Valeurs propres des op\'erateurs de m\'elanges sym\'etris\'es

TL;DR

This paper develops a combinatorial method using symmetric group representation theory to compute eigenvalues of symmetrized shuffling operators, extending previous work and solving several conjectures.

Contribution

It provides a comprehensive eigenvalue computation method for all operators in the family, building on and generalizing prior results, and offers new proofs and conjectures.

Findings

Eigenvalues are all nonnegative integers.

The operators commute, as previously conjectured.

A new combinatorial algorithm for eigenvalue computation.

Abstract

English title: Eigenvalues of Symmetrized Shuffling Operators The random-to-random shuffling operator explains, for example, the evolution of a deck of cards subject to the following random process: draw a card randomly from the deck and reinsert it at a random position. If one instead draws more than one card at a time before reinserting, then the resulting operator is an example of a family of symmetrized shuffling operators studied by Victor Reiner, Franco Saliola and Volkmar Welker. This thesis describes a way to obtain the eigenvalues of these operators. We build on the work of Anton Dieker and Franco Saliola, who computed the eigenvalues of the random-to-random shuffle. Here, we compute the eigenvalues for all the operators of the family. We proceed with the help of the representation theory of the symmetric group. We decompose the vector space on which the shuffles act into simple modules for the symmetric group. These modules correspond to standard Young tableaux, and the algorithm to compute the eigenvalues is combinatorial because it computes the eigenvalues directly from the standard Young tableaux. As a corollary of our main result, we solve several conjectures of Reiner, Saliola and Welker, including showing that the eigenvalues are all nonnegative integers. Furthermore, the techniques used here allow us to give a new proof of their result that these symmetrized shuffling operators commute. Knowing the eigenvalues is the key step in one method of computing the number of shuffles one needs to execute to get a perfectly shuffled deck, which is briefly explored. We also study a second family of shuffles introduced by Reiner, Saliola and Welker. We present many conjectures about their eigenvalues.