesis we study Kolmogorov complexity, which is a measure of randomness of finite strings, and its resource-bounded variants. We show that sets consisting of strings of high resource-bounded Kolmogorov complexity provide examples of sets that are complete for complexity classes in which they naturally lie under probabilistic and non-uniform reductions whereas they are provably not complete under the usual many-one reductions.

Further, our results imply that under certain intractability assumptions, resource-bounded Kolmogorov complexity of a string as well as circuit size, formula size and branching program size of a Boolean function can not be efficiently approximated.

We also study unbounded Kolmogorov complexity. We show completeness results in that setting and point out certain inconsistencies caused by the usual definition of Kolmogorov complexity. We also consider the problem of deterministically generating a Kolmogorov random string when the set of Kolmogorov random strings is given to us as an oracle.

Contributors
  • Charles University
  • Rutgers University–New Brunswick
Please enable JavaScript to view thecomments powered by Disqus.

Recommendations