arXiv++ Combinatorics

A latin square of order $n$ is an $n\times n$ array in which each of $n$ symbols occurs exactly once in each row and column. A transversal in such a square is a selection of $n$ entries that includes one representative of each row and column, and one of each symbol. For all even orders $n\ge 28$ except $n=30$, we construct a latin square of order $n$ in which every pair of transversals share at least one entry. We conjecture that in our squares there is no single entry that is common to all transversals. We prove this conjecture for $n\le10\,000$ by finding transversals using an algorithm that is likely to be of independent interest. We say that a transversal is dominant if it intersects every other transversal of the same latin square. We show that there exist latin squares of order $n$ that have a dominant transversal for $n\in\{5,7\}$ and also for all $n\ge8$ such that $n\not\equiv3\bmod4$.

A $k$-plane of a $d$-dimensional array is a subarray formed by fixing $d-k$ coordinates and allowing the remaining $k$ coordinates to vary freely. A Latin hypercube of dimension $d$ and order $n$ is an $n\times n\times\cdots\times n$ array of dimension $d$ containing symbols from an $n$-set, such that each $1$-plane contains each of the possible entries exactly once. A transversal in a Latin hypercube of order $n$ is a set of $n$ entries of the hypercube, no pair of which agree in any coordinate or contain the same symbol. The aim of this paper is to construct Latin hypercubes that have transversals but which have many entries that are not in any transversal, or for which the number of disjoint transversals is limited. We show the following results in the case when the dimension $d$ is even. For all even $n\ge 10$ there exists a Latin hypercube of order $n$ that contains a transversal but for which all transversals hit one $(d-2)$-plane. For $n\in\{6,8\}$ there exists a Latin hypercube of order $n$ that contains a transversal but for which all transversals hit one of two $(d-2)$-planes. For even $d>2$ there is a Latin hypercube of order $n=4$ that contains a transversal but has $2^d$ entries that are not in any transversal. Our constructions use a quasigroup $(Q,\ast)$ to increase the dimension of a Latin hypercube using the rule $H_d(x_1,\dots,x_d)=H_{d-1}(x_1,\dots,x_{d-1})\ast x_d$. We give several characterisations which allow us to diagnose which entries of $H_d$ are in transversals in terms of properties of $H_{d-1}$ and $Q$.

Paratopism is a well known action of the wreath product $\mathcal{S}_n\wr\mathcal{S}_3$ on Latin squares of order $n$. A paratopism that maps a Latin square to itself is an autoparatopism of that Latin square. Let $\mathrm{Par}(n)$ denote the set of paratopisms that are an autoparatopism of at least one Latin square of order $n$. We prove a number of general properties of autoparatopisms. Applying these results, we determine $\mathrm{Par}(n)$ for $n\le17$. We also study the proportion of all paratopisms that are in $\mathrm{Par}(n)$ as $n\rightarrow\infty$.

An $n\times n\times\dots\times n$ hypercube is made from $n^d$ unit hypercubes. Two unit hypercubes are neighbours if they share a $(d-1)$-dimensional face. In each step of a dismantling process, we remove a unit hypercube that has precisely $d$ neighbours. A move is balanced if the neighbours are in $d$ orthogonal directions. In the extremal case, there are $n^{d-1}$ independent unit hypercubes left at the end of the dismantling. We call this set of hypercubes a solution. If a solution is projected in $d$ orthogonal directions and we get the entire $[n]^{d-1}$ hypercube in each direction, then the solution is perfect. We show that it is possible to use a greedy algorithm to test whether a set of hypercubes forms a solution. Perfect solutions turn out to be precisely those which can be reached using only balanced moves. Every perfect solution corresponds naturally to a Latin hypercube. However, we show that almost all Latin hypercubes do not correspond to solutions. In three dimensions, we find at least $n$ perfect solutions for every $n$, and we use our greedy algorithm to count the perfect solutions for $n\le6$. We also construct an infinite family of imperfect solutions and show that the total size of its three orthogonal projections is asymptotic to the minimum possible value. Our results solve several conjectures posed in a proceedings paper by Barát, Korondi and Varga. If our dismantling process is reversed we get a build-up process very closely related to well-studied models of bootstrap percolation. We show that in an important special case our build-up reaches the same maximal position as bootstrap percolation.

A binary frequency square of type $(n;λ_0,λ_1)$ is a $(0,1)$-matrix of order $n$ with $λ_0$ zeros and $λ_1$ ones in each row and in each column. Two such squares are orthogonal if there are exactly $λ_1^2$ cells where both squares contain ones. A set of binary MOFS is a set of binary frequency squares in which each pair is orthogonal. A set of binary MOFS of type $(n;λ_0,λ_1)$ is type maximal if there is no square of the type $(n;λ_0,λ_1)$ that is orthogonal to every square in the set. A Doehlert-Klee design consists of points $V$ and blocks $B$, where every pair of points occurs in precisely $Λ$ blocks and every point occurs in precisely $R$ blocks, where $R^2=Λ|B|$. We show that sets of binary MOFS are equivalent to a particular kind of Doehlert-Klee design. In a distinct application, Doehlert-Klee designs can also be used to construct sets of binary MOFS that are cyclically generated from their first rows. We use these connections to find new constructions for sets of type-maximal binary MOFS.

A perfect $1$-factorisation of a graph is a decomposition of that graph into $1$-factors such that the union of any two $1$-factors is a Hamiltonian cycle. A Latin square of order $n$ is row-Hamiltonian if for every pair $(r,s)$ of distinct rows, the permutation mapping $r$ to $s$ has a single cycle of length $n$. We report the results of a computer enumeration of the perfect $1$-factorisations of the complete bipartite graph $K_{11,11}$. This also allows us to find all row-Hamiltonian Latin squares of order $11$. Finally, we plug a gap in the literature regarding how many row-Hamiltonian Latin squares are associated with the classical families of perfect $1$-factorisations of complete graphs.

A Latin hypercuboid of order $n$ is a $d$-dimensional matrix of dimensions $n\times n\times\cdots\times n\times k$, with symbols from a set of cardinality $n$ such that each symbol occurs at most once in each axis-parallel line. If $k=n$ the hypercuboid is a Latin hypercube. The Latin hypercuboid is \emph{completable} if it is contained in a Latin hypercube of the same order and dimension. It is \emph{extendible} if it can have one extra layer added. In this note we consider which Latin hypercuboids are completable/extendible. We also consider a generalisation that involves multidimensional arrays of sets that satisfy certain balance properties. The extendibility problem corresponds to choosing representatives from the sets in a way that is analogous to a choice of a Hall system of distinct representatives, but in higher dimensions. The completability problem corresponds to partitioning the sets into such SDRs. We provide a construction for such an array of sets that does not have the property analogous to completability. A related concept was introduced by Häggkvist under the name $(m,m,m)$-array. We generalise a construction of $(m,m,m)$-arrays credited to Pebody, but show that it cannot be used to build the arrays that we need.

We prove that, for all even $n\geq10$, there exists a latin square of order $n$ with at least one transversal, yet all transversals coincide on $ \big\lfloor n/6 \big\rfloor$ entries. These latin squares have at least $ 19 n^2/36 + O(n)$ transversal-free entries. We also prove that for all odd $m\geq 3$, there exists a latin square of order $n=3m$ divided into nine $m\times m$ subsquares, where every transversal hits each of these subsquares at least once.

A quasigroup is a pair $(Q, *)$ where $Q$ is a non-empty set and $*$ is a binary operation on $Q$ such that for every $(a, b) \in Q^2$ there exists a unique $(x, y) \in Q^2$ such that $a*x=b=y*a$. Let $(Q, *)$ be a quasigroup. A pair $(x, y) \in Q^2$ is a commuting pair of $(Q, *)$ if $x * y = y * x$. Recently, it has been shown that every rational number in the interval $(0, 1]$ can be attained as the proportion of ordered pairs that are commuting in some quasigroup. For every positive integer $n$ we establish the set of all integers $k$ such that there is a quasigroup of order $n$ with exactly $k$ commuting pairs. This allows us to determine, for a given rational $q \in (0, 1]$, the spectrum of positive integers $n$ for which there is a quasigroup of order $n$ whose proportion of commuting pairs is equal to $q$.

A sequence covering array, denoted \textsf{SCA}$(N;t,v)$, is a set of $N$ permutations of $\{0, \dots, v-1 \}$ such that each sequence of $t$ distinct elements of $\{0, \dots, v-1\}$ reads left to right in at least one permutation. The minimum number of permutations such a sequence covering array can have is $t!$ and Levenshtein conjectured that if a sequence covering array with $t!$ permutations exists, then $v \in \{t,t+1\}$. In this paper, we prove that if an \textsf{SCA}$(7!;7,v)$ exists, then $v \leq 9$. We do this by analysing connections between sequence covering arrays and a special kind of covering array called an excess coverage array. A strength 2 excess coverage array, denoted \textsf{CA}$_{X}(N;2,k,v)$, is an $N \times k$ array with entries from $\{0, \dots, v - 1\}$ such that every ordered pair of distinct symbols appear at least once in each pair of columns and all other pairs appear at least twice. We demonstrate computationally that there is a unique \textsf{CA}$_{X}(42;2,5,6)$, and we prove that this array alone does not satisfy necessary conditions we establish for the existence of an \textsf{SCA}$(7!;7,10)$. Furthermore, we find the maximum possible number of columns for strength 2 excess coverage arrays with symbol sets of sizes between 2 and 6, and more broadly investigate binary excess coverage arrays.

Suppose that $k$ is a function of $n$ and $n\to\infty$. We show that with probability $1-O(1/n)$, a uniformly random $k\times n$ Latin rectangle contains no proper Latin subsquare of order $4$ or more, proving a conjecture of Divoux, Kelly, Kennedy and Sidhu. We also show that the expected number of subsquares of order 3 is bounded and find that the expected number of subsquares of order 2 is $\binom{k}{2}(1/2+o(1))$ for all $k\le n$.

A Latin square of order $n$ is an $n\times n$ matrix in which each row and column contains each of $n$ symbols exactly once. For $ε>0$, we show that with high probability a uniformly random Latin square of order $n$ has no proper subsquare of order larger than $n^{1/2}\log^{1/2+ε}n$. Using this fact we present a canonical labelling algorithm for Latin squares of order $n$ that runs in average time bounded by a polynomial in $n$. The algorithm can be used to solve isomorphism problems for many combinatorial objects that can be encoded using Latin squares, including quasigroups, Steiner triple systems, Mendelsohn triple systems, $1$-factorisations, nets, affine planes and projective planes.

A $d$-dimensional Latin hypercube of order $n$ is a $d$-dimensional array containing symbols from a set of cardinality $n$ with the property that every axis-parallel line contains all $n$ symbols exactly once. We show that for $(n, d) \notin \{(4,2), (6,2)\}$ with $d \geq 2$ there exists a $d$-dimensional Latin hypercube of order $n$ that contains no $d$-dimensional Latin subhypercube of any order in $\{2,\dots,n-1\}$. The $d=2$ case settles a 50 year old conjecture by Hilton on the existence of Latin squares without proper subsquares.

A \emph{frequency square} is a matrix in which each row and column is a permutation of the same multiset of symbols. Two frequency squares $F_1$ and $F_2$ with symbol multisets $M_1$ and $M_2$ are \emph{orthogonal} if the multiset of pairs obtained by superimposing $F_1$ and $F_2$ is $M_1\times M_2$. A set of MOFS is a set of frequency squares in which each pair is orthogonal. We first generalise the classical bound on the cardinality of a set of MOFS to cover the case of \emph{mixed type}, meaning that the symbol multisets are allowed to vary between the squares in the set. A frequency square is \emph{binary} if it only uses the symbols 0 and 1. We say that a set $\mathcal{F}$ of MOFS is \emph{type-maximal} if it cannot be extended to a larger set of MOFS by adding a square whose symbol multiset matches that of at least one square already in $\mathcal{F}$. Building on pioneering work by Stinson, several recent papers have found conditions that are sufficient to show that a set of binary MOFS is type-maximal. We generalise these papers in several directions, finding new conditions that imply type-maximality. Our results cover sets of binary frequency squares of mixed type. Also, where previous papers used parity arguments, we show the merit of arguments that use moduli greater than 2.

A \emph{Latin square} is a matrix of symbols such that each symbol occurs exactly once in each row and column. A Latin square $L$ is \emph{row-Hamiltonian} if the permutation induced by each pair of distinct rows of $L$ is a full cycle permutation. Row-Hamiltonian Latin squares are equivalent to perfect $1$-factorisations of complete bipartite graphs. For the first time, we exhibit a family of Latin squares that are row-Hamiltonian and also achieve precisely one of the related properties of being column-Hamiltonian or symbol-Hamiltonian. This family allows us to construct non-trivial, anti-associative, isotopically $L$-closed loop varieties, solving an open problem posed by Falconer in 1970.

Let $\mathbb{F}$ be a finite field of odd order and $a,b\in\mathbb{F}\setminus\{0,1\}$ be such that $χ(a) = χ(b)$ and $χ(1-a)=χ(1-b)$, where $χ$ is the extended quadratic character. Let $Q_{a,b}$ be the quasigroup upon $\mathbb{F}$ defined by $(x,y)\mapsto x+a(y-x)$ if $χ(y-x) \ge 0$, and $(x,y)\mapsto x+b(y-x)$ if $χ(y-x) = -1$. We show that $Q_{a,b} \cong Q_{c,d}$ if and only if $\{a,b\}= \{α(c),α(d)\}$ for some $α\in \textrm{aut}(\mathbb{F})$. We also characterise $\textrm{aut}(Q_{a,b})$ and exhibit further properties, including establishing when $Q_{a,b}$ is a Steiner quasigroup or is commutative, entropic, left or right distributive, flexible or semisymmetric. In proving our results we also characterise the minimal subquasigroups of $Q_{a,b}$.

A relation on a $k$-net$(n)$ (or, equivalently, a set of $k-2$ mutually orthogonal Latin squares of order $n$) is an $\mathbb{F}_{2}$ linear dependence within the incidence matrix of the net. Dukes and Howard (2014) showed that any 6-net(10) satisfies at least two non-trivial relations, and classified the relations that could appear in such a net. We find that, up to equivalence, there are $18\,526\,320$ pairs of MOLS satisfying at least one non-trivial relation. None of these pairs extend to a triple. We also rule out one other relation on a set of $3$-MOLS from Dukes and Howard's classification.

A PSCA$(v, t, λ)$ is a multiset of permutations of the $v$-element alphabet $\{0, \dots, v-1\}$ such that every sequence of $t$ distinct elements of the alphabet appears in the specified order in exactly $λ$ of the permutations. For $v \geq t \geq 2$, we define $g(v, t)$ to be the smallest positive integer $λ$ such that a PSCA$(v, t, λ)$ exists. We show that $g(6, 3) = g(7, 3) = g(7, 4) = 2$ and $g(8, 3) = 3$. Using suitable permutation representations of groups we make improvements to the upper bounds on $g(v, t)$ for many values of $v \leq 32$ and $3\le t\le 6$. We also prove a number of restrictions on the distribution of symbols among the columns of a PSCA.

We survey the impact of Lieb's influential paper "Proofs of some conjectures on permanents" [J. Math. Mech. 16 1966, 127-134], which introduced the famous permanental dominance conjecture. This conjecture has defied all attacks for over half a century, although a number of related conjectures have recently been resolved.

A Latin square has six conjugate Latin squares obtained by uniformly permuting its (row, column, symbol) triples. We say that a Latin square has conjugate symmetry if at least two of its six conjugates are equal. We enumerate Latin squares with conjugate symmetry and classify them according to several common notions of equivalence. We also do similar enumerations under additional hypotheses, such as assuming the Latin square is reduced, diagonal, idempotent or unipotent. Our data corrected an error in earlier literature and suggested several patterns that we then found proofs for, including (1) The number of isomorphism classes of semisymmetric idempotent Latin squares of order $n$ equals the number of isomorphism classes of semisymmetric unipotent Latin squares of order $n+1$, and (2) Suppose $A$ and $B$ are totally symmetric Latin squares of order $n\not\equiv0\bmod3$. If $A$ and $B$ are paratopic then $A$ and $B$ are isomorphic.