Abstract
A problem is presented with deterministic VLSI complexity AT 2 det =Ω(N2), but Las Vegas complexity only AT 2Las Vegas =O (N poly(logN)). (The Las Vegas algorithm always decides correctly, but T is only the expected running time; A is the area of the chip). Previously AT 2Las Vegas =O(N3/2 poly(logN)) has been shown for a similar problem with a more complicated algorithm. Here, we use a simple universal hashing technique based on random linear functions. We hope this will give rise to other applications of universal hashing in VLSI.
Our algorithm is very practical, because the random bits can even be wired into the chip. For every sequence of inputs during a chip's lifetime, the chances are high that the same short random bit string will always produce the result quickly.
Preview
Unable to display preview. Download preview PDF.
References
J. L. Carter and M.N. Wegman, Universal Classes of Hash Functions, JCSS 18 (1979) 143–154.
M. Fürer, The Power of Randomness for Communication Complexity, 19th STOC 1987, 178–181.
K. Mehlhorn and E. M. Schmidt, Las Vegas is better than Determinism in VLSI, 14th STOC 1982, 330–337.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1988 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Fürer, M. (1988). Universal hashing in VLSI. In: Reif, J.H. (eds) VLSI Algorithms and Architectures. AWOC 1988. Lecture Notes in Computer Science, vol 319. Springer, New York, NY. https://doi.org/10.1007/BFb0040398
Download citation
DOI: https://doi.org/10.1007/BFb0040398
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-0-387-96818-6
Online ISBN: 978-0-387-34770-7
eBook Packages: Springer Book Archive