Deterministic stack-sorting for set partitions

TL;DR

This paper introduces a new deterministic stack-sorting method for sock sequences, characterizes its sorting bounds, and enumerates sortable sock patterns, advancing understanding of pattern-avoiding stack-sorting for set partitions.

Contribution

It defines a novel $\sigma$-avoiding stack-sorting map for sock sequences and analyzes its sorting efficiency and pattern enumeration, especially for $\sigma=aba$.

Findings

The map sorts any sock sequence with n distinct socks in at most n iterations.

The number of 1-stack-sortable sock patterns under $\sigma=aba$ is explicitly enumerated and asymptotically analyzed.

For most patterns, the sorting map cannot sort all sequences unless they are already sorted.

Abstract

A sock sequence is a sequence of elements, which we will refer to as socks, from a finite alphabet. A sock sequence is sorted if all occurrences of a sock appear consecutively. We define equivalence classes of sock sequences called sock patterns, which are in bijection with set partitions. The notion of stack-sorting for set partitions was originally introduced by Defant and Kravitz. In this paper, we define a new deterministic stack-sorting map $\phi_{\sigma}$ for sock sequences that uses a $\sigma$-avoiding stack, where pattern containment need not be consecutive. When $\sigma = aba$, we show that our stack-sorting map sorts any sock sequence with $n$ distinct socks in at most $n$ iterations, and that this bound is tight for $n \geq 3$. We obtain a fine-grained enumeration of the number of sock patterns of length $n$ on $r$ distinct socks that are $1$-stack-sortable under $\phi_{aba}$, and we also obtain asymptotics for the number of sock patterns of length $n$ that are $1$-stack-sortable under $\phi_{aba}$. Finally, we show that for all unsorted sock patterns $\sigma \neq a\cdots a b a \cdots a$, the map $\phi_{\sigma}$ cannot eventually sort all sock sequences on any multiset $M$ unless every sock sequence on $M$ is already sorted.