A Monotone Path Proof of an Extremal Result for Long Markov Chains

Entropy (Basel). 2019 Mar 13;21(3):276. doi: 10.3390/e21030276.

Authors

Jia Wang¹, Jun Chen²

Affiliations

¹ Department of Electronic Engineering, Shanghai Jiao Tong University, Shanghai 200240, China.
² Department of Electrical and Computer Engineering, McMaster University, Hamilton, ON L8S 4K1, Canada.

Abstract

We prove an extremal result for long Markov chains based on the monotone path argument, generalizing an earlier work by Courtade and Jiao.

Keywords: Karush–Kuhn–Tucker; Markov chain; entropy power inequality; mean squared error; semidefinite programming.

Grants and funding

61771305/National Natural Science Foundation of China