The Dispersion of Slepian-Wolf Coding
The authors characterize second-order coding rates (or dispersions) for distributed lossless source coding (the Slepian-Wolf problem). They introduce a fundamental quantity known as the entropy dispersion matrix, which is analogous to scalar dispersion quantities. They show that if this matrix is positive definite, the optimal rate region under the constraint of a fixed block-length and non-zero error probability has a curved boundary compared to being polyhedral for the Slepian-Wolf case. In addition, the entropy dispersion matrix governs the rate of convergence of the non-asymptotic region to the asymptotic one.