\documentclass[11pt]{article} \usepackage[margin=1in]{geometry} \usepackage[utf8]{inputenc} \usepackage[T1]{fontenc} \usepackage{lmodern} \usepackage{amsmath,amssymb,amsthm} \usepackage{booktabs} \usepackage[title]{appendix} \usepackage[colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pdfauthor={P. Graham},pdftitle={A four-dimensional 4-reptile simplex}]{hyperref} \newtheorem{theorem}{Theorem} \newtheorem{corollary}[theorem]{Corollary} \newtheorem{lemma}[theorem]{Lemma} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} \setlength{\emergencystretch}{3em} \title{A four-dimensional 4-reptile simplex} \author{P. Graham\\ \small\url{https://orcid.org/0009-0007-8493-1505}} \date{First published November 8, 2026; this version November 11, 2026 (v2)\\[2pt] \small Zenodo archive (\texttt{checks}, \texttt{data}): \url{https://doi.org/10.5281/zenodo.24128596}} \begin{document} \maketitle \begin{abstract} A $d$-dimensional simplex is a $k$-reptile if it can be tiled by $k$ mutually congruent simplices similar to it. Matoušek and Safernová conjectured that for $d\ge3$ a $k$-reptile $d$-simplex exists only when $k=m^d$ for an integer $m$, and proved this for $d=3$. Kynčl and Patáková proved that in dimension four $k$ must be a square, noted that every $k$-reptile simplex known in dimension $d\ge3$ has $k=m^d$, and asked whether a four-dimensional $k$-reptile simplex exists with $k$ a square but not a fourth power. We show that the simplex with vertices $(0,0,0,0)$, $(1,0,0,0)$, $(\tfrac12,\tfrac12,0,0)$, $(\tfrac23,\tfrac13,\tfrac13,0)$ and $(\tfrac34,\tfrac14,\tfrac14,\tfrac14)$, the fundamental alcove of the affine Weyl group of type $F_4$, is a $4$-reptile: it is tiled by four congruent copies of itself scaled by $1/\sqrt2$, two of them mirror images. This answers the question of Kynčl and Patáková with $k=4$ and disproves the conjecture of Matoušek and Safernová in dimension four. The subdivision itself is due to Felikson and Tumarkin, who found in 2004 that the affine reflection group of type $F_4$ has a reflection subgroup of index $4$ whose fundamental simplex is similar to the alcove. What we add is the observation that this settles the reptile question, a short direct proof of the tiling, and exact computer checks of it. This work has not yet been reviewed by experts. The proofs, the computer checks and the text were produced by AI models (from OpenAI and Anthropic), working under the author's direction. \end{abstract} \medskip \noindent\textit{Keywords:} reptile simplex, self-similar tiling, alcove, affine Weyl group, $F_4$. \noindent\textit{MSC 2020:} 52C22, 05B45, 20F55. \section{Introduction} \subsection{Main result} A tiling of a closed set $X\subset\mathbb{R}^d$ is a decomposition of $X$ into finitely many closed sets with nonempty, pairwise disjoint interiors. Following Kynčl and Patáková \cite{KP}, a $d$-dimensional simplex $S$ is a \emph{$k$-reptile} if it can be tiled by $k$ simplices that are all mutually congruent and similar to $S$; as they emphasize, ``congruence includes mirror symmetries'' \cite[p.~1]{KP}. Throughout, let \begin{gather*} S=\operatorname{conv}\{v_0,v_1,v_2,v_3,v_4\},\qquad\text{where}\\ v_0=(0,0,0,0),\quad v_1=(1,0,0,0),\quad v_2=(\tfrac12,\tfrac12,0,0),\quad v_3=(\tfrac23,\tfrac13,\tfrac13,0),\quad v_4=(\tfrac34,\tfrac14,\tfrac14,\tfrac14). \end{gather*} \begin{theorem}\label{thm:main} The simplex $S\subset\mathbb{R}^4$ is a $4$-reptile. More precisely, $S$ is the union of four simplices $S_0,\dots,S_3$ with pairwise disjoint interiors, each the image of $S$ under a similarity of ratio $1/\sqrt2$. Two of the four similarities preserve orientation and two reverse it. \end{theorem} Since $4=2^2$ and $4$ is not a fourth power, Theorem~\ref{thm:main} answers the question of Kynčl and Patáková stated below, with $k=4$, and gives a counterexample, in dimension $d=4$, to the conjecture of Matoušek and Safernová. The simplex $S$ is the fundamental alcove of the affine Weyl group of type $F_4$ (Section~\ref{sec:origin}). A linear map $\varphi$ that is $\sqrt2$ times an isometry carries the hyperplane arrangement of this group into itself, so $\varphi(S)$ is a union of alcoves, $4$ of them by volume. Section~\ref{sec:proof} proves Theorem~\ref{thm:main} directly, by writing $\varphi(S)$ and its four pieces as solution sets of linear inequalities; the proof needs no Lie theory and no computer. Section~\ref{sec:verification} describes exact computer checks of the same tiling. \subsection{The question} Hertel \cite[pp.~650--652]{He} proved that the simplices he calls H-simplices, and more generally his H-simplotopes, are $m^d$-replicating (tiled by $m^d$ congruent copies of themselves scaled down) for every integer $m\ge2$. His Conjecture~2 \cite[p.~652]{He} asserts that, for $d>2$, an $f$-replicating $d$-simplex must have $f=m^d$ for an integer $m\ge2$ and must be an H-simplex. Theorem~\ref{thm:main} also refutes this conjecture in dimension four, since $4$ is not a fourth power. Matoušek and Safernová \cite[Conjecture~1.2]{MS} conjectured: ``If $d\ge3$ and a $d$-dimensional $k$-reptile simplex exists, then $k=m^d$ for some natural number $m\ge2$.'' They proved it for $d=3$; Kynčl and Patáková gave a simpler proof \cite[Theorem~1.2]{KP} and proved that a four-dimensional $k$-reptile simplex can exist only when $k$ is the square of a positive integer \cite[Theorem~1.3]{KP}. In their abstract they note that ``the only $k$-reptile simplices that are known for $d\geqslant3$, have $k=m^d$''. They then ask \cite[Problem~1.4; Problem~2 in the arXiv version]{KP} whether there is a four-dimensional $k$-reptile simplex with $k$ the square of a non-square integer, that is, with $k$ a square but not a fourth power. Theorem~\ref{thm:main} answers this with $k=4$, the square of $2$. It is consistent with \cite[Theorem~1.3]{KP}, since $k=4$ is a square, and it leaves the case $d=3$ untouched. \subsection{Credit} The tiling is not new as geometry. Felikson and Tumarkin \cite[Lemma~10]{FT} consider Coxeter simplices $F\ne P$ in $\mathbb{E}^n$ with the same Coxeter diagram, $\Sigma(F)=\Sigma(P)$, and with $G_P\subset G_F$ for the groups generated by the reflections in their facets. They prove that $1<[G_F:G_P]<2^n$ holds exactly in three cases, the last being ``$\Sigma(F)=\Sigma(P)=\widetilde F_4$, $[G_F:G_P]=4$''. Equal Coxeter diagrams make $P$ similar to $F$. A fundamental simplex of an index-$4$ reflection subgroup of $G_F$ is a union of $4$ fundamental simplices of $G_F$, which are congruent. So their $\widetilde F_4$ case is a simplex similar to the $F_4$ alcove and tiled by four alcoves: in substance, Theorem~\ref{thm:main}. Their paper does not discuss reptiles, and we found no reference connecting it to the reptile question. Our contribution is that connection, the explicit tiling with a direct proof, and its checks. The other two cases of their lemma, of types $\widetilde C_2$ and $\widetilde G_2$, give the classical $2$-reptile isosceles right triangle and $3$-reptile $30$-$60$-$90$ triangle. \section{Proof of Theorem~\ref{thm:main}}\label{sec:proof} We describe $S$ by inequalities. For $x=(x_1,x_2,x_3,x_4)$ let \[ \ell_0(x)=1-x_1-x_2,\quad \ell_1(x)=x_2-x_3,\quad \ell_2(x)=x_3-x_4,\quad \ell_3(x)=x_4,\quad \ell_4(x)=x_1-x_2-x_3-x_4 . \] \begin{lemma}\label{lem:facets} $S=\{x\in\mathbb{R}^4:\ \ell_i(x)\ge0 \text{ for } i=0,\dots,4\}$, that is, $S=\{x:\ x_2\ge x_3\ge x_4\ge0,\ x_1\ge x_2+x_3+x_4,\ x_1+x_2\le1\}$. \end{lemma} \begin{proof} Each vertex of $S$ makes exactly one of the five functions nonzero, and that one positive: $\ell_0(v_0)=1$, $\ell_4(v_1)=1$, $\ell_1(v_2)=\tfrac12$, $\ell_2(v_3)=\tfrac13$ and $\ell_3(v_4)=\tfrac14$, and every other value $\ell_i(v_j)$ is $0$. The points $v_j$ are affinely independent, since $v_1,\dots,v_4$ are linearly independent. Let $\lambda_0,\dots,\lambda_4$ be the barycentric coordinates with respect to $v_0,\dots,v_4$. For each vertex there is one function, say $\ell_{\sigma(j)}$ for $v_j$, that is nonzero on it; the affine functions $\ell_{\sigma(j)}/\ell_{\sigma(j)}(v_j)$ and $\lambda_j$ agree on all five vertices, hence everywhere. So $S=\{\lambda_j\ge0\ \forall j\}=\{\ell_i\ge0\ \forall i\}$. \end{proof} Let $\varphi(x)=(x_1+x_2,\ x_1-x_2,\ x_3+x_4,\ x_3-x_4)$. Its matrix $\Phi$ satisfies $\Phi^{\mathsf T}\Phi=2I$, so $\varphi/\sqrt2$ is an orthogonal map and $T=\varphi(S)$ is similar to $S$ with ratio $\sqrt2$; also $\det\Phi=4$. Substituting $x_1=\frac{y_1+y_2}2$, $x_2=\frac{y_1-y_2}2$, $x_3=\frac{y_3+y_4}2$, $x_4=\frac{y_3-y_4}2$ into Lemma~\ref{lem:facets} gives \begin{equation}\label{eq:T} T=\{y\in\mathbb{R}^4:\ y_1\le1,\ y_1\ge y_2+y_3+y_4,\ y_2\ge y_3\ge y_4\ge0\}. \end{equation} Its vertices are $\varphi(v_0)=(0,0,0,0)$, $\varphi(v_1)=(1,1,0,0)$, $\varphi(v_2)=(1,0,0,0)$, $\varphi(v_3)=(1,\tfrac13,\tfrac13,\tfrac13)$ and $\varphi(v_4)=(1,\tfrac12,\tfrac12,0)$. Note that $S\subset T$ by Lemma~\ref{lem:facets} and~\eqref{eq:T}. Define four isometries of $\mathbb{R}^4$: \begin{align*} g_0(x)&=x, & g_1(x)&=(1-x_2,\,1-x_1,\,x_3,\,x_4),\\ g_2(x)&=(1-x_3,\,1-x_1,\,x_2,\,x_4), & g_3(x)&=(1-x_4,\,1-x_1,\,x_2,\,x_3). \end{align*} Each is a signed permutation of coordinates followed by a translation. The linear parts of $g_0,g_1,g_2,g_3$ have determinants $1,-1,1,-1$. \begin{lemma}\label{lem:pieces} Let $P_i=g_i(S)$. Then \begin{align*} P_0&=T\cap\{y:\ y_1+y_2\le1\}, & P_1&=T\cap\{y:\ y_1+y_3\le1\le y_1+y_2\},\\ P_2&=T\cap\{y:\ y_1+y_4\le1\le y_1+y_3\}, & P_3&=T\cap\{y:\ 1\le y_1+y_4\}. \end{align*} \end{lemma} \begin{proof} Since $g_i$ is a bijection, $y\in P_i$ exactly when $x=g_i^{-1}(y)$ satisfies the five inequalities of Lemma~\ref{lem:facets}. We write them out in $y$ for each $i$. For $i=0$: $y_2\ge y_3\ge y_4\ge0$, $y_1\ge y_2+y_3+y_4$ and $y_1+y_2\le1$. The remaining inequality $y_1\le1$ of~\eqref{eq:T} follows from $y_1\le1-y_2\le1$. For $i=1$, put $x=(1-y_2,\,1-y_1,\,y_3,\,y_4)$. The five inequalities become $y_1+y_3\le1$, $y_3\ge y_4$, $y_4\ge0$, $y_1\ge y_2+y_3+y_4$ and $y_1+y_2\ge1$. The inequalities $y_1\le1$ and $y_2\ge y_3$ of~\eqref{eq:T} follow: $y_1\le1-y_3\le1$, and $y_1+y_2\ge1\ge y_1+y_3$. For $i=2$, put $x=(1-y_2,\,y_3,\,1-y_1,\,y_4)$. The inequalities become $y_1+y_3\ge1$, $y_1+y_4\le1$, $y_4\ge0$, $y_1\ge y_2+y_3+y_4$ and $y_2\ge y_3$. The inequalities $y_3\ge y_4$ and $y_1\le1$ follow from $y_1+y_3\ge1\ge y_1+y_4$ and $y_1\le1-y_4\le1$. For $i=3$, put $x=(1-y_2,\,y_3,\,y_4,\,1-y_1)$. The inequalities become $y_3\ge y_4$, $y_1+y_4\ge1$, $y_1\le1$, $y_1\ge y_2+y_3+y_4$ and $y_2\ge y_3$. The inequality $y_4\ge0$ follows from $y_4\ge1-y_1\ge0$. In each case the inequalities obtained, together with the ones shown to follow, are exactly those of~\eqref{eq:T} and of the stated region. \end{proof} \begin{proof}[Proof of Theorem~\ref{thm:main}] On $T$ we have $y_2\ge y_3\ge y_4$, so $y_1+y_2\ge y_1+y_3\ge y_1+y_4$. For $y\in T$, the number $1$ lies above $y_1+y_2$, between two consecutive terms of this chain, or below $y_1+y_4$, so $y$ lies in at least one of the four regions of Lemma~\ref{lem:pieces}. Hence $T=P_0\cup P_1\cup P_2\cup P_3$. Within $T$, two different regions intersect only inside one of the hyperplanes $y_1+y_j=1$ ($j=2,3,4$), so the $P_i$ have pairwise disjoint interiors. Each $P_i=g_i(S)$ is congruent to $S$. Apply $\varphi^{-1}$. The sets $S_i=\varphi^{-1}(P_i)$ tile $\varphi^{-1}(T)=S$, and $S_i$ is the image of $S$ under the similarity $\varphi^{-1}\circ g_i$ of ratio $1/\sqrt2$. These four similarities differ from one another by isometries, so the $S_i$ are mutually congruent. Since $\det\Phi>0$, the similarity $\varphi^{-1}\circ g_i$ preserves orientation exactly when $g_i$ does, that is, for $i=0,2$. \end{proof} The tiling is a chain: consecutive pieces share a facet. The vertices of the pieces are listed in Appendix~\ref{app:vertices}. \begin{remark}\label{rem:chiral} The mirror images cannot be avoided by relabeling: $S$ has no nontrivial symmetry, since its ten edge lengths admit no nonidentity permutation of the vertices preserving them. Whether some other tiling of $S$ by four copies uses only directly congruent ones is not settled here. \end{remark} \section{Where the tiling comes from}\label{sec:origin} Take the root system of type $F_4$ in Bourbaki's coordinates \cite[Plate~VIII]{Bou}: its $48$ roots are $\pm e_i$, $\pm e_i\pm e_j$ ($i