---
title: Interlaced Arithmetic Progressions
url: https://www.emergentmind.com/topics/interlaced-arithmetic-progressions
type: topic
---

# Interlaced Arithmetic Progressions

“Interlaced arithmetic progressions” does not appear in the cited literature as a single universally fixed technical definition. Across current work, the phrase is most naturally used as an umbrella term for configurations in which two or more arithmetic-progression structures coexist in a coupled way: the same data may support arithmetic progression structure simultaneously in indices and values, arithmetic progressions may form structured families in parameter space, several progressions may cover or pack a common ambient set, or a single object may tile or encode two distinct progression systems at once [1304.1829] [1905.02591] [2211.16058].

## 1. Terminological scope and formal models

The literature supports several precise formalizations that capture different aspects of interlacing. Some papers explicitly note that they do not define “interlaced arithmetic progressions” as a standalone notion and instead study nearby structures such as arithmetic progressions of parameter pairs, overlapping progression families, or simultaneous progressions centered at a common value [1905.02591] [1210.6612] [1902.04176].

| Formalization | Basic object | Representative result |
|---|---|---|
| Double progression | Indices and values both form APs | Bounded-gap sequences and additive squares |
| Parameter-space progression | \((a,b)\) with \(\{a,a+b,\dots,a+lb\}\subseteq A\) | The parameter set is itself large |
| Coupled progression systems | Several APs through a common base point or two lattices \(a\mathbb Z,\beta\mathbb Z\) | Orbit-covering rigidity and simultaneous tilings |
| Geometric parametrization | Centered triples on a conic | Rational points on an elliptic curve classify centered 3-APs |
| Packing and decomposition | Unions or disjoint shifted copies of APs | Asymptotic packing bounds and FPT algorithms |

This multiplicity of models matters. In some settings, “interlaced” means synchronized arithmetic structure in two coordinates; in others it means arithmetic progressions of arithmetic progressions, or disjoint/overlapping coexistence inside a host set. A common thread is that the object of study is no longer one isolated progression but a higher-order system of progressions whose parameters, supports, or incidence relations are constrained simultaneously.

## 2. Double arithmetic progressions and additive-square structure

The most direct formalization is the double arithmetic progression. For an increasing sequence \(a_1<a_2<\cdots\), a double \(k\)-term arithmetic progression consists of indices \(p_1<\cdots<p_k\) such that both \(\{p_1,\dots,p_k\}\) and \(\{a_{p_1},\dots,a_{p_k}\}\) are arithmetic progressions. For \(k=3\), this is the simultaneous system
\[
i+k=2j,\qquad a_i+a_k=2a_j.
\]
The problem is therefore intrinsically “interlaced”: one AP lives in the index line and one AP lives in the value line [1304.1829].

This formulation is equivalent to additive squares in the gap word. If \(x_i=a_{i+1}-a_i\), then two adjacent blocks of equal length and equal sum correspond exactly to a double \(3\)-term arithmetic progression in the cumulative sequence. The paper “On Double 3-Term Arithmetic Progressions” develops this equivalence in detail and proves that the bounded-gap sequence problem, the additive-square problem for infinite words over a finite positive alphabet, and a covering formulation for \(\mathbb N\) are equivalent. It also shows that if every bounded-gap sequence contains a double \(3\)-term progression, then every finite coloring of \(\mathbb N\) has a color class containing one, and that existence of double \(3\)-term progressions is equivalent to existence with arbitrarily large index spacing [1304.1829].

The finite extremal theory is already nontrivial. The exact value
\[
w^\ast(2,3)=17
\]
is established, while lower bounds
\[
w^\ast(3,3)\ge 414,\qquad w^\ast(2,4)\ge 30830
\]
show how rapidly the combinatorics becomes difficult. The same paper conjectures that \(w^\ast(3,3)\) exists and that \(w^\ast(2,4)\) does not. In this line of work, “interlacing” is not metaphorical: it is the requirement that two arithmetic regularities occur on the same selected terms.

## 3. Abundance phenomena and arithmetic progressions of arithmetic progressions

A second major formalization treats interlacing at the level of progression parameters. If \(A\) is a subset of a commutative semigroup \(S\) and \(l\in\mathbb N\) is fixed, one considers
\[
B=\{(a,b): \{a,a+b,a+2b,\dots,a+lb\}\subseteq A\}.
\]
The key question is whether largeness of \(A\) implies largeness of \(B\).

For A.P.-rich sets, the answer is affirmative in an elementary form: if \(A\subseteq S\) contains arithmetic progressions of arbitrary finite length, then \(B\subseteq S\times S\) is itself A.P. rich. Equivalently, one obtains arbitrarily long arithmetic progressions of parameter pairs \((a,b)\), hence families
\[
a+ie,\ a+ie+(b+if),\ \dots,\ a+ie+l(b+if)\in A
\]
with linearly varying starting points and linearly varying common differences [1905.02591]. This is one of the clearest precise realizations of an interlaced progression family: the progressions are not independent, but coupled by an affine rule in the family index \(i\).

The same phenomenon persists for stronger largeness notions. For \(\mathcal{CR}\)-sets in a commutative semigroup \((S,+)\), if \(A\) is a \(\mathcal{CR}\)-set, then
\[
\{(a,b): \{a,a+b,a+2b,\ldots,a+lb\}\subseteq A\}
\]
is a \(\mathcal{CR}\)-set in \((S\times S,+)\); if \(A\) is an essential \(\mathcal{CR}\)-set in \(\mathbb N\), then the same parameter set is an essential \(\mathcal{CR}\)-set in \((\mathbb N\times\mathbb N,+)\) [2211.12372]. These results extend the Furstenberg–Glasner perspective from piecewise syndetic sets to broader algebraic-Ramsey classes.

The conceptual significance is that the “collection of progressions in \(A\)” is itself progression-rich. In this sense, the subject moves from single arithmetic progressions to arithmetic progressions in parameter space and even arithmetic progressions in the space of progression-tuples. That higher-order viewpoint is central to modern uses of the term.

## 4. Coupled progression systems: dynamics, tilings, conics, and additive constraints

Another large cluster of results studies systems in which two or more arithmetic progressions are coupled by an external structure. In arithmetic dynamics, for a polynomial \(f\in\mathbb Z[x]\) of degree at least \(2\), one asks whether the orbit
\[
\operatorname{Orb}_f(t)=\{t,f(t),f^{\circ 2}(t),\dots\}
\]
can be covered by finitely many arithmetic progressions through the common base point \(t\). The answer is highly rigid: if
\[
\delta_{f,t}\!\left(\bigcup_{i=1}^k (t+n_i\mathbb Z)\right)=1,
\]
then one of those progressions already contains the entire orbit. More generally, the relative density covered by \(k\) such progressions is universally bounded by
\[
\delta_k=1-\prod_{i=1}^k\left(1-\frac1{p_i}\right),
\]
where \(p_i\) is the \(i\)-th prime, although densities arbitrarily close to \(1\) become accessible when \(k\) is allowed to grow [2403.04397]. Here interlacing means several congruence-based arithmetic threads anchored at the same point.

A different simultaneous-progression problem appears in translational tilings. A measurable \(f\) on \(\mathbb R\) may tile simultaneously by two arithmetic progressions \(a\mathbb Z\) and \(\beta\mathbb Z\) at levels \(p\) and \(q\). The arithmetic relation between \(a\) and \(\beta\) is decisive. If \(a,\beta\) are rationally independent, then any \(p,q\in\mathbb C\) are possible and one can realize them with
\[
\operatorname{mes}(\operatorname{supp}f)\le a+\beta,
\]
but if \((p,q)\) is not proportional to \((\beta,a)\), bounded support is impossible. For positive coprime integers \(p,q\), the support infimum is
\[
a+\beta-\min\left\{\frac{a}{q},\frac{\beta}{p}\right\},
\]
and it is not attained. In the rationally dependent case, reduced to \((n,m)\in\mathbb N^2\), one necessarily has \((p,q)\propto(m,n)\), and the exact minimum support is
\[
n+m-\gcd(n,m)
\]
[2211.16058]. This is a literal theory of one function compatible with two arithmetic-progression translation systems.

Geometric Diophantine problems furnish yet another model. On a conic
\[
\mathcal C\subset \mathbb P^2
\]
with a linear rational map \(\ell:\mathcal C\to \mathbb P^1\), one studies triples \(P_1,P_2,P_3\in \mathcal C(K)\) such that
\[
\ell(P_1)=t_0-\delta,\qquad \ell(P_2)=t_0,\qquad \ell(P_3)=t_0+\delta.
\]
For fixed \(t_0\), the admissible centered \(3\)-term progressions are parametrized by rational points on the elliptic curve
\[
\mathcal E_k:\quad Y^2+4XY+4kY=X^3+kX^2,
\]
with
\[
\delta= -\frac{\operatorname{Disc}(t_0)}{\operatorname{Disc}'(t_0)} \cdot \frac{4XY}{Y^2+2XY+kX^2}.
\]
Thus multiple centered progressions with the same middle term correspond to multiple rational points on one elliptic curve [1210.6612]. In analytic number theory, a related two-progression coupling appears in Goldbach problems where \(m\equiv a_1\pmod{q_1}\) and \(l\equiv a_2\pmod{q_2}\); the average representation count has main term
\[
\frac{X^2}{2\varphi(q_1)\varphi(q_2)}
\]
together with single-modulus zero terms and an exceptional-zero interaction term [2407.18266].

## 5. Unions, packings, and algorithmic decomposition

Interlaced progressions also arise when several APs are placed inside one ambient set. A classical extremal question asks how small the union of many progressions can be when the common differences are distinct. If \(u_\ell(n)\) denotes the minimum size of the union of \(n\) arithmetic progressions, each of length \(\ell\), with pairwise distinct differences, then for every \(\varepsilon>0\) there is \(c_1(\varepsilon)>0\) such that
\[
u_\ell(n)\ge c_1(\varepsilon)\begin{cases}
\ell^2 & \text{for }\ell<n^{1/2-\varepsilon},\\[1mm]
n^{1-\varepsilon}\ell & \text{for }n^{1/2-\varepsilon}\le \ell<n^{1-\varepsilon},\\[1mm]
n^{2-\varepsilon} & \text{for }n^{1-\varepsilon}<\ell.
\end{cases}
\]
In particular, the union of any \(n\) arithmetic progressions, each of length \(n\), with pairwise distinct differences has size at least \(c(\varepsilon)n^{2-\varepsilon}\). Yet there are \(n\) such progressions with union size \(o(n^2)\), realized by the multiplication-table construction \(\{j,2j,\dots,nj\}\) [1310.4348]. The resulting picture is one of heavy overlap without total collapse.

A closely related but disjoint version is studied in packing problems. If
\[
A_d=D_d\cap[n],\qquad D_d=\{d,2d,3d,\dots\},
\]
and \(m(\mathcal F)\) is the minimum interval length containing pairwise disjoint shifted copies of a family \(\mathcal F\), then for the full bounded-diameter family one has
\[
m(n)=\Theta\!\left(\frac{n^{3/2}}{\ln n}\right),
\]
more precisely
\[
\left(\frac{4}{3}-o(1)\right)\frac{n^{3/2}}{\ln n}\le m(n)\le \left(\frac{5}{3}+o(1)\right)\frac{n^{3/2}}{\ln n}.
\]
For the equal-size family
\[
B_d=D_d\cap[nd],
\]
one has
\[
M(n)=\Theta\!\left(\frac{n^3}{\ln n}\right),
\]
with bounds
\[
\left(\frac16-o(1)\right)\frac{n^3}{\ln n}\le M(n)\le \left(0.526+o(1)\right)\frac{n^3}{\ln n},
\]
and for fixed \(n\) and sufficiently large \(k\),
\[
M_k(n)<3kn
\]
[2603.02786]. These results treat interlacing as the problem of arranging many shifted AP-patterns so that their occupied points are disjoint while their containing intervals overlap maximally.

The same theme has an algorithmic counterpart. In Cover by Arithmetic Progressions (CAP), one asks whether a finite set \(X\subseteq\mathbb Z\) is the union of \(k\) arithmetic progressions contained in \(X\); in Exact Cover by Arithmetic Progressions (XCAP), those progressions must be disjoint. CAP admits a
\[
2^{O(k^2)}\,poly(n)
\]
algorithm, XCAP admits a
\[
2^{O(k^3)}\,poly(n)
\]
algorithm, and CAP parameterized below the guarantee \(|X|/2\) admits a
\[
2^{O(k)}n^{O(1)}
\]
algorithm. Over \(\mathbb Z_p\), the modular versions of Cover and Exact Cover by Arithmetic Progressions are strongly NP-complete [2312.06393]. In CAP, overlap is allowed; in XCAP, interlacing is constrained by exact ownership of each point.

## 6. Sparse, random, and nonlinear ambient sets

The theme also appears inside highly structured ambient sets. Powerful numbers are one example. A positive integer is powerful if each prime divisor occurs to exponent at least \(2\), equivalently \(n=a^2b^3\) with \(b\) squarefree. These numbers contain arithmetic progressions of every finite length. For \(3\)-term progressions
\[
N,\ N+d,\ N+2d
\]
of powerful numbers, the Pell-type construction yields infinitely many examples with
\[
d\le 3\sqrt N,
\]
so \(\theta_3\le 1/2\), while the \(abc\)-conjecture implies
\[
d\gg_\varepsilon N^{1/2-\varepsilon},
\]
hence \(\theta_3=1/2\) under \(abc\) [2210.00281]. This identifies a square-root compression threshold for \(3\)-term progressions inside a very sparse multiplicative set.

In random subsets \([n]_p\), the interaction of arithmetic progressions becomes probabilistic. If \(X_\ell\) counts \(\ell\)-term APs in \([n]_p\), then for a pair \((X_\ell,X_{\ell'})\) the covariance is asymptotically governed by two dominant overlap types: loose pairs with \(|T\cap T'|=1\), and overlap pairs with \(T'\subseteq T\). The regime is controlled by
\[
\psi_\ell=np^{\ell-1}\ell.
\]
Depending on the asymptotics of \(\psi_\ell\), the normalized counts are asymptotically uncorrelated, nontrivially correlated, or perfectly correlated [1902.04176]. This is a statistical theory of interacting APs indexed by intersection structure rather than by deterministic geometry.

Nonlinear sequences provide another setting in which several arithmetic strands coexist. For slightly curved sequences, if
\[
a(n)=f(n)+o\!\big((\log\log n)^{1/c_k}\big),\qquad f''(x)=O(1/x^\alpha),
\]
then the graph \(\{(n,a(n))\}\) contains a \(k\)-term arithmetic progression. If \(A\subseteq\mathbb N\) has positive upper Banach density and the error is \(O(1)\), then the graph of \(\{a(n)\}_{n\in A}\) contains arbitrarily long APs; in particular, this applies to \(\{\lfloor n^a\rfloor\}_{n\in A}\) for \(1\le a<2\). By contrast, for every \(a\ge 2\), the graph of \(\{\lfloor n^a\rfloor\}\) contains no \(3\)-term AP [1807.06971]. In modified Ulam sequences \(V(2,n)\), when \(n\ge 5\) is odd and \(n-1\) is not a power of \(2\), the sequence is eventually a finite union of arithmetic progressions, and the parity recursion
\[
1(u)=1(u-2)+1(u-2n)
\]
together with the Pascal-mod-\(2\) / Sierpiński-triangle structure explains why the tail can be viewed as several interlaced arithmetic strands [1804.09594]. A numeration-theoretic analogue appears for Zeckendorf-Niven and Lucas-Niven numbers: every arithmetic progression contains infinitely many of them, but every arithmetic progression also contains infinitely many subsequences of \(\ell\) consecutive non-Zeckendorf-Niven terms for every \(\ell\ge 1\) [2606.24006].

Taken together, these results show that interlaced arithmetic progressions are best understood not as a single theorem class but as a family of higher-order arithmetic configurations. The dominant formal paradigms are synchronized index/value progressions, largeness of progression-parameter sets, coupled AP systems arising from dynamics or tilings, and extremal or algorithmic questions about unions and packings. The main open directions in the cited literature remain correspondingly varied: existence of double \(3\)-term progressions in every bounded-gap sequence, sharper bounds for longer progressions in sparse sets such as the powerful numbers, exact constants in AP packing problems, and more explicit structural theories of overlap beyond parameter-set largeness.

Source: https://www.emergentmind.com/topics/interlaced-arithmetic-progressions