Green–Tao Theorem: Prime Arithmetic Progressions

• Green–Tao Theorem is a groundbreaking result establishing that the set of primes contains arbitrarily long arithmetic progressions through a transference approach.
• It employs analytic number theory and combinatorial methods by constructing pseudorandom majorants and verifying strict linear forms conditions.
• The theorem’s framework has been extended to include quantitative bounds, sparse set applications, and innovative generalizations in arithmetic combinatorics.

The Green–Tao theorem establishes that the set of prime numbers contains arbitrarily long arithmetic progressions. Since its original proof, significant advancements and generalizations have been developed, extending both the scope of the theorem and the quantitative aspects of its conclusions. The Green–Tao machinery, rooted in advances in combinatorics, analytic number theory, and higher-order Fourier analysis, continues to drive progress in arithmetic combinatorics and related fields.

1. The Green–Tao Theorem: Statement and Framework

The fundamental theorem asserts: for every $k \geq 3$, there exist arbitrarily long arithmetic progressions of prime numbers. Explicitly, for any $k$ and $N$, there exist primes $p, p+d, p+2d, \dotsc, p+(k-1)d$ with $p + (k-1)d$ as large as desired (Conlon et al., 2014).

The proof proceeds by a transference principle: it reduces the task to finding such progressions in suitable pseudorandom measure spaces, abstracting the crucial property of "random-like" sets that enable Szemerédi-theoretic methods. The high-level framework consists of:

• The construction of a pseudorandom majorant, typically via the $W$-trick and smoothed/truncated divisor sum weights.
• Verification that this majorant satisfies strong linear forms conditions.
• Application of a relative version of Szemerédi's theorem, ensuring that sufficiently dense subsets of the pseudorandom set also contain long progression patterns.

2. Relative Szemerédi Theorem and Transference

The technical core is the relative Szemerédi theorem for sparse sets. Conlon, Fox, and Zhao (Conlon et al., 2013, Zhao, 2013) demonstrated that it is sufficient for the majorant $\nu:\mathbb{Z}_N \to [0, \infty)$ to satisfy the $k$-linear forms condition:

$\E_{\mathbf{x}^{(0)},\mathbf{x}^{(1)} \in \mathbb{Z}_N^k} \prod_{j=1}^k \prod_{\omega \in \{0,1\}^{k}\setminus\{\mathbf{e}_j\}} \nu\Big(\sum_{i=1}^k (j-i)x_i^{(\omega_i)}\Big) = 1 + o(1),$

where the averages and products enumerate all the generalized patterns underlying $k$-term arithmetic progressions.

If $f: \mathbb{Z}_N \to [0, \nu(x)]$ with $\E f \geq \delta$, then

$\E_{x,d} \prod_{i=0}^{k-1} f(x+id) \geq c(k,\delta) - o_{k,\delta}(1),$

where $c(k, \delta)>0$ is explicit, and the error vanishes as $N$ tends to infinity.

The proof employs:

• Sparse weak regularity: approximation of $f$ by bounded-complexity functions.
• Sparse counting lemma: quantitative transfer of $k$-AP counts from the bounded-complexity approximation.
• Relative removal lemma: conversion of the relative count into concrete existence.

This structure removes the need for classical Gowers uniformity or "correlation" conditions, relying solely on the linear forms condition, which streamlines qualitative and quantitative arguments.

3. Construction and Verification of Pseudorandom Majorants

A pivotal technical component is the construction of majorants $\nu$ dominating the indicator function of (shifted, $W$-tricked) primes, and establishing their pseudorandomness as per the $k$-linear forms condition. The standard choice involves the truncated divisor sum majorants:

$$\Lambda_{W,R}(n) = \frac{\phi(W)}{W} \left(\sum_{d\mid n,\,d\leq R} \mu(d)\log\frac{R}{d}\right)^2,$$

where $W$ is a product of small primes, "sieving out" arithmetic structure modulo small moduli. The majorant is tailored: $$\nu(n) = \begin{cases} \Lambda_{W,R}(Wn + b)/\log R & \gcd(Wn+b, W)=1, \ 0 & \text{otherwise}. \end{cases}$$

Verifying that these satisfy the necessary conditions rests on:

• Sieve-theoretic divisor sum expansions and local correlation control (Conlon et al., 2013, Conlon et al., 2014).
• The Goldston–Yıldırım–Tao estimates for short divisor sums.
• Control of the $k$-linear forms for all systems of bounded-complexity affine-linear forms.

4. Extensions and Generalizations

The Green–Tao philosophy and methodology have been extended to a wide variety of settings:

• Sparse/quantitative bounds: Teräväinen–Wang (Teräväinen et al., 10 Mar 2026) established that the necessary density for a subset of the primes to guarantee nontrivial $k$-term APs is exponentially small in $\log\log\log N$ for $k\geq 5$, improving the earlier log-iterated quantitative bounds (Rimanic et al., 2017), which built on Conlon–Fox–Zhao and Henriot's sieve weights.
• Short intervals: Le Duc Hieu (Hieu, 5 Sep 2025) showed that for any fixed $\theta>17/30$, every interval $[x,x+x^\theta]$ with $x$ sufficiently large contains many $k$-term prime APs, via refinement of the transference method and application of uniform zero-density estimates for Dirichlet $L$-functions.
• Special prime sets: Primes defined by additional constraints, including Piatetski–Shapiro primes for $p = \lfloor n^{1/\gamma} \rfloor$ with $\gamma$ close to 1 (Li et al., 2019), primes of the form $x^2 + y^2 + 1$ (Sun et al., 2017), and other polynomial or quadratic sets (Bienvenu, 2016), have been shown to contain arbitrarily long arithmetic progressions under variable sparsity thresholds.
• Multiplicative functions: Similar transference techniques have been adapted to study correlations and configurations among values of bounded or divisor–bounded multiplicative functions, extending the Green–Tao type structure (Matthiesen, 2016).
• Positive density patterns in higher dimensions: Cook, Magyar, and Titichetrakun (Cook et al., 2013) obtain higher-dimensional analogues, establishing that positive density subsets of prime $d$-tuples contain affine copies of any fixed finite set.
• Polynomials and other number rings: Analogues in rings of integers of number fields, function fields, and coordinate rings of affine curves over finite fields have been achieved (Kai, 2021), with appropriate adaptation of $W$-tricks and hypergraph regularity arguments.

5. Quantitative and Density Thresholds

The effective density thresholds depend on $k$, the progression length, and on the pseudorandomness quality of the majorant. For the primes,

$p=|S|/N \approx 1/\log N = N^{-o(1)},$

one can take $p \geq N^{-c_k}$ for some $c_k = c(k, \delta) > 0$ (Conlon et al., 2013, Rimanic et al., 2017).

Recent advances (Teräväinen et al., 10 Mar 2026, Rimanic et al., 2017) have produced explicit, nearly best-possible quantitative bounds of the form

$$\delta \ll \exp\left(-(\log\log\log N)^{c_k}\right),$$

for the minimal density $\delta$ required in subsets of the primes to guarantee a $k$-term AP free set is small.

The size of the majorant's supremum, the complexity of the required linear forms, and the cut norms (or Gowers norms) control the quantitative error terms in the relative Szemerédi and transference steps. Improved bounds are obtained by leveraging quasipolynomial inverse theorems and optimized dense model arguments (Leng–Sah–Sawhney, Teräväinen–Tao).

6. Methodological Advances and New Directions

• Cut/Discrepancy norms vs. Gowers norms: The Conlon–Fox–Zhao approach develops discrepancy/cut-type norms as a coarser substitute for Gowers uniformity, sufficient for arithmetic progression problems, allowing more streamlined dense model and counting lemmas (Zhao, 2013).
• Hypergraph regularity and removal lemmas: For multidimensional generalizations and complex patterns (constellations), the hypergraph removal lemma, extended to weighted/pseudorandom cases, undergirds higher-dimensional and polynomial generalizations (Cook et al., 2013).
• Pseudorandomness regimes: The continuing goal is to push effective transference arguments to ever sparser pseudorandom models (e.g., for Piatetski–Shapiro, lower $\gamma$ regimes), or for sets with entropic or nonuniform randomness, as dictated by exponential sum bounds (Li et al., 2019).
• Arithmetic progressions in function fields and algebraic structures: Sophisticated versions of the $W$-trick and sieve methods are adapted to the intricacies of rings beyond $\mathbb{Z}$, such as function fields and smooth projective curves (Kai, 2021).

7. Impact, Open Questions, and Outlook

The Green–Tao theorem and its extensions have fundamentally transformed approaches to arithmetic combinatorics, with implications for random-like behavior in arithmetic sets, higher order Fourier analysis, and model-theoretic stability. Key open directions include:

• Further reducing density and sparsity thresholds required (e.g., pushing beyond current exponential decay, improving polynomial dependence in $k$).
• Transferring structural results to more complicated polynomial and non-linear configurations in the primes or in prime-like sets.
• Quantitative polynomial patterns, with density bounds parallel to those achieved for linear progressions.
• Deepening understanding of the "true" random-like nature required for such arithmetic patterns, including for general multiplicative or ergodic structures.

Continued synthesis of analytic, combinatorial, and ergodic techniques promises further advances, enhancing both the theoretical landscape and explicit quantitative control over arithmetic patterns in primes and related sets.