🔢 mathematics

Differences of squares of upper-triangular $2\times 2$ integer matrices

This paper provides a complete characterization and classification of upper-triangular $2\times 2$ integer matrices that can be expressed as the difference of two squares of upper-triangular integer matrices, using criteria based on the properties of their diagonal elements and a divisibility condition on their off-diagonal element.

Original authors: Andrej Dujella, Zrinka Franušić

Published 2026-04-28

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Andrej Dujella, Zrinka Franušić

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Mathematical "Difference of Squares" Mystery: A Simple Guide

Imagine you are a master chef, and you have a very specific rule for your kitchen: Every dish you serve must be the result of a "Subtraction Battle."

In this battle, you take two "Square Dishes" (think of these as perfectly symmetrical, square-shaped cakes) and you subtract one from the other to see what’s left.

In basic math, this is easy. If you want to know if the number $7$ can be made this way, you ask: "Is there a square number minus another square number that equals $7$?" (Yes: $4^2 - 3^2 = 16 - 9 = 7$ ). But if you want to make the number $6$, you’re out of luck. No two square cakes will ever leave you with exactly $6$ pieces.

This paper is about taking this "Subtraction Battle" and moving it from simple numbers into the world of "Matrix Cakes."

1. The "Matrix Cake" (The Setting)

Instead of just a single number, the authors are looking at Upper-Triangular $2 \times 2$ Matrices.

Think of a matrix as a layered sandwich.

The top layer has two ingredients.
The bottom layer has one ingredient on the left and one on the right.
But in an "upper-triangular" sandwich, the bottom-left corner is always empty (it’s just a zero).

The researchers want to know: If I give you a specific sandwich, can you find two "Square Sandwiches" that, when subtracted, result in exactly the sandwich I gave you?

2. The "Diagonal" vs. The "Secret Ingredient" (The Problem)

A matrix sandwich has two main parts:

The Diagonals (The Bread): These are the numbers running from the top-left to the bottom-right.
The Upper-Right Entry (The Secret Sauce): This is the number tucked in the top-right corner.

The authors discovered that for a sandwich to be "representable" (meaning it can be made via the subtraction battle), it has to pass two different tests.

Test 1: The Bread Test
The two numbers on the diagonal must be able to exist as "differences of squares" on their own. If one of your diagonal numbers is a "forbidden number" (like the number $6$ in our earlier example), the whole sandwich is impossible. The bread fails, so the sandwich fails.

Test 2: The Secret Sauce Test (The Hard Part)
Even if the bread is perfect, the "Secret Sauce" (the top-right number) is tricky. It isn't just any number; it is mathematically "tethered" to how you made the bread.

Imagine you have two square cakes. The way you slice them to get your diagonal numbers determines how much "crumbs" (the secret sauce) are left over. If the sandwich you were given has a "sauce" amount that doesn't match the "crumbs" produced by your bread, the sandwich is impossible.

3. The "Master Recipe" (The Result)

The authors spent the paper doing the heavy lifting to find the Master Recipe. They looked at every possible combination of numbers—odd numbers, even numbers, numbers divisible by $4$, and even numbers divisible by $16$.

They created a "Classification System." It’s like a giant flowchart for chefs:

Is the bread odd? If yes, follow path A.
Is the bread a multiple of 4? If yes, follow path B.
Does the sauce match the crumbs? If yes, you’ve made a valid matrix!

Why does this matter? (The "So What?")

You might ask, "Who cares about square sandwich subtraction?"

The authors mention that this isn't just a game. This math is a stepping stone to studying Diophantine Quadruples. These are special sets of numbers that have incredibly deep connections to how patterns work in the universe. By solving the "Matrix Sandwich" problem, they are helping mathematicians unlock the secrets of how these complex number patterns are built.

In short: They turned a complex, messy question about matrix structures into a clear, organized rulebook that tells you exactly which "sandwiches" can be made and which cannot.

Technical Summary: Differences of Squares of Upper-Triangular $2 \times 2$ Integer Matrices

Authors: Andrej Dujella and Zrinka Franušić
Subject Area: Number Theory, Matrix Theory, Diophantine Equations

1. The Problem

The paper investigates the characterization of upper-triangular $2 \times 2$ integer matrices that can be expressed as the difference of two squares of other upper-triangular integer matrices.

Formally, given a matrix $T = \begin{pmatrix} p & r \\ 0 & q \end{pmatrix} \in UT_2(\mathbb{Z})$ , the authors seek to determine the necessary and sufficient conditions on the integers $p, q,$ and $r$ such that there exist matrices $A, B \in UT_2(\mathbb{Z})$ satisfying:
$T = A^2 - B^2$
While the classical problem for integers (where $n = x^2 - y^2$ iff $n \not\equiv 2 \pmod 4$ ) is well-understood, the matrix version introduces an additional arithmetic dependency between the diagonal entries ( $p, q$ ) and the upper-right entry ( $r$ ).

2. Methodology

The authors employ a combination of algebraic manipulation, Diophantine analysis, and modular arithmetic:

Direct Computation: By squaring general upper-triangular matrices $A = \begin{pmatrix} a & b \\ 0 & d \end{pmatrix}$ and $B = \begin{pmatrix} x & y \\ 0 & u \end{pmatrix}$ , they derive the fundamental relation:
$A^2 - B^2 = \begin{pmatrix} a^2 - x^2 & b(a + d) - y(x + u) \\ 0 & d^2 - u^2 \end{pmatrix}$
Linear Diophantine Equations: The problem of finding $b$ and $y$ is reduced to solving the linear equation $b(a + d) - y(x + u) = r$ . The solvability of this equation depends on the greatest common divisor (gcd) of the coefficients.
Modular Classification: To provide a complete classification, the authors study the problem in the ring $UT_2(\mathbb{Z}_m)$ for $m=4$ and $m=16$ . This allows them to identify congruence-based obstructions that prevent a matrix from being representable in $\mathbb{Z}$ .
Minimization Function: They define a function $g(p, q)$ as the minimum possible value of $\gcd(a+d, x+u)$ over all possible integer representations of $p$ and $q$ as differences of squares.

3. Key Contributions and Results

The paper provides a complete characterization of representable matrices through several progressive theorems:

The Main Criterion (Theorem 1): A matrix $T$ is representable if and only if $p$ and $q$ are representable as differences of squares in $\mathbb{Z}$ , and there exist representations such that $\gcd(a+d, x+u)$ divides $r$ (or $r=0$ if the gcd terms are zero).
The $g(p, q)$ Characterization (Theorem 11): This is a significant theoretical refinement. It states that $T$ is representable if and only if $p, q$ are representable in $\mathbb{Z}$ and $g(p, q) \mid r$ . This proves that the "difficulty" of the representation is entirely captured by the interaction of the diagonal entries.
Complete Congruence Classification (Theorem 10): The authors provide a definitive list of all possible congruence classes for $p, q,$ $p, q,$ and $r$ $r$ that allow for representation. This includes:
- Odd Diagonals: If $p, q$ are odd, $r$ must be even if $p \equiv q \pmod 4$ , but $r$ can be anything if $p \not\equiv q \pmod 4$ .
- Mixed Parity: If one diagonal is odd and the other is a multiple of 4, $r$ can be any integer.
- Even Diagonals: If $p, q \equiv 0 \pmod 4$ , the requirements on $r$ become more stringent (depending on $p, q \pmod{16}$ ).

4. Significance

The research holds importance in two main areas:

Diophantine Quadruples: The study of $D(n)$ -quadruples (sets where $xy+n$ is a square) is deeply linked to the representability of $n$ as a difference of squares. By characterizing these representations in the non-commutative (but structured) ring of upper-triangular matrices, the authors provide tools for future research into $D(n)$ -tuples in matrix rings.
Algebraic Structure Theory: The paper demonstrates how the transition from a commutative ring ( $\mathbb{Z}$ ) to a non-commutative ring ( $UT_2(\mathbb{Z})$ ) complicates the "difference of squares" problem by introducing coupling between the diagonal and off-diagonal elements, requiring more sophisticated tools than simple modular arithmetic.