Not All Universal Source Codes Are Pointwise Universal

Let Cn be a lossless code for n-tuples over a finite alphabet and ln denote its length function. The code sequence {Cn} is said to be universal if for every stationary source X = (X1, X2, . . .) 1 n Eln(X ) −→ H(X), where X = (X1, . . . , Xn) and H(X) denotes the entropy rate of X. It is said to be pointwise universal if for every stationary and ergodic source, with probability one, 1 n ln(X ) −→ H(X). Pointwise universality implies universality. We show that the converse is not true by constructing a universal sequence of schemes that is not pointwise universal, hence establishing the latter as a stronger notion of universality.

[1]  David L. Neuhoff,et al.  Simplistic Universal Coding. , 1998, IEEE Trans. Inf. Theory.

[2]  R. Durrett Probability: Theory and Examples , 1993 .

[3]  Thomas M. Cover,et al.  Elements of Information Theory , 2005 .

[4]  Ioannis Kontoyiannis,et al.  Pointwise redundancy in lossy data compression and universal lossy data compression , 2000, IEEE Trans. Inf. Theory.

[5]  Abraham Lempel,et al.  Compression of individual sequences via variable-rate coding , 1978, IEEE Trans. Inf. Theory.

[6]  John C. Kieffer,et al.  Sample converses in source coding theory , 1991, IEEE Trans. Inf. Theory.

[7]  Amir Dembo,et al.  Source coding, large deviations, and approximate pattern matching , 2001, IEEE Trans. Inf. Theory.

[8]  Sang Joon Kim,et al.  A Mathematical Theory of Communication , 2006 .

[9]  Michelle Effros,et al.  Universal lossless source coding with the Burrows Wheeler transform , 1999, Proceedings DCC'99 Data Compression Conference (Cat. No. PR00096).