\documentclass[11pt]{article} \usepackage[T1]{fontenc} \usepackage{lmodern,microtype,amsmath,amssymb,amsthm,mathtools,booktabs} \usepackage[margin=1.05in]{geometry} \usepackage{graphicx,tikz,placeins,needspace,float} \usepackage[font=small,labelfont=bf]{caption} \setlength{\intextsep}{9pt plus 2pt minus 2pt} \setlength{\textfloatsep}{13pt plus 2pt minus 2pt} \input{figures/box-and-labels} \usepackage[colorlinks=true,linkcolor=blue!40!black,citecolor=blue!40!black, urlcolor=blue!40!black]{hyperref} \hypersetup{pdftitle={Boxes with cyclic tiling groups have one layer or size 2×2×4}, pdfauthor={twinheterodox}} \newtheorem{theorem}{Theorem} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{corollary}[theorem]{Corollary} \newcommand{\Z}{\mathbb Z} \newcommand{\Q}{\mathbb Q} \newcommand{\F}{\mathbb F} \newcommand{\ord}{\operatorname{ord}} \newcommand{\coker}{\operatorname{coker}} \newcommand{\laur}{\Q[q,q^{-1}]} \newcommand{\pos}[1]{(#1)_+} \title{\Large\bfseries Boxes with cyclic tiling groups have one layer or size \(2\times2\times4\)} \author{twinheterodox\\[2pt]\normalsize Prepared with Codex} \date{19 November 2026} \makeatletter \renewcommand{\@maketitle}{% \begin{center} {\@title\par}\vspace{.6em} {\large\begin{tabular}{c}\@author\end{tabular}\par}\vspace{.45em} {\large\@date\par} \end{center}} \makeatother \begin{document} \maketitle \begin{abstract} Kuperberg asked how a box's dimensions determine its tiling group. We prove Conlon's conjecture that this group is cyclic exactly when one side has length one or the box has dimensions \(2\times2\times4\) in some order. For every box, we also show that the minimum number of generators of its tiling group equals the largest multiplicity of a cyclotomic factor of prime-power index in MacMahon's volume polynomial. This polynomial counts stacks by the number of cubes they contain. It also determines the Smith factors over \(\Q[q,q^{-1}]\) of the box's weighted Carlitz matrix, which presents the tiling group at \(q=1\). Each cyclotomic factor of multiplicity \(e\) occurs in exactly the last \(e\) Smith factors, once in each. \end{abstract} \section{Which boxes have a cyclic tiling group?} Stack unit cubes in a corner of an \(a\times b\times c\) box so that column heights weakly decrease away from that corner in both horizontal directions. This \emph{boxed plane partition} projects to a hexagon tiled by diamonds called \emph{lozenges} (Figure~\ref{fig:stack}). The matrix \(K\) of possible tile placements presents a finite abelian group \(G(a,b,c)\), the \emph{tiling group}, with one generator for each row and one relation for each column (Figure~\ref{fig:unit-placement}). Its order is \(\lvert\det K\rvert\), the number of possible stacks in that box~\cite[p.~265]{Propp}. In 1998 Kuperberg asked which group of that order occurs for each box~\cite[Question~2]{Kuperberg1998}. \begin{theorem}[Conlon's conjecture on cyclic tiling groups]\label{thm:cyclic} For positive integers \(a,b,c\), the group \(G(a,b,c)\) is cyclic if and only if some side has length one or \((a,b,c)\) is a permutation of \((2,2,4)\). The exceptional group \(G(2,2,4)\) is isomorphic to \(\Z/105\Z\). \end{theorem} Theorem~\ref{thm:cyclic} tells us when one generator suffices. How many generators does an arbitrary box need? Give each stack of \(v\) cubes weight \(q^v\) and sum over all stacks to obtain MacMahon's volume polynomial \(M(q)\)~\cite[\S429]{MacMahon}. Write \(e(d)\) for the multiplicity in \(M(q)\) of \(\Phi_d(q)\), the cyclotomic polynomial whose roots are the primitive \(d\)th roots of unity. \begin{theorem}[Minimum number of generators]\label{thm:generators} The tiling group \(G(a,b,c)\) needs exactly \(\max_{p,j} e(p^j)\) generators, where \(p\) runs over the primes and \(j\) over the positive integers. \end{theorem} \enlargethispage{1.1\baselineskip} \begin{figure}[H] \centering\StackAndTiling \caption{One of the 105 stacks in the \(2\times2\times4\) box. Its four columns have heights \(4,2,1,0\). Viewed along a body diagonal, each visible unit square, whether the face of a cube or part of the box's floor or walls, projects to a lozenge. Together these lozenges tile the hexagon on the right, with the floor and walls shown in pale colors. The dotted segment in the bottom lozenge is its short diagonal. It determines that tile uniquely because the two unit equilateral triangles with this common side form the lozenge.} \label{fig:stack} \end{figure} Kuperberg observed that the tiling group of a one-layer box is cyclic~\cite[Question~2]{Kuperberg1998}. In 2002 he proposed two conjectures that together would determine the group for every box~\cite[\S5.1, Conjectures~11--12]{Kuperberg}. Conlon found the exceptional \(2\times2\times4\) box by computer and conjectured that it was the only exceptional case~\cite[\S7]{Conlon}. MacMahon's count of the stacks~\cite[\S429]{MacMahon} gives the group order \(105=3\cdot5\cdot7\), so by Cauchy's theorem there are elements of orders \(3\), \(5\), and \(7\). Since these elements commute and their orders are pairwise coprime, their product has order \(105\) and generates the group. Section~\ref{sec:cyclic-proof} rules out the remaining boxes by a gcd calculation when the shortest side is two and by a bound on element orders when it is larger. Section~\ref{sec:volume} illustrates Theorem~\ref{thm:generators} on two boxes with twenty stacks, and Section~\ref{sec:ordinary} proves it. \subsection*{From the picture to the group} The three orientations of a lozenge correspond to squares parallel to the floor or to either wall. Projection gives a bijection between stacks in an \(a\times b\times c\) box and lozenge tilings of the equiangular hexagon with successive side lengths \(a,b,c,a,b,c\)% ~\cite[\S5.1, Fig.~10]{Kuperberg}. Marking the short diagonal of every lozenge records the whole tiling. To list the possible marks, divide the hexagon into unit triangles. In the orientation of Figures~\ref{fig:stack} and~\ref{fig:unit-placement}, the right-pointing triangles (\(\triangleright\)) index the rows of a matrix \(K\) and the left-pointing ones (\(\triangleleft\)) its columns. Put \(1\) for a shared side, the short diagonal of a possible tile, and \(0\) otherwise. A choice gives a tiling exactly when its tiles cover each triangle once, selecting one \(1\) per row and column~\cite[p.~261]{Propp}. The Kasteleyn--Percus formula counts these tilings by \(|\det K|\)~\cite[p.~265]{Propp}. The generators of \(G(a,b,c)\) are indexed by the rows of \(K\). For each column, set the sum of the generators at its \(1\)'s equal to zero. In the one-cube box (Figure~\ref{fig:unit-placement}), column \(A\) gives \(y=-x\), and then column \(B\) gives \(z=-y=x\). Substituting in column \(C\) leaves one generator \(x\) with the single relation \(2x=0\). With \(m\) triangles of each orientation, this construction gives the quotient \(G=\Z^m/K\Z^m\), the integer cokernel of \(K\). Its order is the lattice index \(|\det K|\). Kuperberg also asked how its elements could be interpreted in terms of the tilings~\cite[p.~265, Problem~5]{Propp}. \begin{figure}[!htbp] \centering\includegraphics[width=\linewidth]{figures/one-cube-placement-matrix.pdf} \caption{The marked diagonals of each tiling select one \(1\) in every row and column of \(K\). Triangle \(A\) shares a lozenge with \(x\) in the empty box and with \(y\) in the full box. In column \(A\) these two lozenges give a circled \(1\) in row \(x\) and a boxed \(1\) in row \(y\). Adding those row labels gives the group relation \(x+y=0\).} \label{fig:unit-placement} \end{figure} \enlargethispage*{\baselineskip} \subsection*{Does the stack count determine cyclicity?} Both boxes in Figure~\ref{fig:twenty} have twenty stacks, and only the flat box's tiling group is cyclic. Each colored line traces the boundary of a layer halfway up its cubes, continuing along the box walls where needed. Adding a unit cube interchanges two consecutive steps in its layer's path and leaves its endpoints fixed. The \(\binom63=20\) shortest paths between opposite floor corners give all the flat stacks. In the pictured \(2\times2\times2\) box, three unit cubes form an L and a fourth sits above its corner. Figure~\ref{fig:volumes} shows all twenty stacks in this box. Kuperberg's path matrix presents the same group as the triangle matrix \(K\)~\cite[Corollary~7 and Theorem~14]{Kuperberg}. Its rows index path starts (generators) and its columns index path ends (relations). Each entry counts paths between its endpoints by rising or falling steps to the right. The flat box gives \(20x=0\). In the two-layer box, two of four steps rise for a level finish (six choices), and one or three rise for a change of height (four choices). The upper path start gives \(x\) and the lower gives \(y\). In \(mx+ny\), both relations preserve the parities of \(m,n\) because their coefficients are even. All four pairs occur at \(0,x,y,x+y\). Multiples of any \(g\) give only the parities of \(0\) or \(g\), so two generators are necessary. \begin{figure}[!htbp] \centering\includegraphics[width=\linewidth]{figures/equal-count-boxes.pdf} \caption{Both boxes have twenty stacks. \(G(3,3,1)\) needs one generator and \(G(2,2,2)\) needs two. The colored paths trace the layers on the dotted grid. The lower right records the coefficient parities in \(mx+ny\), so \(2x+y\) gives (even, odd). Adding \(x\) changes only the parity of \(m\), and adding \(y\) only that of \(n\).} \label{fig:twenty} \end{figure} \FloatBarrier \section{What the volume polynomial remembers}\label{sec:volume} A cyclic group contains an element of order \(|G|\). To rule this out for every box with shortest side at least three, we bound the orders of its elements using the volume polynomial. Figure~\ref{fig:volumes} sorts the twenty stacks in the \(2\times2\times2\) box by volume. A rule that always adds or removes one unit cube sends all twelve even-volume stacks into the eight odd-volume stacks. Such a rule cannot realize addition by a group element, which is a bijection. \begin{figure}[htbp] \centering\includegraphics[width=.97\linewidth]{figures/volume-stacks.pdf} \caption{The twenty stacks in the \(2\times2\times2\) box, arranged by number of cubes. The darker faces belong to the cubes and the pale faces belong to the box. Setting \(q=1\) counts all twenty stacks.} \label{fig:volumes} \end{figure} \FloatBarrier The four stacks in the middle column contribute \(4q^4\). Reading across the figure gives the volume polynomial \[ M(q)=1+q+3q^2+3q^3+4q^4+3q^5+3q^6+q^7+q^8=A^2B, \] where \(A=1+q^2=\Phi_4(q)\) and \(B=1+q+q^2+q^3+q^4=\Phi_5(q)\). The factor \(\Phi_4\) has prime-power index \(4=2^2\) and occurs twice, so by Theorem~\ref{thm:generators} the group needs exactly two generators. For the flat \(3\times3\times1\) box, MacMahon's product gives \(M(q)=\Phi_2(q)\Phi_4(q)\Phi_5(q)\Phi_6(q)\). Each factor of prime-power index occurs once, so one generator suffices. Both polynomials give \(20\) at \(q=1\). \Needspace{6\baselineskip} The map \(mx+ny\mapsto(m\bmod2,(m+n)\bmod10)\) respects both relations for the \(2\times2\times2\) box in Figure~\ref{fig:twenty} and takes all twenty values. Since the group has order twenty, this gives \(G(2,2,2)\cong\Z/2\Z\oplus\Z/10\Z\). To see these summand orders in the volume polynomial, number the four steps of each path in Figure~\ref{fig:twenty} from \(0\) to \(3\). Give a rise weight \(1\) and a fall at position \(r\) weight \(q^r\), and multiply the weights along the path. Choosing \(q^r u\) in \(\prod_{r=0}^3(1+q^r u)\) records a fall at \(r\), so the coefficient \(e_k\) of \(u^k\) counts paths with \(k\) falls by weight. From the upper start \(x\) to the lower endpoint \(B\) there are three falls, giving \(e_3\) in the upper-right entry. At \(q=1\) every path has weight one. \[ C(q)=\begin{pmatrix}e_2&e_3\\e_1&e_2\end{pmatrix}, \qquad C(1)=\begin{pmatrix}6&4\\4&6\end{pmatrix}. \] The determinant of \(C(q)\) is \(M(q)\) times a power of \(q\). Allow rational coefficients and powers of \(q\) of either sign. Invertible row and column operations then give a diagonal matrix with each entry dividing the next. These entries are the \emph{Smith factors}, determined up to nonzero rational constants and powers of \(q\). The first Smith factor divides the second, so its square divides \(A^2B\). Since \(A\) and \(B\) are distinct irreducibles, the only possibilities are \((1,A^2B)\) and \((A,AB)\). At \(q=i\), where \(A=0\), the first has rank one and the second has rank zero. The row and column operations remain invertible at \(i\), so we can read the rank from \(C(i)\). At this value the product defining the entries becomes \((1+u)(1+iu)(1-u)(1-iu)=1-u^4\). All three middle coefficients vanish, so \(C(i)=0\). This selects \((A,AB)\), giving the summand orders \(2\) and \(10\) at \(q=1\). \Needspace{7\baselineskip} \subsection*{The rule for every box} For the general box, fix the lengths \(a,b,c\), put \(N=b+c\), and let \(e_k\) be the coefficient of \(x^k\) in \(\prod_{j=0}^{N-1}(1+q^j x)\), with \(e_k=0\) outside \(0\le k\le N\). The weighted Carlitz matrix and MacMahon's volume formula \cite[\S429]{MacMahon} are \[ C=(e_{b-i+j})_{0\le i,jN\) and \(v=0\). We have found at least \(e(d)\) dimensions of kernel. In Smith form the nullity at \(\zeta\) counts the invariant factors divisible by \(\Phi_d\). Their valuations sum to \(e(d)\) by \eqref{eq:det}, so there are exactly \(e(d)\) of them, each of valuation one. Divisibility puts them in the last \(e(d)\) positions. All nonunit irreducibles of the determinant are cyclotomic, which proves Theorem~\ref{thm:smith}. There are exactly \(\min(a,b,c)\) nonunit rational Laurent factors. For \(a\le b\le c\), at a primitive \((b+c)\)-th root the product defining the elementary functions has only its constant and top coefficients, and every matrix entry is intermediate. Thus the entire matrix vanishes. Permuting the box sides preserves \(M\), so the number \(\max_d e(d)\) has the same value in every orientation. This also gives the count proposed in \cite[\S9]{Conlon} for an integral Laurent Smith form if it exists, since primitivity of the determinant precludes additional nonunit constant factors. \section{Consequences for the integer matrix}\label{sec:integer} Let \(G=\coker C(1)\), the quotient of \(\Z^a\) by the image of the integer Carlitz matrix. Its order is \(M(1)\). For \(0\le k\le a\) put \[ D_k(q)=\prod_{d\ge2}\Phi_d(q)^{(e(d)-a+k)_+}. \] Over \(\laur\) the product of the first \(k\) Smith factors divides every \(k\)-rowed minor. That product is \(D_k(q)\). Its coprimality to \(q\) lets us remove powers of \(q\) over \(\Q[q]\). Division by the monic polynomial \(D_k(q)\) then proves divisibility of the actual minors in \(\Z[q]\). This makes specialization legitimate even if rational row operations would introduce denominators. \begin{corollary}\label{cor:int} Every \(k\)-rowed minor of \(C(1)\) is divisible by \(D_k(1)\). In particular, the exponent of \(G\) divides \[ \prod_{\substack{d\ge2\\e(d)>0}}\Phi_d(1) \quad\text{which divides}\quad \operatorname{lcm}(1,\ldots,a+b+c-1). \] \end{corollary} \begin{proof} The exponent of \(G\), the largest order of one of its elements, is its last integer Smith factor. This is \(M(1)\) divided by the gcd of the cofactors, with gcd 1 in size one. For the \(2\times2\times2\) box the diagonal cofactors are \(6\) and the off-diagonal cofactors are \(-4\). Their gcd is \(2\), so the largest element order is \(20/2=10\). In general \(D_{a-1}(1)\) divides the cofactor gcd. Dividing \(M(1)\) by \(D_{a-1}(1)\) gives the stated product. All indices \(d\) occurring are at most \(\max E=a+b+c-1\). For \(d\ge2\), \(\Phi_d(1)=p\) if \(d\) is a positive power of a prime \(p\), and is 1 otherwise. This gives both the lcm bound and \(\ord_p |G|=\sum_{j\ge1}e(p^j)\). \end{proof} \section{Proof of the cyclicity classification}\label{sec:cyclic-proof} \begin{proof}[Proof of Theorem~\ref{thm:cyclic}] Order the sides \(a\le b\le c\). Permuting them gives stably equivalent presentations of the same tiling group~\cite[Problem~5]{Propp}. If \(a=1\), the Carlitz matrix has a single entry and the group is cyclic. If \(a=2\), the group is cyclic exactly when the entries of the two-by-two Carlitz matrix have gcd \(1\), since their gcd is its first Smith factor. Put \(U=\binom{b+c}{b}\). The diagonal entries are \(U\), and the other two are \(Ub/(c+1)\) and \(Uc/(b+1)\). Their gcd is at least \(U/((b+1)(c+1))\), because multiplication by \((b+1)(c+1)\) makes all three entries integer multiples of \(U\). The lower bound increases with either \(b\) or \(c\) for \(b,c\ge2\) and equals \(20/16\) at \(b=c=3\). For \(b=2\) it is \((c+2)/6>1\) whenever \(c\ge5\). Finally at \(b=2,\ c=2,3,4\) the gcds are respectively \(2,5,1\). Only the last box gives a cyclic group. Its matrix \(\left(\begin{smallmatrix}15&20\\6&15\end{smallmatrix}\right)\) has determinant \(105\). The map \((x,y)\mapsto5y-9x\bmod105\) sends both columns to zero and the class of \((1,2)\) to \(1\), giving an explicit isomorphism with \(\Z/105\Z\). Suppose now that \(a\ge3\). A finite abelian group is cyclic exactly when its exponent equals its order. We show that the order outgrows an upper bound for the exponent. Write \(t=a+b-1\) and \(L(t)=\operatorname{lcm}(1,\ldots,t)\). Each active power \(p^j\) divides a numerator index in the symmetric version of MacMahon's formula having height \(c\). These indices range from \(c+1\) to \(c+t\). The number of active levels is at most the largest active \(j\), so Corollary~\ref{cor:int} gives \[ \exp(G)\mid\operatorname{lcm}(c+1,\ldots,c+t) \mid L(t)\binom{c+t}{t}. \] For the second divisibility, every power of a prime beyond \(t\) that divides some \(c+i\) contributes to the valuation of the binomial, and the smaller powers are covered by \(L(t)\). In the product for \(|G|\), the cells of the first row and last column together give one factor \(1+c/k\) for every \(1\le k\le t\). Thus \[ F(a,b,c)=\frac{|G|}{\binom{c+t}{t}} =\prod_{i=2}^{a}\prod_{j=1}^{b-1}\left(1+\frac{c}{i+j-1}\right). \] It suffices that \(F(a,b,c)>L(t)\), and \(F\) increases with \(c\). For \(b\ge7\) it suffices to compare \(F(a,b,b)\) with the upper bound \(4^{t-1}\) for \(L(t)\). Start at \((3,7,7)\), increase \(b\) along \((3,b,b)\), and then increase \(a\) at fixed \(b\). Each step adds one to \(t=a+b-1\), multiplying the comparison bound by four. To prove \(L(n)\le4^{n-1}\) by induction, observe that \(L(2m)/L(m)\) divides \(\binom{2m}{m}\), and \(L(2m+1)/L(m+1)\) divides \(\binom{2m+1}{m}\). Any prime power missing from the smaller lcm gives a factor in the binomial, and at most one power per prime is missing. Both binomials are at most \(4^m\), using the two equal central coefficients in the odd row. For \(F_b=F(3,b,b)\), multiplying its two rows gives \[ F_b=\frac{2(2b+1)\binom{2b}{b}^2}{(b+1)^3(b+2)},\qquad \frac{F_{b+1}}{F_b}= \frac{4(2b+3)(2b+1)(b+1)}{(b+2)^2(b+3)}>4 \quad(b\ge2). \] The last inequality reduces to \(3b^3+5b^2-5b-9>0\). Since \(F_7=306735/4>4^8\), we have \(F_b>4^{b+1}\). Each new row up to \(a\le b\) increases \(F(a,b,b)\) by a factor of at least \((3/2)^{b-1}>4\). Therefore \(F(a,b,c)>4^{a+b-2}\ge L(t)\). For \(4\le b\le6\) the first pair of columns below covers \(a=3\). The second covers every \(4\le a\le b\), by monotonicity of \(F\) and \(L(t)\le L(2b-1)\). \[ \begin{array}{c|rr|rr} b &F(3,b,b)&L(b+2)&F(4,b,b)&L(2b-1)\\ \hline 4&588/5&60&3528/5&420\\ 5&924&420&13068&2520\\ 6&56628/7&840&2024451/7&27720 \end{array} \] For \(a=b=3\) and \(c\ge6\), \(F(3,3,6)=90>60=L(5)\). For \(c=3,4,5\) the respective group orders are \(980,4116,14112\), while \(\operatorname{lcm}(c+1,\ldots,c+5)\) is \(840,2520,2520\). This proves noncyclicity for every box with minimum side at least three and completes the classification. \end{proof} Another prediction in Conlon's \S7 concerns the number of generators for an \(n\times n\times n\) box. It requires \(n\) generators exactly when \(n\) is a power of two, including \(n=1\). Indeed this happens when some prime divides every entry. The entries comprise all interior binomial coefficients of row \(2n\). If \(2n=p^h u\), \(p\nmid u\), then \((1+X)^{2n}=(1+X^{p^h})^u\) in \(\F_p[X]\). All interior coefficients vanish when \(u=1\), while for \(u>1\) the coefficient at \(p^h\) is \(u\ne0\). Hence \(2n\) must be a prime power, and the prime must be 2. \section{The ordinary group at a prime}\label{sec:ordinary} The four coefficient-parity pairs in Figure~\ref{fig:twenty} form \(G/2G\) for the \(2\times2\times2\) box. Both relations vanish modulo \(2\), leaving the images of \(x,y\) as a basis over \(\F_2\). For any box and prime \(p\), reducing the generators and relations modulo \(p\) gives \(G/pG\), with one dimension for each nontrivial cyclic summand of the \(p\)-primary subgroup \(G_p\). Call \(j\ge1\) active if \(e(p^j)>0\), and let \(t_p\) count the active indices. Corollary~\ref{cor:int} gives \(\lvert G_p\rvert\) and \(p^{t_p}G_p=0\). Since \(C(1)\) is square, its nullity modulo \(p\) equals \(\dim_{\F_p}(G/pG)\). We compute it using Han's theorem on relations between three powers of linear forms. \begin{theorem}\label{thm:ordinary} For every \(p\) the dimension of \(G/pG\) over \(\F_p\) is \(\max_{j\ge1} e(p^j)\), with empty maximum zero. If \(t_p\le2\), list the positive multiplicities \(e(p^j)\) and pad the list with zeros to obtain two numbers \(u\le v\). Then \[ G_p \ \simeq\ (\Z/p)^{\,v-u}\oplus(\Z/p^2)^{\,u}. \] This gives the complete \(p\)-primary group for every \(p^3>a+b+c-1\), and also for smaller primes having at most two active indices. \end{theorem} \begin{proof}[Deduction of Theorem~\ref{thm:generators}] Write \(G\) as a direct sum of cyclic groups of orders \(1T-1\). It follows that \[ \dim_k\ker C(1)=(T-\alpha)_++(T-\beta)_+ =(\beta-\alpha)/2 . \] Han's theorem computes this degree gap by a distance. We use the statement in Brenner and Kaid~\cite[Theorem~2.3]{BK}, which attributes the formula to Han~\cite{Han} and also gives Monsky's proof reference~\cite[Corollary~23]{Monsky}. Put \(\mathbf v=(W,B,A)\) and let \(\mathcal O\) be the triples of integers with odd sum. Write \(\operatorname{dist}_1\) for the sum-of-absolute-values distance to a set. The coordinates of \(\mathbf v\) satisfy the strict triangle inequalities. If \(s\) is the least integer for which \(\operatorname{dist}_1(p^s\mathbf v,\mathcal O)<1\), the gap is \[ \beta-\alpha=p^{-s}\bigl(1-\operatorname{dist}_1(p^s\mathbf v,\mathcal O)\bigr). \] It is zero when there is no such \(s\). Coordinate permutations preserve the formula. The sum of the coordinates of \(\mathbf v\) is even, so only \(s<0\) can qualify. For any \(d\ge2\) we claim that \[ \bigl(d-\operatorname{dist}_1(\mathbf v,d\mathcal O)\bigr)_+ =2e(d). \] Reducing the sides modulo \(d\) changes \(\mathbf v/d\) by an integer triple with even sum. We can therefore use \(\mathbf v'=(y+z,x+z,x+y)\). If an odd center \(d\mathbf h\) lies at distance \(1\), take their closest odd centers \(d\mathbf h,D\mathbf H\). The triangle inequality gives \(\|\mathbf h-l\mathbf H\|_1N\bmod d.\end{cases} \] Here \(d\ge2\), and these are all the irreducible factors. The quotients \(B_k/g_I\) therefore belong to \(\Z[q]\) and have constant term 1. Suppose their ideal were proper. A rational polynomial B\'ezout identity gives a nonzero integer in that ideal after denominators are cleared, so a maximal ideal containing it has residue characteristic \(p>0\). Its residue field is a finite extension of \(\F_p\). The constant terms force the image of \(q\) to be nonzero. Denote its multiplicative order by \(n\). The cyclotomics vanishing there have indices \(d=np^j\), \(j\ge0\), with \(d=1\) omitted from consideration. This follows from \(X^{mp}-1=(X^m-1)^p\) in characteristic \(p\) and the distinct roots of \(X^n-1\) when \(p\nmid n\). For each such \(d\), call an index good if \(k\bmod d\le N\bmod d\). If goodness is constant on \(I\) for every \(d\), any \(k\in I\) attains all the minimum valuations. Otherwise take the largest \(D\) on this chain for which there are both good and bad indices. Such \(D\) exists because all indices are good for \(d>N\). The good indices modulo \(D\) occur in the consecutive blocks \([tD,tD+(N\bmod D)]\). Since \(I\) contains both kinds of index, it contains a good endpoint of one of these blocks. That endpoint is congruent to either \(0\) or \(N\) modulo \emph{every} divisor of \(D\), so it is good for every smaller \(d\) on the chain. At every larger level the choice of index is immaterial. Thus a single \(B_k\) attains the minimum at every vanishing cyclotomic. Its quotient by \(g_I\) is nonzero in the residue field, a contradiction. \end{proof} For \(C\) take \(I=[b-a+1,b+a-1]\cap[0,N]\) in the integers. Powers of \(q\) are units in \(\Z[q,q^{-1}]\), so the entries generate \((s_1(q))\) over the integral Laurent ring. In particular the smallest ordinary Smith factor always equals \(s_1(1)\). The two-by-two weighted presentation also has a Smith form in each local ring at a maximal ideal, since after division by \(s_1\) one entry is a unit and serves as a pivot. \paragraph{Relation to earlier work.} Kuperberg's integral Laurent problem also asks that the diagonal form exist over \(\Z[q,q^{-1}]\). Theorem~\ref{thm:smith} determines it after rational coefficients are allowed. In Stanley's Theorem~3.1 the alphabet length varies through \(y=q^N\), with \(y\) independent and \(q\) in the coefficient field \(\Q(q)\)~\cite{Stanley}. Here \(N\) is a fixed integer and the zeros in the variable \(q\) determine the invariant factors. Chen, Guo, Jin and Liu studied the integer Carlitz matrices and used skew Schur polynomials to relate minors of Toeplitz matrices~\cite{CGJL}. Huang and Yang's path formula and interval decompositions \cite[\S\S4.2, 4.4]{HY} give a representation-theoretic method for computing ranks at a fixed root of unity. The length-\(N\) path between grades \(a-1\) and \(a+N-1\) on the tensor product with interval lengths \(a+b-1\) and \(a+c-1\) gives our transposed matrix up to row and column units. For \(k=b+i-j\), the exponent adjustment from their \(q^{j(N-k)}B_k(q)\) to \(e_k\) is \(\binom{b+i}{2}-\binom j2-Nj\). \paragraph{Kuperberg's question.} In each cyclic case of Theorem~\ref{thm:cyclic}, the group is \(\Z/M(1)\Z\). Choosing a side of length two for \(a\) gives a two-row presentation. Lemma~\ref{lem:interval} gives the first integer Smith factor \(s_1(1)\), and the determinant gives the second, \(M(1)/s_1(1)\). For any box, Theorem~\ref{thm:ordinary} determines the entire \(p\)-primary group whenever at most two of the polynomials \(\Phi_{p^j}\), \(j\ge1\), divide \(M(q)\). With three or more such indices, we still know the number of cyclic \(p\)-power summands. Corollary~\ref{cor:int} supplies a common divisor of the minors of \(C(1)\) at each intermediate size \(1