Pairwise Reflection Symmetry in Frequency Rectangles

arXiv:2606.28315 · cs.DM, math.CO · Submitted 2026-06-26 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Computer Science. The day in computer science: software, systems and theory on arXiv.

Tom: I'm Tom.

Jane: And I'm Jane, here with Lu, Meng and our in-house large language model, Lalam.

Tom: We're looking at the paper: "Pairwise Reflection Symmetry in Frequency Rectangles".

Jane: Here's the gist of the paper. Pairwise reflection-symmetry in generalized Latin rectangles requires that on every pair of columns, each ordered symbol pair (p, q) occurs as often as its reversal (q, p) This order-balance is precisely what makes headto-head comparisons unbiased The paper examines structural properties of this symmetry in generalized Latin rectangles, which extend Latin squares to matrices with controlled column frequencies Existence of such objects for odd λ turns out to be remarkably more subtle than for even λ The authors obtain the insight that a column multiplicity of λ = 1 can be achieved if and only if n is a power of two Structural Definitions and Properties The paper introduces the concept of pairwise reflection-symmetry, which demands that for every pair of distinct positions j1, j2, the number of times symbol pair (p, q) appears in columns j1 and j2 matches the number of times (q, p) appears This is formalized by defining a bivariate frequency matrix F(j1,j2)(A) which must be symmetric with respect to transposition A key observation is that if A is pairwise reflection-symmetric, then for any choice of columns j1!= j2 we have f(j1) = f(j2), meaning the column-frequency vector for each column is independent of the column index Furthermore, Observation 3 states that any A in URS(n, λ) has an even row-count λn Existence and Constructions The paper establishes existence results for various multiplicities λ, noting that for even n and even λ ≥ 2, there exists an URS(n, λ) by vertically stacking copies of a matrix from Lemma 2 Theorem 1 provides a construction for every even n ≥ 2 where the row-count is λ = (n - 3)!!

Tom: The existence for odd λ is more subtle, leading to the definition of kURS(n), the smallest odd multiplicity such that an URS(n, λ) exists Corollary 4 shows that for any fixed odd r, the sequence 2k+1r admits a uniform bound kURS2k+1r ≤ (2r − 3)!!

Jane: Group-Theoretic Characterization A powerful group-theoretic characterization is presented in Theorem 4, which states that Mat

G: ∈ URS(n, λ) if and only if G is generously transitive This links the combinatorial property directly to the properties of a subgroup G of Sn Corollary 5 further shows that for every n ≥ 3 and every abelian group A of order n not being an elementary abelian 2-group, Mat

Dih(A): ∈ URS(n, 2) Computational Insights and Bounds The paper utilizes a quadratically constrained integer program (QCP) to computationally search for these objects Experiments show that many such matrices possess an underlying group-theoretic structure The enumeration experiments in Table 1 reveal the feasibility of finding representatives, showing that infeasibility of the QCP for n = 12 is shown approximately 16 times faster than for n = 10 The paper also relates pairwise reflection-symmetry to perfect sequence covering arrays, noting that every A in URS(n, λ) is a PSCA of strength 2 whose row-inverse is also a PSCA of strength 2 The investigation concludes by highlighting the dichotomy between the even regime, where existence is guaranteed, and the odd regime, which presents a subtle challenge regarding kURS(n) Several open questions remain regarding whether the superexponential bound on kURS(n) is tight or if a simpler bound applies Future directions suggest exploring optimization variants and information-theoretical density to better understand these structures The paper also provides an explicit lower bound on the number of reduced elements in URS(n, 2) using group theory The final result shows that distinct group structures yield distinct reduced representatives in URS(n, 2) The paper concludes by stating that pairwise reflection-symmetry forms a natural generalization of 2-subsquare completeness Page 1 --- Pairwise Reflection Symmetry in Generalized Latin Rectangles Enrico Iurlano, Günther R.

Tom: Raidl Algorithms and Complexity Group, TU Wien, Austria Abstract Many combinatorial designs ask for equal distribution of given symbols across the entries of a matrix The paramount examples are Latin squares, where each symbol from the set 1 to n appears once per row and column of an n x n matrix Generalized Latin rectangles extend this to λn x n matrices with repeated symbols under controlled column frequencies In this more general setting, we examine structural properties of pairwise reflection-symmetry, which requires that on every pair of columns, each ordered symbol pair (p, q) occurs as often as its reversal (q, p) This order-balance is precisely what makes headto-head comparisons unbiased Page 2 --- A related idea is classical in statistical experimental design under the heading of residual (or, in modern terminology, carry-over) effects Pairwise reflection-symmetry is conceptually related, yet different in two respects: It requires a certain balance not only between adjacent positions but between every pair of columns, each of which must realize the symmetry in isolation; on the other hand, it alters the Latin square property that each column is a permutation by requiring λ-fold repetition of each symbol per column Page 3 --- We will later encounter particular permutations in Sn: An involution π ∈ Sn satisfies π ◦π = id and a derangement π ∈ Sn is free of fixed points, i.e., π(j)!= j for all j in

n: We recall that every permutation in Sn can be decomposed into a finite number of cycles and that an involutory derangement consists exclusively of length-2 cycles Page 4 --- Definition 1 (adapted from

1: ).

Title and authors: Tom: We just touched on the core idea of "Pairwise Reflection Symmetry in Frequency Rectangles," which is about balancing symbol pairs across all column combinations, and how that links to group theory and existence conditions. Now let's look at what the authors actually proved about it.

Jane: The summary says they introduce this symmetry by defining a bivariate frequency matrix, which has to be symmetric when you transpose it; that’s the formal way they capture that head-to-head comparison being unbiased.

Lu: They observe that if you have this property, then the frequency vector for each column becomes independent of its position in the matrix. This is a key structural simplification they found, moving away from needing to check every pair of columns individually.

Meng: So, instead of checking all pairs, they only need to check the structure once because that one check implies something about all other pairs. That’s a huge efficiency win if that holds true across the whole matrix.

Tom: It does hold for them, and this leads to Observation three: any matrix in this class, URS(n, lambda), must have an even row count of lambda times n <ref:2606.28315#pg3>. This is a necessary condition they found for existence.

Jane: That’s a neat constraint because it immediately rules out any matrices where the repetition factor would result in an odd total count across all rows and columns combined. It sets a boundary on what kind of structures are even possible to build under these rules.

Lu: And then they dive into existence proofs. For even sizes n and even lambda greater than or equal to two, they show you can always construct one by stacking copies of a smaller matrix that's already known to have this property.

Meng: So for the easy cases, if we pick an even size and an even repetition number, the construction is straightforward. That’s something I could actually use in a deployment scenario if it worked reliably.

Tom: But they admit that existence for odd lambda is much more subtle than for even lambda, which means that when you're dealing with odd repetition factors, the math gets much deeper and harder to pin down.

Jane: That's exactly what they say. The subtlety in the odd case suggests there might be some structural roadblocks or specific constraints that prevent us from building these matrices easily. This points toward the next part of their work.

The paper's summary: Tom: Moving beyond just proving existence, the authors suggest a few ways to improve how we understand these objects, which mostly involves group theory and relating them to other known combinatorial structures.

Jane: They bring in the idea that a matrix has this symmetry if and only if it's generated by a generously transitive group. That’s a way to translate this complex frequency requirement into the language of permutation groups, which is powerful because we have tools for those groups.

Lu: That connection is very strong because it links the combinatorial property directly to the algebraic properties of subgroups within Sn, the symmetric group on n elements. It means if you find a subgroup that isn't generously transitive, you know immediately that no matrix generated by it will satisfy this symmetry.

Meng: So, instead of brute-forcing every possible matrix structure to check for symmetry, we can just check the group properties of whatever structure we build first. That shifts the effort from checking matrices to checking groups.

Tom: They also connect this symmetry directly to perfect sequence covering arrays, which is a different type of combinatorial design called a PSCA, showing that every matrix in this class has very specific properties with respect to these covers.

Jane: That's useful because PSCA are often used in areas like experimental design where you want to ensure certain patterns appear frequently enough. This gives them another way to view the symmetry property—it’s not just about symbol pairs, it’s also about covering sequences.

Lu: And they show that every matrix in this class is a PSCA of strength two, and its row inverse also has that same strength two property. That provides a nice structural link between the forward and backward operations on these matrices.

Meng: For production, knowing it’s a PSCA of strength two tells us something about how robust the resulting design is against certain types of errors or deviations during execution. It's a measure of its stability.

The paper's improvements: Tom: So, to wrap up on this paper "Pairwise Reflection Symmetry in Frequency Rectangles," the main point is that we have a clear structural characterization using generous transitivity groups and strong links to sequence covering arrays.

Jane: The implication for us is that when designing these complex combinatorial designs, we don't need to test every single combination of entries; we can check the properties of the underlying group instead.

Lu: And for those working on odd repetition factors, the existence proofs are much harder, which sets a boundary on what kind of structures you can realistically expect to find in that regime.

Meng: From an engineering standpoint, knowing these structural constraints helps us avoid wasting time building matrices that fundamentally can't satisfy the symmetry requirement. It’s about pruning the search space early.

Tom: Exactly. And for anyone looking to build these things, the main thing to remember is that you need to check if your chosen group is generous enough before you start constructing anything based on this paper.

Jane: So, even though we're not building a finished product yet, this research gives us a much sharper tool for understanding why certain combinatorial structures exist and others don't. That’s really important for the theory.

Lu: It definitely provides a clearer roadmap for where the research needs to go next, especially concerning those hard odd multiplicity cases.

Meng: Yeah, we need more practical tools that can help us navigate those harder existence proofs or maybe some information theoretical bounds to make sure our constructions actually work in practice when scaled up.

Tom: Alright then, that’s our take on "Pairwise Reflection Symmetry in Frequency Rectangles." We'll be moving on to the next paper shortly.

Conclusion: Tom: So, this paper, "Pairwise Reflection Symmetry in Frequency Rectangles," is really about imposing a specific balance on how symbols are distributed across columns in these generalized Latin rectangles.

Jane: It looks at this order balance—where every pair of columns has the symbol pair p-q appearing exactly as often as q-p—and it shows that this symmetry forces a lot of things to be independent.

Lu: The structural insight here is that if you have this pairwise reflection symmetry, then the frequency vector for any single column becomes independent of which column index you are looking at. That simplifies the whole analysis a lot.

Meng: So, instead of checking every possible pair combination for every matrix entry, we only need to check that one property once and it tells us about everything else. That sounds like something you could use in a high-throughput system if that holds true at scale.

Lalam: From my perspective, the real meat is the existence part—they prove that for even column counts and even repetition numbers, you can always construct these matrices by stacking smaller ones together.

Tom: That's good to know because it means we have a solid recipe for building these things when we stick to those even parameters. But what about the odd repetition numbers?

Jane: That’s where things get tricky; the authors admit that existence for odd repetition numbers is much more subtle and less straightforward than for the even ones.

Lu: Exactly. They introduce a new concept called kURS, which is basically the smallest odd repetition number you need to have to find an object in that class. They give us some bounds on this, like kURS for a certain size N being at most something related to two plus the remainder of N divided by three.

Meng: I worry about those bounds when I think about deployment; we need concrete numbers on how hard it is to find that smallest odd repetition number because if it's huge, the construction becomes completely unfeasible for our current hardware.

Tom: The group theory connection is also interesting here; they show that a matrix has this symmetry if and only if the underlying permutation group is generous transitive.

Jane: That links the counting problem to the structure of permutations in a way that’s really deep, showing it's not just about symbol positions but about how those positions are organized by group actions.

Lu: And they provide a very specific result for abelian groups; any matrix built from the direct product of two cyclic groups has this symmetry if that group isn't an elementary abelian two-group. That’s a concrete condition we can use to filter out possibilities quickly.

Meng: It’s helpful to have those filters, but I still need to know how much computational power is required for the actual search process when we try to find these matrices using quadratic constrained integer programming.

Tom: The experiments they ran suggest that while finding these objects isn't always easy, their search method can be quite efficient on certain sizes, showing feasibility even if it takes some time.

Jane: So the main thing to take away is the contrast between the even and odd cases—one side is a guaranteed construction method, and the other side requires deep structural bounds.

Lu: They also linked this symmetry to perfect sequence covering arrays, which shows this isn't just an abstract math curiosity; it connects directly to designs used in experimental setup.

Meng: That connection to experimental design is what matters for us; if we can use these matrices to design better tests, that’s where the real value is.

Tom: Alright, so we covered the mechanics of pairwise reflection symmetry in generalized Latin rectangles today and how it links group theory to sequence covering arrays.

Jane: The paper highlights that while even repetition numbers have guaranteed construction methods, odd ones require a more complex structural search constrained by bounds on kURS.

Lu: The main implication is the clear group-theoretic characterization, giving us a way to test symmetry using subgroup properties instead of checking every entry.

Meng: For deployment, it means we need to be very careful about the column frequencies when designing these kinds of combinatorial structures at scale.

Tom: We'll keep an eye on those bounds for odd repetition numbers as we look at how they might affect future design models.

Enrico Iurlano, Günther R. Raidl

TU Wien

cs.DM, math.CO

Submitted: 2026-06-26

Updated: 2026-11-03

Comments: 18 pages, 2 figures; Julia source code provided as ancillary file

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 66/100

The gist: Pairwise reflection-symmetry in generalized Latin rectangles requires that on every pair of columns, each ordered symbol pair (p, q) occurs as often as its reversal (q…

Key concepts

Pairwise Reflection-Symmetry
This is a condition requiring that for any two columns, the frequency of observing symbol pair (p, q) must equal the frequency of observing its reversal (q, p). This order-balance ensures unbiased comparisons between any two columns.
Generalized Latin Rectangle (URS(n, $\lambda$))
These are matrices where each row and column has a controlled frequency of symbols. The specific class URS(n, $\lambda$) is defined by the requirement that every pair of columns exhibits pairwise reflection-symmetry.
Generously Transitive Group
A permutation group $G$ is generously transitive if it possesses certain structural properties related to how it acts on the set of indices. The paper proves that a matrix belongs to URS(n, $\lambda$) if and only if its underlying permutation group $G$ is generously transitive.
kURS(n)
This denotes the smallest odd multiplicity $\lambda$ for which a pairwise reflection-symmetric generalized Latin rectangle of size $n \times n$ exists. The paper provides bounds for this value, showing it is related to the prime factorization of $n$. (e.g., kURS(2k+1r) $\le$ (2r - 3)!!)

Terminology

Summary

Pairwise reflection-symmetry in generalized Latin rectangles requires that on every pair of columns, each ordered symbol pair (p, q) occurs as often as its reversal (q, p) This order-balance is precisely what makes headto-head comparisons unbiased The paper examines structural properties of this symmetry in generalized Latin rectangles, which extend Latin squares to matrices with controlled column frequencies Existence of such objects for odd λ turns out to be remarkably more subtle than for even λ The authors obtain the insight that a column multiplicity of λ = 1 can be achieved if and only if n is a power of two

Structural Definitions and Properties

The paper introduces the concept of pairwise reflection-symmetry, which demands that for every pair of distinct positions j1, j2, the number of times symbol pair (p, q) appears in columns j1 and j2 matches the number of times (q, p) appears This is formalized by defining a bivariate frequency matrix F(j1,j2)(A) which must be symmetric with respect to transposition A key observation is that if A is pairwise reflection-symmetric, then for any choice of columns j1!= j2 we have f(j1) = f(j2), meaning the column-frequency vector for each column is independent of the column index Furthermore, Observation 3 states that any A in URS(n, λ) has an even row-count λn

Existence and Constructions

The paper establishes existence results for various multiplicities λ, noting that for even n and even λ ≥ 2, there exists an URS(n, λ) by vertically stacking copies of a matrix from Lemma 2 Theorem 1 provides a construction for every even n ≥ 2 where the row-count is λ = (n - 3)!! The existence for odd λ is more subtle, leading to the definition of kURS(n), the smallest odd multiplicity such that an URS(n, λ) exists Corollary 4 shows that for any fixed odd r, the sequence 2k+1r admits a uniform bound kURS2k+1r ≤ (2r − 3)! p. 9!

Group-Theoretic Characterization

A powerful group-theoretic characterization is presented in Theorem 4, which states that Mat[G] ∈ URS(n, λ) if and only if G is generously transitive This links the combinatorial property directly to the properties of a subgroup G of Sn Corollary 5 further shows that for every n ≥ 3 and every abelian group A of order n not being an elementary abelian 2-group, Mat[Dih(A)] ∈ URS(n, 2)

Computational Insights and Bounds

The paper utilizes a quadratically constrained integer program (QCP) to computationally search for these objects Experiments show that many such matrices possess an underlying group-theoretic structure The enumeration experiments in Table 1 reveal the feasibility of finding representatives, showing that infeasibility of the QCP for n = 12 is shown approximately 16 times faster than for n = 10 The paper also relates pairwise reflection-symmetry to perfect sequence covering arrays, noting that every A in URS(n, λ) is a PSCA of strength 2 whose row-inverse is also a PSCA of strength 2

Conclusion and Open Questions

The investigation concludes by highlighting the dichotomy between the even regime, where existence is guaranteed, and the odd regime, which presents a subtle challenge regarding kURS(n) Several open questions remain regarding whether the superexponential bound on kURS(n) is tight or if a simpler bound applies Future directions suggest exploring optimization variants and information-theoretical density to better understand these structures The paper also provides an explicit lower bound on the number of reduced elements in URS(n, 2) using group theory The final result shows that distinct group structures yield distinct reduced representatives in URS(n, 2) The paper concludes by stating that pairwise reflection-symmetry forms a natural generalization of 2-subsquare completeness

--- Page 1 ---

Pairwise Reflection Symmetry in Generalized Latin Rectangles

Enrico Iurlano, Günther R. Raidl Algorithms and Complexity Group, TU Wien, Austria Abstract Many combinatorial designs ask for equal distribution of given symbols across the entries of a matrix The paramount examples are Latin squares, where each symbol from the set 1 to n appears once per row and column of an n x n matrix Generalized Latin rectangles extend this to λn x n matrices with repeated symbols under controlled column frequencies In this more general setting, we examine structural properties of pairwise reflection-symmetry, which requires that on every pair of columns, each ordered symbol pair (p, q) occurs as often as its reversal (q, p) This order-balance is precisely what makes headto-head comparisons unbiased

--- Page 2 ---

A related idea is classical in statistical experimental design under the heading of residual (or, in modern terminology, carry-over) effects Pairwise reflection-symmetry is conceptually related, yet different in two respects: It requires a certain balance not only between adjacent positions but between every pair of columns, each of which must realize the symmetry in isolation; on the other hand, it alters the Latin square property that each column is a permutation by requiring λ-fold repetition of each symbol per column

--- Page 3 ---

We will later encounter particular permutations in Sn: An involution π ∈ Sn satisfies π ◦π = id and a derangement π ∈ Sn is free of fixed points, i.e., π(j)!= j for all j in [n] We recall that every permutation in Sn can be decomposed into a finite number of cycles and that an involutory derangement consists exclusively of length-2 cycles

--- Page 4 ---

Definition 1 (adapted from [1]). A matrix A ∈ [n] λn×µn is called a generalized Latin rectangle if (Aij = qj=1 = λ and (Aij = qi=1 = µ for all q in [n], i. p. 2e., the columnwise as well as the rowwise frequencies per symbol are equally distributed) For the special case λ = µ = 1 we recover the notion of a classical Latin square

--- Page 5 ---

Definition 3 We say that A ∈ [n]L×M is pairwise reflection-symmetric if for all (j1, j2) ∈ [M] 2 with j1!= j2 we have for any (p, q) ∈ [n] 2 with p!= q that Aij1 = p, Aij2 = q = Aij1 = q, Aij2 = p This definition enforces that the bivariate frequency matrix associated to each column-pair (j1,j2) is symmetric

--- Page 6 ---

Observation 1 If A is pairwise reflection-symmetric, then for any choice of columns j1!= j2 we have f(j1) = f(j2), in particular the entry of the column-frequency associated to symbol q is independent of the column index

--- Page 7 ---

Lemma 1 Assume A ∈ [n]λn×n has just permutations of [n] as rows, i.e., f(i) = (1)q∈[n] for i = 1,..., λn Then A is pairwise reflection-symmetric iff A ∈ URS(n, λ)

--- Page 8 ---

Proposition 1 The following three assertions are valid (i) Every URS(n, 1) consists of n pairwise distinct rows (ii) For each odd n ≥ 3, every URS(n, 2) consists of 2n pairwise distinct rows (iii) Let n ≥ 3 be odd, A ∈ URS(n, λ), and suppose there are precisely ρ rows hosting the same permutation. Then we have ρ ≤ floor(λn/(n + 3))

--- Page 9 ---

Definition 4 Let us denote by URS(n, λ) the class of all matrices A in [n]λnxµn which are pairwise reflection-symmetric, and of uniform f(j) = (λ)q∈[n], j = 1... µ

--- Page 10 ---

Observation 2 Permuting rows of an URS(n, λ) preserves pairwise reflection-symmetry, and permuting its columns by some π in Sn replaces F(j1,j2) by F(π-1(j1),π-1(j2)) Observation 3 Any A ∈ URS(n, λ) has an even row-count λn

--- Page 11 ---

Lemma 3 We have A in URS(n, 1) iff A is a 2-subsquare complete Latin square

--- Page 12 ---

Proposition 2 An URS(n, 1) exists iff n = 2k for some k ∈ N Uset of zero

--- Page 13 ---

Definition 7 Let G be a finitely generated abelian group. Then G ∼= Zr ⊕ Zd1 ⊕ Zd2 ⊕ · · ·oplus Zd s exists, such that divisibility chain d1 d2 · ··· ds exist

--- Page 14 ---

Corollary 3 Every finite elementary abelian 2-group G is isomorphic to Zk squared for some k in N Uset of zero

--- Page 15 ---

Theorem 3 We have A □⊠∼A ∈ URS(n n, λ λ), up to identification of the entries with [n u]-valued scalars

--- Page 16 ---

Corollary 4 Let N = 2k+1·r with k ≥ 0 and r ≥ 1 odd. Then kURS(N) ≤ (2r − 3)! p. 5 p. 5!, with the convention (−1)! p. 1 p. 1!:= 1 for r = 1

--- Page 17 ---

Theorem 4 Let G ≤ Sn be a subgroup of order G = λn. Then Mat[G] ∈ URS(n, λ) iff G is generously transitive

--- Page 18 ---

Proposition 3 For n ≥ 3, let L(n) be the number of labeled abelian groups of order n with a fixed identity, i.e., the abelian Cayley tables of order n

--- Page 19 ---

Corollary 5 For every n ≥ 3 and every abelian group A of order n not being an elementary abelian 2-group, we have Mat[Dih(A)] ∈ URS(n, 2)

--- Page 20 ---

Theorem 4 Let G ≤ Sn be a subgroup of order G = λn. Then Mat[G] ∈ URS(n, λ) iff G is generously transitive

--- Page 21 ---

Corollary 5 For every n ≥ 3 and every abelian group A of order n not being an elementary abelian 2-group, we have Mat[Dih(A)] ∈ URS(n, 2)

--- Page 13 ---

Definition 7 Let G be a finitely generated abelian group. Then G ∼= Zr ⊕ Zd1 ⊕ Zd2 ⊕ · · ·oplus Zd s exists, such that divisibility chain d1 d2 ds exist

--- Page 16 ---

Corollary 4 Let N = 2k+1·r with k ≥ 0 and r ≥ 1 odd. Then kURS(N) ≤ (2r − 3)!!, with the convention (−1)!!

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. The AI system can be used to construct matrices in URS(n, λ) by leveraging group-theoretic constructions like Mat[G] where G is a generously transitive subgroup of Sn with order lambda n. This allows for explicit generation of objects satisfying pairwise reflection-symmetry and uniform column distribution.

  2. The system can computationally search for the smallest possible odd multiplicity kURS(n) using a quadratically constrained integer program to find representatives of URS(n, λ). This provides a method to determine the structural limits on existence for even n.

  3. The AI can perform direct product constructions of URS matrices, specifically A times ~A, which preserves pairwise reflection-symmetry and uniformity in the resulting larger matrix size lambda n by lambda. This enables the construction of larger combinatorial designs from smaller known ones.

  4. For any given pair of columns (j1, j2) and symbols (p, q), the system can determine if a bivariate frequency matrix F(j1, j2) is symmetric by checking if F(j1, j2)[p,q] equals F(j1, j2)[q,p], which is equivalent to checking the condition for generous transitivity of the underlying permutation group G.

  5. The system can verify if a Latin square A in URS(n, 1) is a 2-subsquare complete Latin square using Lemma 3 and its proof structure. This allows for quick structural checks on smaller combinatorial designs to confirm their symmetry properties.

Related papers