\documentclass[11pt]{amsart} \usepackage[T1]{fontenc} \usepackage{lmodern,amsmath,amssymb,amsthm,mathtools} \usepackage[margin=1.05in]{geometry} \usepackage{enumitem,booktabs,needspace,array} \usepackage{tikz} \usepackage[colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue]{hyperref} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{conjecture}[theorem]{Conjecture} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{remark}[theorem]{Remark} \newtheorem{question}[theorem]{Question} \newcommand{\OP}{\mathcal O} \newcommand{\CP}{\mathcal C} \newcommand{\sd}{\mathbin{\triangle}} \newcommand{\Iso}{\mathrm{Iso}} \DeclareMathOperator{\diam}{diam} \setlist[enumerate]{leftmargin=*,label=\textup{(\roman*)}} \emergencystretch=2em \raggedbottom \hypersetup{pdftitle={The Lee (2026) conjecture on the diameters of order and chain polytopes fails in both directions, and other miscellanea},pdfauthor={Dave Goel}} \title[Diameters of order and chain polytopes]{The Lee (2026) conjecture on the diameters of order and chain polytopes fails in both directions, and other miscellanea} \author{Dave Goel} \subjclass[2020]{Primary 52B05; Secondary 06A07, 05C12} \keywords{order polytope, chain polytope, finite poset, graph of a polytope, diameter} \dedicatory{\normalfont\footnotesize Manuscript version 4.30, October 28, 2026.} \begin{document} \begin{abstract} For a finite poset $P$, Stanley's order polytope $\OP(P)$ and chain polytope $\CP(P)$ have the same number of vertices and, by a theorem of Hibi, Li, Sahara and Shikama, the same number of edges. J.~Lee recently conjectured that for a \emph{connected} poset the diameters of their graphs differ by at most $1$. We show that this fails in both directions. For every $k\ge1$ we describe a connected poset $B_k$ with $3k$ elements on three consecutive antidiagonals of the grid $\mathbb Z^2$, for which the graph of $\OP(B_k)$ has diameter exactly $k$, while the graph of $\CP(B_k)$ has diameter exactly $\lfloor2k/3\rfloor+1$. Their difference is therefore $\lceil k/3\rceil-1$, and their ratio tends to $3/2$. Hence no constant bounds $\diam\OP(P)-\diam\CP(P)$ on connected posets. In the other direction, for every $n\ge4$ we construct a connected poset $P_n$ with $2n^2+5n+1$ elements such that $\diam\OP(P_n)=2$ and $\diam\CP(P_n)=4$. We also show that $\diam\OP(P)\le2$ implies $\diam\CP(P)\le4$ for every finite poset $P$, so a difference of $2$ is the largest possible at order diameter $2$. \end{abstract} \maketitle \section{Introduction}\label{sec:intro} Let $P$ be a finite poset. An \emph{ideal} of $P$ is a subset that contains, with each of its elements, every smaller element; an \emph{antichain} is a set of pairwise incomparable elements. Stanley~\cite{Sta86} attached two polytopes to $P$: the \emph{order polytope} $\OP(P)$, whose vertices are the indicator vectors of the ideals\footnote{Stanley uses filters; passing to complements changes nothing in what follows.}, and the \emph{chain polytope} $\CP(P)$, whose vertices are the indicator vectors of the antichains. The two polytopes have the same number of vertices, and Hibi, Li, Sahara and Shikama~\cite{HLSS17} proved that their graphs have the same number of edges. Lee~\cite{Lee26} refined this edge theorem and compared further invariants of the two graphs, among them the diameter; we write $\diam$ for the diameter of the graph (the $1$-skeleton) of a polytope, and we call $\diam\OP(P)$ and $\diam\CP(P)$ the \emph{order diameter} and the \emph{chain diameter} of $P$. Both diameters are additive over disjoint unions of posets, since $\OP(P\sqcup Q)=\OP(P)\times\OP(Q)$, $\CP(P\sqcup Q)=\CP(P)\times\CP(Q)$ and the graph of a product of polytopes is the Cartesian product of their graphs. There are small connected posets on which the two diameters differ by $1$, in either direction \cite[Example~5]{Lee26}; so they can differ by any amount on posets with many components. For connected posets Lee reports a computer search that found no difference larger than $1$, and states the following conjecture, which we call the Lee (2026) conjecture. \begin{conjecture}[{\cite[Conjecture~29]{Lee26}}]\label{conj:lee} For every connected finite poset $P$, \[ |\diam\OP(P)-\diam\CP(P)|\le1. \] \end{conjecture} Versions~2 and~3 of~\cite{Lee26} withdraw the conjecture, and version~3 cites an earlier version of this paper. We show that the conjecture fails, and that it cannot be repaired by a larger constant. For example, the band $B_7$ of Definition~\ref{def:band}, with $21$ elements, has order diameter $7$ and chain diameter $5$, and the poset $P_4$ of Section~\ref{rv-sec:construction}, with $53$ elements, has order diameter $2$ and chain diameter $4$. The word \emph{connected} matters: for posets with many components a large difference is immediate from additivity. For a connected poset $P$ that is not a chain, both diameters lie between $2$ and the width $w(P)$ \cite[Corollary~25]{Lee26}, so they differ by at most $w(P)-2$. Our examples show that the difference can indeed grow linearly with the width. \needspace{6\baselineskip} \medskip\noindent\textbf{Results.} Our results are listed below, and Table~\ref{tab:results} compares them with the best earlier results. \begin{itemize}[leftmargin=*] \item The order diameter can exceed the chain diameter by any amount on connected posets: the bands $B_k$ of Theorem~\ref{thm:main}. This disproves Conjecture~\ref{conj:lee} in the direction $\diam\OP(P)-\diam\CP(P)\le1$, and no larger constant can replace~$1$; for disconnected posets the difference is unbounded, by additivity~\cite{Lee26}. \item The chain diameter can exceed the order diameter by $2$ on connected posets: the posets $P_n$ \mbox{($n\ge4$)} of Theorem~\ref{rv-thm:main}. This disproves the other half of Conjecture~\ref{conj:lee}, $\diam\CP(P)-\diam\OP(P)\le1$; the small examples of~\cite[Example~5]{Lee26} reach a difference of $1$ in either direction. \item Order diameter at most two forces chain diameter at most four, for every finite poset (Theorem~\ref{rv-thm:upper}). So $P_n$ $(n\ge4)$ has the largest possible difference among posets of order diameter two; unlike the bound of both diameters by the width~\cite[Corollary~25]{Lee26}, this bound does not depend on $w(P)$. \item The ratio $\diam\OP(B_k)/\diam\CP(B_k)$ tends to $3/2$ (Theorem~\ref{thm:main}); for a connected poset $P$ that is not a chain, the bounds of~\cite[Corollary~25]{Lee26} give only $\diam\OP(P)/\diam\CP(P)\le w(P)/2$. Whether ratios larger than $3/2$, or an unbounded difference $\diam\CP(P)-\diam\OP(P)$, occur on connected posets is open (Section~\ref{sec:questions}). \end{itemize} \medskip \begin{table}[ht] \centering\small \caption{Results of this paper and the best results before it.}\label{tab:results} \begin{tabular}{>{\raggedright\arraybackslash}p{0.47\linewidth}>{\raggedright\arraybackslash}p{0.43\linewidth}} \toprule New result & Best result before this paper\\ \midrule $\diam\OP(P)-\diam\CP(P)$ is unbounded on connected posets; $\diam\OP(B_k)=k$ and $\diam\CP(B_k)=\lfloor2k/3\rfloor+1$ (Theorem~\ref{thm:main}) & no difference above $1$ in Lee's computer search; at most $w(P)-2$ if $P$ is not a chain (by \cite[Corollary~25]{Lee26})\\\addlinespace $\diam\CP(P)-\diam\OP(P)=2$ for the connected posets $P=P_n$, $n\ge4$ (Theorem~\ref{rv-thm:main}) & difference $1$ \cite[Example~5]{Lee26}\\\addlinespace $\diam\OP(P)\le2$ implies $\diam\CP(P)\le4$ (Theorem~\ref{rv-thm:upper}) & $\diam\CP(P)\le w(P)$ \cite[Corollary~25]{Lee26}\\ \bottomrule \end{tabular} \end{table} \begin{definition}\label{def:band} For $k\ge1$ the \emph{band} $B_k$ is the poset on the $3k$ elements $a_i,b_i,c_i$ $(1\le i\le k)$ whose cover relations are \[ a_ij\}. \] By Lemma~\ref{lem:pairs} the only comparable pairs with one element in $L_j$ and one in $R_j$ are $a_{j-1}