Technical Summary: Partition Regularity of Pythagorean Pairs

1. Problem Statement and Context

The paper addresses a fundamental open problem in Ramsey theory concerning the partition regularity of the Pythagorean equation $x^2 + y^2 = z^2$. While Schur's theorem (1916) established that $x+y=z$ is partition regular, and Rado's theorem (1933) characterized partition regularity for linear systems, the status of non-linear polynomial equations has remained largely elusive. Specifically, the question of whether every finite coloring of the positive integers $\mathbb{N}$ contains a monochromatic Pythagorean triple ($x, y, z$) has been a notorious problem posed by Erdős and Graham.

Prior to this work, the only known result for Pythagorean triples was a computer-assisted proof for the specific case of 2-colorings (2016). Previous theoretical attempts, such as those by the first author and Host [21], utilized Gowers uniformity properties of multiplicative functions but failed to resolve the Pythagorean case because the relevant algebraic expressions lacked the necessary "positivity" properties when $n=0$.

The authors define a Pythagorean pair as $(x, y) \in \mathbb{N}^2$ such that there exists $z \in \mathbb{N}$ satisfying either $x^2 + y^2 = z^2$ or $x^2 + z^2 = y^2$. The primary goal is to prove that such pairs are partition regular, and to extend this to density regularity and level sets of multiplicative functions.

2. Methodology

The proof strategy combines ergodic theory, the theory of multiplicative functions, and novel concentration estimates. The approach proceeds through the following stages:

2.1. Ergodic Reformulation

Using the Furstenberg correspondence principle, the combinatorial problem is reformulated into an ergodic setting. The existence of monochromatic solutions is reduced to proving the positivity of certain multiple recurrence integrals involving measure-preserving actions of the multiplicative semigroup $(\mathbb{N}, \times)$. Specifically, for a set $A$ of positive measure, one must show:
$$\mu(T^{-1}_{\ell(m^2-n^2)}A \cap T^{-1}_{\ell' mn}A) > 0$$
for distinct $m, n$.

2.2. Decomposition of Multiplicative Functions

The core of the argument relies on the decomposition of the space of completely multiplicative functions $\mathcal{M}$ into two classes:

1. Aperiodic functions: Functions that do not correlate with any Dirichlet character or Archimedean character ($n^{it}$).
2. Pretentious functions: Functions that "pretend" to be a twisted Dirichlet character $\chi \cdot n^{it}$.

The authors utilize the fact that for aperiodic functions, the relevant averages vanish (Proposition 2.4, 2.10). The challenge lies in the pretentious case, where the averages do not automatically vanish and require careful analysis.

2.3. Novel Concentration Estimates

A critical innovation in this paper is the development of nonlinear concentration estimates for multiplicative functions evaluated on quadratic forms.

• Type I (Difference of squares): The authors adapt existing linear concentration estimates (from [21, 35]) to handle expressions like $f((Qm+1)^2 - (Qn)^2)$.
• Type II (Sum of squares): The authors prove a new, non-trivial concentration estimate (Proposition 2.11, 5.1) for expressions of the form $f((Qm+1)^2 + (Qn)^2)$. This estimate relies on the fact that primes $p \equiv 1 \pmod 4$ split in the field $\mathbb{Q}(i)$, allowing the authors to control the behavior of $f$ on sums of squares using a "pretentious distance" restricted to these primes.

2.4. Weighted Averages and Positivity

To overcome the lack of positivity in the integrands (a failure point of previous approaches), the authors introduce specific weight functions $w_\delta$ and $\tilde{w}_{\delta, c}$. These weights are designed to be supported on regions where the logarithmic ratios of the terms are close to specific constants, ensuring that the real part of the integral remains positive when restricted to the trivial character (the identity function).

3. Key Contributions and Results

3.1. Partition Regularity of Pythagorean Pairs

Theorem 1.1: For every finite coloring of $\mathbb{N}$, there exist distinct $x, y$ of the same color and $z \in \mathbb{N}$ such that $x^2 + y^2 = z^2$ (or $x^2 + z^2 = y^2$).

• This resolves the question of whether Pythagorean pairs are partition regular.
• The result is generalized to equations of the form $ax^2 + by^2 = cz^2$ where $a, b, c$ are perfect squares.

3.2. Density Regularity

Theorem 1.2: The authors establish a stronger density version. If a set $\Lambda \subset \mathbb{N}$ has positive upper multiplicative density (with respect to a multiplicative Følner sequence), then $\Lambda$ contains distinct $x, y$ such that $ax^2 + by^2 = cz^2$ for some $z$.

• This rules out additive density as the correct notion for this problem (since the set of odd numbers has additive density 1/2 but contains no Pythagorean triples).

3.3. Pythagorean Triples on Level Sets

Theorem 1.5: Let $f: \mathbb{N} \to S^1$ be a completely multiplicative function taking finitely many values. Then there exist distinct $x, y, z$ such that $x^2 + y^2 = z^2$ and $f(x) = f(y) = f(z) = 1$.

• This provides strong evidence for the full partition regularity of Pythagorean triples, as level sets of such functions represent a broad class of "structured" colorings.
• The result is extended to equations $ax^2 + by^2 = cz^2$ under specific conditions on $a, b, c$ (e.g., $a=c$, $b=c$, or $a+b=c$).

3.4. Generalizations

The methodology is shown to be flexible enough to handle:

• Other dilation-invariant pairs (Theorem 1.8).
• General linear forms $L_1(m,n)L_2(m,n)$ and $L_3(m,n)L_4(m,n)$ (Section 1.5.2).
• More general expressions involving powers and products of linear forms (Section 1.5.3).

4. Significance and Claims

The authors claim that their work resolves the partition regularity of Pythagorean pairs, a problem that had remained open despite significant prior efforts. They explicitly state that their approach overcomes the specific obstruction in [21] where the relevant expressions failed to be non-negative.

The paper does not claim to have solved the full partition regularity of Pythagorean triples (i.e., finding $x, y, z$ all of the same color) for arbitrary finite colorings. Instead, it proves this for:

1. Pairs $(x, y)$ with a third variable $z$ of any color.
2. Triples $(x, y, z)$ where the coloring is generated by the level sets of finite-valued completely multiplicative functions.

The authors identify the remaining gap: proving partition regularity for triples in general colorings would require extending their results to cases where the coefficients $a, b, c$ in $ax^2 + by^2 = cz^2$ do not satisfy specific square conditions or Rado's condition, or where the parametrization involves quadratic forms that do not factor into linear forms (as noted in Problem 1 and Problem 2 of Section 1.6).

The work is presented as a "general approach" that combines Gowers uniformity with new concentration estimates, opening the door to solving other previously intractable partition regularity problems involving non-linear patterns.