On the Strong Converse Exponent and Error Exponent of the Classical Soft Covering

TL;DR

This paper determines the exact strong converse exponent for the classical soft covering problem, introduces a new two-parameter information measure, and discusses improvements over random coding and uniform message distributions.

Contribution

It establishes the precise strong converse exponent using a novel information quantity and proposes a non-uniform message distribution framework to improve error exponents.

Findings

Exact strong converse exponent characterized for classical soft covering.

New two-parameter information measure introduced for exponent expression.

Deterministic codes outperform random codes in noiseless channels.

Abstract

This paper establishes the exact strong converse exponent of the soft covering problem in the classical setting. This exponent characterizes the slowest achievable convergence speed of the total variation to one when a code of rate below mutual information is applied to a discrete memoryless channel for synthesizing a product output distribution. The proposed exponent is expressed through a new two-parameter information quantity, differing from the more commonly studied R\'enyi divergence or R\'enyi mutual information. In addition, we demonstrate the non-tightness of random coding for rates both below and above mutual information. Discussions on the latter start with noiseless channels, where we develop a deterministic code construction that outperforms random codes in error exponents. We further observe that the conventional formulation, which assumes a uniform distribution over messages, inherently introduces a discrepancy in error exponents depending on whether the components of the target distribution are rational or irrational numbers. To eliminate this discrepancy, we propose a new formulation in which messages are allowed to be distributed non-uniformly, and the rate is given by the logarithm of the smallest nonzero message probability (corresponding to R\'enyi entropy $H_{-\infty}$ of order $-\infty$). The exact error exponent is characterized in this formulation for noiseless channels. Furthermore, for noisy channels, we provide a high-rate improvement in achievability and derive a converse bound on the error exponent.