---
title: Generalized Pythagorean Partition Regularity
url: https://www.emergentmind.com/topics/generalized-pythagorean-partition-regularity-conjecture
type: topic
---

# Generalized Pythagorean Partition Regularity

The generalized Pythagorean partition regularity problem concerns monochromatic solutions to homogeneous quadratic equations of the form
\[
a x^2+b y^2=c z^2,\qquad a,b,c\in \mathbb{Z}\setminus\{0\}.
\]
In current arXiv literature, the phrase “Generalized Pythagorean Partition Regularity Conjecture” is used in two closely related senses. One is a **pair-partition-regularity** conjecture: under natural square conditions on the coefficients, every finite coloring of $\mathbb{N}$ should contain a solution in which a prescribed pair of variables, typically $(x,y)$, has the same color [2407.08360]. The other is a stronger **full partition regularity** conjecture: for $a,b,c\in\mathbb{N}$, the equation should admit a monochromatic triple $(x,y,z)$ in every finite coloring of $\mathbb{N}$ if and only if $(a,b,c)$ is a Rado triple, meaning $a=c$, $b=c$, or $a+b=c$ [2508.09778]. The contemporary theory establishes broad pair-regular regimes, strong structured full-triple results for pretentious multiplicative colorings and actions, and several obstructions and negative instances, while leaving the unrestricted full-triple problem open.

## 1. Formulations, terminology, and scope

For the equation
\[
a x^2+b y^2=c z^2,
\]
partition regularity with respect to the pair $(x,y)$ means that for every finite coloring of $\mathbb{N}$ there exist distinct $x,y\in\mathbb{N}$ of the same color and some $z\in\mathbb{N}$ such that the equation holds. The analogous notions for $(x,z)$ and $(y,z)$ are defined by symmetry. Full partition regularity is stronger: it requires $x,y,z$ all to be monochromatic. Full partition regularity implies pair regularity for all three pairs, but the converse fails; a standard example is $x^2+y^2=4z^2$, which is pair-regular for each pair but not fully partition regular [2407.08360].

For the pair problem, working over $\mathbb{N}$ or over $\mathbb{Q}_{>0}$ is equivalent: pair partition regularity over $\mathbb{N}$ holds if and only if it holds over $\mathbb{Q}_{>0}$ [2407.08360]. This equivalence reflects the multiplicative character of the analytic framework used in recent work.

A second notion central to the subject is **density regularity**. Given a multiplicative Følner sequence $\mathfrak D=(\mathcal D_K)$, one defines upper multiplicative density by
\[
d_{\mathfrak D}^*(A)=\limsup_{K\to\infty}\frac{|A\cap \mathcal D_K|}{|\mathcal D_K|}.
\]
An equation is density regular with respect to $(x,y)$ if every subset $A\subset\mathbb N$ of positive upper multiplicative density contains distinct $x,y\in A$ and some $z\in\mathbb N$ satisfying the equation. Density regularity implies pair partition regularity because any finite coloring has a color class of positive multiplicative density [2407.08360].

The stronger, full-triple conjecture is formulated in explicitly Rado-theoretic terms. In that formulation, a triple $(a,b,c)\in\mathbb N^3$ is called a **Rado triple** if $a=c$, $b=c$, or $a+b=c$, and the conjecture asserts that $a x^2+b y^2=c z^2$ is partition regular exactly in those cases [2508.09778]. The necessity is motivated by the linear shadow $a x+b y=c z$, since full partition regularity of the quadratic equation would force partition regularity of the associated linear equation.

## 2. Natural coefficient conditions and conjectural classifications

The pair-regularity conjecture proposed in "Partition regularity of generalized Pythagorean pairs" identifies a natural square-condition trichotomy. It predicts that if at least one of $ac$, $bc$, or $(a+b)c$ is a square, then
\[
a x^2+b y^2=c z^2
\]
is partition regular with respect to $(x,y)$ [2407.08360]. By symmetry, the analogous condition for other pairs is obtained by renaming variables. Canonical model cases are $x^2+2y^2=z^2$ and $x^2+y^2=2z^2$.

The same paper also gives necessary conditions. If the equation is pair partition regular with respect to $(x,y)$, then at least one of
\[
ac,\quad bc,\quad (a+b)c,\quad ab(a+b)c
\]
must be a perfect square [2407.08360]. This does not prove the conjectured sufficiency, but it sharply narrows the admissible coefficient regimes.

Congruence obstructions show that the residual regime in which only $ab(a+b)c$ is square is delicate. If $a\equiv 2\pmod 4$, $b$ is odd, and $ab(a+b)c$ is a square, then the equation is not pair partition regular with respect to $(x,y)$; the example $x^2+2y^2=6z^2$ is explicitly negative. If $a,b$ are odd, $c\equiv 2\pmod 4$, and $ab(a+b)c$ is a nonzero square, then pair partition regularity for $(x,y)$ forces $ab\equiv 1\pmod 8$; thus $3x^2+5y^2=30z^2$ fails pair regularity [2407.08360].

The full-triple conjecture uses a different classification principle. In that setting, the proposed criterion is not a square condition on coefficient products, but the Rado condition $a=c$, $b=c$, or $a+b=c$ [2508.09778]. These two conjectural schemes are compatible but logically distinct: the pair problem asks for same-color coincidence in a chosen pair, whereas the full problem asks for a monochromatic solution in all three variables.

| Coefficient regime | Conclusion | Status |
|---|---|---|
| $ac$ or $bc$ square | Pair regularity for $(x,y)$ | Unconditional |
| $(a+b)c$ square | Pair regularity for $(x,y)$ | Conditional on Conjecture 2 |
| $a=c$ or $b=c$ or $a+b=c$ | Full monochromatic triple | Conjectural in general |
| Only $ab(a+b)c$ square | Mixed: some cases fail | Partly obstructed |

## 3. Established pair-regularity results

The central 2024 advance is the unconditional theorem that if either $ac$ or $bc$ is a square, then
\[
a x^2+b y^2=c z^2
\]
is partition regular with respect to $(x,y)$ [2407.08360]. By symmetry, if either $ac$ or $-ab$ is a square, then the equation is partition regular with respect to $(y,z)$. The corresponding density statement is also proved: under the same condition, the equation is density regular with respect to $(x,y)$.

A second theorem addresses the remaining natural regime. Assuming an Elliott-type conjecture about correlations of aperiodic multiplicative functions along irreducible binary quadratic forms, if $(a+b)c$ is a square, possibly zero, then the equation is partition regular with respect to $(x,y)$ [2407.08360]. The same conditional input yields density regularity. Under Rado’s linear condition $a=c$ or $b=c$ or $a+b=c$, the conditional theory further implies pair partition regularity simultaneously for all three pairs $(x,y)$, $(x,z)$, and $(y,z)$.

These results generalize the earlier squares-only theory. "Partition regularity of Pythagorean pairs" proved pair partition regularity for $a x^2+b y^2=c z^2$ when $a,b,c$ are all squares, and also showed that partitions generated by the level sets of a completely multiplicative finite-valued function contain full monochromatic Pythagorean triples [2309.10636]. The 2024 work extends the range of coefficients from the all-squares setting to the more flexible square-product conditions $ac$, $bc$, and $(a+b)c$ [2407.08360].

Several examples illustrate the distinction between unconditional, conditional, and open regimes. The equation $x^2+2y^2=z^2$ is pair partition regular for $(x,y)$ because $ac=1$ is square. The equation $x^2+y^2=2z^2$ fits the conjectural pattern because $(a+b)c=4$ is square, and it is proved conditionally, not unconditionally. The scaled Pythagorean equation $x^2+y^2=4z^2$ is pair-regular for every pair but not fully partition regular. Open examples explicitly listed include
\[
x^2+17y^2=34z^2,\qquad 8x^2+3y^2=66z^2,\qquad 5x^2+7y^2=105z^2
\]
[2407.08360].

## 4. Parametrizations and analytic mechanism

The modern proofs are built around parametrizations of solution families and a reduction from partition regularity to density regularity for pairs of binary quadratic forms. When $ac=d^2$, solutions are parametrized by
\[
x=k c\big(a m^2-b n^2\big),\qquad y=k\cdot 2ac\,mn,\qquad z=k d\big(a m^2+b n^2\big),
\]
and when $(a+b)c=d^2\neq 0$, by
\[
x=kc\,(m^2-2bmn-abn^2),\qquad y=kc\,(m^2+2amn-abn^2),\qquad z=kd\,(m^2+abn^2).
\]
If $a+b=0$, an alternative parametrization is
\[
x=k\,(m^2+ac\,n^2),\qquad y=k\,(m^2-ac\,n^2),\qquad z=k\cdot 2amn
\]
[2407.08360].

These parametrizations reduce the problem to showing that certain pairs $(P_1,P_2)$ of binary quadratic forms are **good for density regularity**. In the unconditional regime, one form is irreducible quadratic and the other is reducible of the shape $\gamma mn$; in the conditional regime, one must handle pairs of distinct irreducible quadratic forms with the same leading $m^2$ coefficient [2407.08360].

The analytic core is a structure/randomness dichotomy for multiplicative functions. Pretentious multiplicative functions are those approximable, in pretentious distance, by $\chi\cdot n^{it}$; non-pretentious ones are aperiodic. For a homogeneous polynomial $P$, the theory introduces a $P$-weighted pretentious distance
\[
D_P(f,g)^2=\sum_{p\in \mathcal P_P}\frac{w_P(p)}{p}\big(1-\Re(f(p)\overline{g(p)})\big),
\]
with weights determined by the local root count of $P(n,1)$ modulo $p$ [2407.08360]. The aperiodic case is handled by vanishing-of-correlation theorems for expressions such as $f_1(P_1(\cdot))f_2(P_2(\cdot))$, while the pretentious case is controlled by concentration estimates along irreducible quadratic forms.

A key innovation of the 2024 paper is uniform concentration for irreducible binary quadratic forms after freezing congruence classes on highly divisible lattices. This produces approximations of the form
\[
f(P_c(Qm+a,Qn+b))\approx \chi(P_c(a,b))\,(P_c(Qm,Qn))^{it}\,e^{G_{P,N}(f,K)},
\]
with explicit control in terms of pretentious norms on prime ranges and an $O(K^{-1/2})$ error [2407.08360]. The independence of the phase from the multiplicative averaging parameter $Q$ is crucial, because it allows averaging over multiplicative Følner sets to eliminate oscillation.

Weighted positivity then supplies the major-arc contribution. Trapezoidal weights
\[
w_{\delta,P_1,P_2}(m,n)=F_\delta\big((P_1(m,n))^{it}(P_2(m,n))^{-it}\big)\mathbf 1_{\{P_1>0,P_2>0\}}
\]
have positive average when either one form is reducible or the two forms share the same leading coefficient [2407.08360]. Combined with minor-arc vanishing, this yields a positivity theorem for weighted correlations on the compact group of completely multiplicative functions, and Theorem 2.4 of the paper converts that analytic positivity into combinatorial density regularity. This framework extends the earlier measure-on-$M$ and concentration method developed for classical Pythagorean pairs [2309.10636].

## 5. Pretentious full-triple recurrence and structured colorings

The strongest current results toward full monochromatic generalized Pythagorean triples occur in the setting of pretentious multiplicative dynamics. "Recurrence for pretentious systems along generalized Pythagorean triples" formulates an ergodic version of the Rado-triple conjecture for multiplicative actions $(X,\mathcal X,\mu,T_n)$ and proves it for pretentious actions, meaning actions whose spectral measures are supported on pretentious completely multiplicative functions [2508.09778].

Its main theorem states that if $(a,b,c)$ is a Rado triple, then for every pretentious multiplicative action and every measurable set $A$ with $\mu(A)>0$, and for every $\varepsilon>0$, there exist distinct $x,y,z\in\mathbb N$ such that
\[
\mu\big(A\cap T_x^{-1}A\cap T_y^{-1}A\cap T_z^{-1}A\big)\ge (\mu(A))^4-\varepsilon
\]
and
\[
a x^2+b y^2=c z^2.
\]
This gives the ergodic-theoretic form of the conjecture in the “critical” structured case [2508.09778].

A concrete combinatorial corollary concerns colorings generated by finitely many pretentious completely multiplicative functions $f_1,\dots,f_s:\mathbb N\to\mathbb S^1$. For every open arc $I\subset\mathbb S^1$ containing $1$, there exist distinct $x,y,z$ such that
\[
a x^2+b y^2=c z^2,\qquad f_j(x),f_j(y),f_j(z)\in I\ \text{for all }j.
\]
Thus any finite coloring coming from level sets of finitely many pretentious multiplicative functions contains a monochromatic generalized Pythagorean triple [2508.09778]. This extends the earlier finite-image multiplicative-coloring result for the classical Pythagorean equation [2309.10636].

The proof strategy is again spectral and pretentious. It combines linear and quadratic concentration estimates, highly divisible multiplicative grids, Chinese Remainder Theorem factorizations that isolate prime supports of parametrizing forms, Archimedean control through polynomial visit-time sets, and characteristic-factor reductions culminating in Chu’s inequality. The theorem also delineates the boundary of the method: outside the Rado regime, multiple recurrence can fail, and the full conjecture for general multiplicative actions remains open [2508.09778].

## 6. Broader context, analogues, and open landscape

The generalized Pythagorean problem sits between several neighboring theories. Over finite fields, a robust monochromatic analogue is known: for every $r$ and exponents $\alpha,\beta,\gamma\in\mathbb N$, every sufficiently large prime field $\mathbb F_p$ admits at least $c_{r,\alpha,\beta,\gamma}p^2$ monochromatic solutions to
\[
x^\alpha+y^\beta=z^\gamma
\]
under any $r$-coloring [1606.07334]. In particular, $x^2+y^2=z^2$ is quantitatively partition regular in all sufficiently large finite fields. This finite-field result is methodologically different, relying on polynomial regularity, counting, and Ramsey lemmas in a compact Abelian model rather than the multiplicative-pretentious framework used over $\mathbb N$.

In higher numbers of variables, a Rado-type classification is available. For diagonal forms in $k$th powers,
\[
\sum_{i=1}^s c_i x_i^k=0,
\]
partition regularity over $\mathbb N$ is characterized, for sufficiently many variables, by the existence of a nonempty subset of coefficients summing to zero. For squares, the threshold is $s_0(2)=5$, yielding partition regularity of
\[
x_1^2+x_2^2+x_3^2+x_4^2=x_5^2
\]
in every finite coloring [1806.05002]. This shows that generalized Pythagorean partition regularity is well understood in the many-variable diagonal regime even though the three-variable case remains subtle.

Nonlinear partition regularity beyond linearity has also been explored from an ultrafilter perspective. "Kernel Partition Regularity Beyond Linearity" proves partition regularity for weak nonlinear Rado systems and multiplicatively decorated Rado systems whenever the linear backbone satisfies the column condition, but it does not settle genuine equations of the form $x^2+y^2=z^2$ or $a x^2+b y^2=c z^2$ with all variables synchronized in one color [2407.05542]. This distinction is important: polynomially augmented linear systems and generalized Pythagorean triples belong to different structural classes.

Historically, the combinatorial geometry of the Pythagorean triple hypergraph was studied through its sum property and the absence of Steiner triple systems. Before later progress on pair regularity and structured monochromaticity, computational work established large finite two-colorings with no monochromatic Pythagorean triple on long initial segments and showed that the hypergraph of Pythagorean triples contains no Steiner triple system [1505.02222]. Those results did not resolve the infinite problem, but they clarified why straightforward finite-obstruction arguments are difficult.

The main open problems are now sharply formulated. For pair regularity, the remaining barrier is the Elliott-type Conjecture 2 on vanishing correlations of aperiodic multiplicative functions along irreducible binary quadratic forms; its resolution would complete the $(a+b)c$-square case and substantially close the pair conjecture [2407.08360]. For full monochromatic triples, no Rado triple is yet known to satisfy unrestricted partition regularity in every finite coloring of $\mathbb N$, and the ergodic and combinatorial conjectures remain open beyond pretentious systems [2508.09778]. A plausible implication is that the decisive obstacle lies not in parametrization itself, but in extending minor-arc orthogonality and major-arc phase control from pair configurations to genuinely triple-synchronized quadratic patterns.

Source: https://www.emergentmind.com/topics/generalized-pythagorean-partition-regularity-conjecture