The theoretical capacity of the Parity Source Coder

Ciliberti, Stefano; Mezard, Marc

doi:10.1088/1742-5468/2005/10/P10003

Full-text links:

Download:

(license)

Current browse context:

cond-mat

< prev | next >

new | recent | 0506

References & Citations

NASA ADS

Bookmark

(what is this?)

Condensed Matter > Disordered Systems and Neural Networks

Title: The theoretical capacity of the Parity Source Coder

Authors: Stefano Ciliberti, Marc Mezard

(Submitted on 24 Jun 2005 (v1), last revised 14 Sep 2005 (this version, v2))

Abstract: The Parity Source Coder is a protocol for data compression which is based on a set of parity checks organized in a sparse random network. We consider here the case of memoryless unbiased binary sources. We show that the theoretical capacity saturate the Shannon limit at large K. We also find that the first corrections to the leading behavior are exponentially small, so that the behavior at finite K is very close to the optimal one.

Comments:	Added references, minor changes
Subjects:	Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Information Theory (cs.IT)
Journal reference:	J. Stat Mech P10003 (2005)
DOI:	10.1088/1742-5468/2005/10/P10003
Cite as:	arXiv:cond-mat/0506652 [cond-mat.dis-nn]
	(or arXiv:cond-mat/0506652v2 [cond-mat.dis-nn] for this version)

Submission history

From: Stefano Ciliberti [view email]
[v1] Fri, 24 Jun 2005 14:17:37 GMT (21kb)
[v2] Wed, 14 Sep 2005 16:51:10 GMT (21kb)

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Link back to: arXiv, form interface, contact.