English

On Two-Stage Guessing

Information Theory 2021-04-16 v2 math.IT Probability

Abstract

Stationary memoryless sources produce two correlated random sequences XnX^n and YnY^n. A guesser seeks to recover XnX^n in two stages, by first guessing YnY^n and then XnX^n. The contributions of this work are twofold: (1) We characterize the least achievable exponential growth rate (in nn) of any positive ρ\rho-th moment of the total number of guesses when YnY^n is obtained by applying a deterministic function ff component-wise to XnX^n. We prove that, depending on ff, the least exponential growth rate in the two-stage setup is lower than when guessing XnX^n directly. We further propose a simple Huffman code-based construction of a function ff that is a viable candidate for the minimization of the least exponential growth rate in the two-stage guessing setup. (2) We characterize the least achievable exponential growth rate of the ρ\rho-th moment of the total number of guesses required to recover XnX^n when Stage 1 need not end with a correct guess of YnY^n and without assumptions on the stationary memoryless sources producing XnX^n and YnY^n.

Cite

@article{arxiv.2104.04586,
  title  = {On Two-Stage Guessing},
  author = {Robert Graczyk and Igal Sason},
  journal= {arXiv preprint arXiv:2104.04586},
  year   = {2021}
}

Comments

Citation for this work: R. Graczyk and I. Sason, "On two-stage guessing," Information, vol. 12, no. 4, paper 159, pp. 1 - 20, April 2021. The published version is available (Open Access) at the link: https://www.mdpi.com/2078-2489/12/4/159

R2 v1 2026-06-24T01:01:25.108Z