A Hopf-power Markov chain on compositions

Research output: Chapter in book/report/conference proceedingConference proceedingpeer-review

Abstract

In a recent paper, Diaconis, Ram and I constructed Markov chains using the coproduct-then-product map of a combinatorial Hopf algebra. We presented an algorithm for diagonalising a large class of these "Hopf-power chains", including the Gilbert-Shannon-Reeds model of riffle-shuffling of a deck of cards and a rock-breaking model. A very restrictive condition from that paper is removed in my thesis, and this extended abstract focuses on one application of the improved theory. Here, I use a new technique of lumping Hopf-power chains to show that the Hopf-power chain on the algebra of quasisymmetric functions is the induced chain on descent sets under riffle-shuffling. Moreover, I relate its right and left eigenfunctions to Garsia-Reutenauer idempotents and ribbon characters respectively, from which I recover an analogous result of Diaconis and Fulman (2012) concerning the number of descents under riffle-shuffling.
Original languageEnglish
Title of host publication25th International Conference on Formal Power Series and Algebraic Combinatorics (FPSAC 2013)
PublisherMaison de l'informatique et des mathematiques discretes
Pages469-480
Number of pages12
DOIs
Publication statusPublished - 1 Jan 2013
Event25th International Conference on Formal Power Series and Algebraic Combinatorics, FPSAC 2013 - Paris, France
Duration: 24 Jun 201328 Jun 2013

Publication series

NameDiscrete Mathematics & Theoretical Computer Science Proceedings
ISSN (Print)1365-8050

Conference

Conference25th International Conference on Formal Power Series and Algebraic Combinatorics, FPSAC 2013
Country/TerritoryFrance
CityParis
Period24/06/1328/06/13

User-Defined Keywords

  • Quasisymmetric functions
  • riffle shuffling
  • descent set
  • combinatorial Hopf algebras

Fingerprint

Dive into the research topics of 'A Hopf-power Markov chain on compositions'. Together they form a unique fingerprint.

Cite this