Compression of Samplable Sources

  • Authors:

  • Luca Trevisan;Salil Vadhan;David Zuckerman

  • Affiliations:

  • University of California at Berkeley;Harvard University;University of Texas

  • Venue:
  • CCC '04 Proceedings of the 19th IEEE Annual Conference on Computational Complexity
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

We study the compression of polynomially samplablesources. In particular, we give efficientprefix-free compression and decompression algorithmsfor three classes of such sources (whosesupport is a subset of {0, 1}^n).1. We show how to compress sources X samplableby logspace machines to expectedlength H(X) + O(1).Our next results concern flat sources whose supportis in P.2. If H(X) 驴 k = n - O(log n), we showhow to compress to length k + 驴 驴 (n - k)for any constant 娄驴 0; in quasi-polynomialtime we show how to compress to length k +O(polylog log(n - k)) even if k = n - polylog(n).3. If the support of X is the witness set for a self-reducibleNP relation, then we show how tocompress to expected length H(X) + 4.