For Roth's theorem on Diophantine approximation of algebraic numbers, see Roth's theorem.

Roth's theorem on arithmetic progressions is a result in additive combinatorics concerning the existence of arithmetic progressions in subsets of the natural numbers. It was first proven by Klaus Roth in 1953.[1] Roth's theorem is a special case of Szemerédi's theorem for the case k = 3.

Statement

A subset A of the natural numbers is said to have positive upper density if

\limsup_{n\to\infty}\frac{|A\cap \{1, 2, 3, \dotsc, n\}|}{n} > 0.
Roth's theorem on arithmetic progressions (infinite version): A subset of the natural numbers with positive upper density contains a 3-term arithmetic progression.

An alternate, more qualitative, formulation of the theorem is concerned with the maximum size of a Salem–Spencer set which is a subset of [N] = \{1, \dots, N\}. Let r_3([N]) be the size of the largest subset of [N] which contains no 3-term arithmetic progression.

Roth's theorem on arithmetic progressions (finitary version): r_3([N]) = o(N).

Improving upper and lower bounds on r_3([N]) is still an open research problem.

History

The first result in this direction was Van der Waerden's theorem in 1927, which states that for sufficiently large N, coloring the integers 1, \dots, n with r colors will result in a k term arithmetic progression.[2]

Later on in 1936 Erdős and Turán conjectured a much stronger result that any subset of the integers with positive density contains arbitrarily long arithmetic progressions. In 1942, Raphaël Salem and Donald C. Spencer provided a construction of a 3-AP-free set (i.e. a set with no 3-term arithmetic progressions) of size \frac{N}{e^{O(\log N/\log \log N)}},[3] disproving an additional conjecture of Erdős and Turán that r_3([N]) = N^{1 - \delta} for some \delta > 0.[4]

In 1953, Roth partially resolved the initial conjecture by proving they must contain an arithmetic progression of length 3 using Fourier analytic methods. Eventually, in 1975, Szemerédi proved Szemerédi's theorem using combinatorial techniques, resolving the original conjecture in full.

Proof techniques

The original proof given by Roth used Fourier analytic methods. Later on another proof was given using Szemerédi's regularity lemma.

Proof sketch via Fourier analysis

In 1953, Roth used Fourier analysis to prove an upper bound of r_3([N]) = O\left(\frac{N}{\log \log N}\right). Below is a sketch of this proof.

Define the Fourier transform of a function f : \mathbb{Z} \rightarrow \mathbb{C} to be the function \widehat{f} : [0, 1) \rightarrow \mathbb{C} satisfying

\widehat{f}(\theta) = \sum_{x \in \mathbb{Z}}f(x)e(-x\theta),

where e(t) = e^{2\pi i t}.

Let A be a 3-AP-free subset of \{1, \dots, N\}. The proof proceeds in 3 steps.

  1. Show that a A admits a large Fourier coefficient.
  2. Deduce that there exists a sub-progression of \{1, \dots, N\} such that A has a density increment when restricted to this subprogression.
  3. Iterate Step 2 to obtain an upper bound on |A|.

Step 1

For functions, f, g, h : \mathbb{Z} \rightarrow \mathbb{C}, define

\Lambda(f, g, h) = \sum_{x, y \in \mathbb{Z}} f(x)g(x + y)h(x + 2y)
Counting Lemma Let f, g : \mathbb{Z} \rightarrow \mathbb{C} satisfy \sum_{n \in \mathbb{Z}}|f(n)|^2, \sum_{n \in \mathbb{Z}}|g(n)|^2 \le M. Define \Lambda_3(f) = \Lambda(f,f,f). Then |\Lambda_3(f) - \Lambda_3(g)| \le 3M\|\widehat{f - g}\|_\infty.

The counting lemma tells us that if the Fourier Transforms of f and g are "close", then the number of 3-term arithmetic progressions between the two should also be "close." Let \alpha = |A|/N be the density of A. Define the functions f = \mathbf{1}_{A} (i.e the indicator function of A), and g = \alpha \cdot \mathbf{1}_{[N]}. Step 1 can then be deduced by applying the Counting Lemma to f and g, which tells us that there exists some \theta such that

\left|\sum_{n=1}^N (1_A - \alpha)(n)e(\theta n) \right| \ge \frac{\alpha^2}{10}N.

Step 2

Given the \theta from step 1, we first show that it's possible to split up [N] into relatively large subprogressions such that the character x \mapsto e(x\theta) is roughly constant on each subprogression.

Lemma 1: Let 0 < \eta < 1, \theta \in \mathbb{R}. Assume that N > C\eta^{-6} for a universal constant C. Then it is possible to partition [N] into arithmetic progressions P_i with length N^{1/3} \le |P_i| \le 2N^{1/3} such that \sup_{x, y \in P_i}|e(x\theta) - e(y\theta)| < \eta for all i.

Next, we apply Lemma 1 to obtain a partition into subprogressions. We then use the fact that \theta produced a large coefficient in step 1 to show that one of these subprogressions must have a density increment:

Lemma 2: Let A be a 3-AP-free subset of [N], with |A| = \alpha N and N > C\alpha^{-12}. Then, there exists a sub progression P \subset [N] such that |P| \ge N^{1/3} and |A \cap P| \ge (\alpha + \alpha^2/40)|P|.

Step 3

We now iterate step 2. Let a_t be the density of A after the tth iteration. We have that \alpha_0 = \alpha, and \alpha_{t + 1} \ge \alpha + \alpha^2/40. First, see that \alpha doubles (i.e. reach T such that \alpha_T \ge 2\alpha_0) after at most 40/\alpha + 1 steps. We double \alpha again (i.e reach \alpha_T \ge 4\alpha_0) after at most 20/\alpha + 1 steps. Since \alpha \le 1, this process must terminate after at most O(1/\alpha) steps.

Let N_t be the size of our current progression after t iterations. By Lemma 2, we can always continue the process whenever N_t \ge C\alpha_t^{-12}, and thus when the process terminates we have that N_t \le C\alpha_t^{-12} \le C\alpha^{-12}. Also, note that when we pass to a subprogression, the size of our set decreases by a cube root. Therefore

N \le N_t^{3^t} \le (C\alpha^{-12})^{3^{O(1/\alpha)}} = e^{e^{O(1/\alpha)}}.

Therefore \alpha = O(1/\log \log N), so |A| = O \left(\frac{N}{\log \log N}\right), as desired. \blacksquare

Unfortunately, this technique does not generalize directly to larger arithmetic progressions to prove Szemerédi's theorem. An extension of this proof eluded mathematicians for decades until 1998, when Timothy Gowers developed the field of higher-order Fourier analysis specifically to generalize the above proof to prove Szemerédi's theorem.[5]

Proof sketch via graph regularity

From the Szemerédi regularity lemma and counting lemma one obtains the graph removal lemma, and as a corollary the following:

Diamond-free Lemma. Any graph G on n vertices in which every edge lies in precisely one unique triangle has o(n^2) edges.

Application to Roth’s theorem

First note that a set A\subseteq\{1,\dots,N\} has a 3-term arithmetic progression iff its reduction modulo M:=2N+1 does.

Fix A\subseteq \{1,\dots,N\} with no 3-term AP. Construct a tripartite graph G with parts X,Y,Z, each a copy of \mathbb{Z}/M\mathbb{Z}. Add edges as follows:

  • x\in X to y\in Y if y-x\in A;
  • y\in Y to z\in Z if z-y\in A;
  • z\in Z to x\in X if (x-z)/2\in A (since 2 is invertible mod M).

If x,y,z form a triangle, then y-x,\ \frac{x-z}{2},\ z-y\in A. These three numbers form an arithmetic progression in that order with step (x+z-2y)/2. Because A has no nontrivial 3-term AP, they are equal, so (x+z-2y)/2=0. Therefore, if we fix any two vertices of G connected by an edge, every triangle extending this edge must satisfy x+z-2y=0, yielding a unique choice of the third vertex. Furthermore, such a choice guarantees the existence of the other two edges in that triangle. Thus, the diamond-free lemma applies, and e(G)=o((3M)^2)=o(N^2).

Notice now, that each element a \in A contributes exactly one edge for each vertex in a part: for instance, for every x \in X, there is exactly one y \in Y such that y - x \equiv a \pmod{M}, so a gives exactly M edges between X and Y. The same holds for the other two pairs, hence e(G)=3|A|M. Consequently,

|A| = \frac{e(G)}{3M} = \frac{o(N^2)}{3(2N+1)} = o(N)

proving Roth’s theorem.

Extensions and generalizations

Szemerédi's theorem resolved the original conjecture and generalized Roth's theorem to arithmetic progressions of arbitrary length. Since then it has been extended in multiple fashions to create new and interesting results.

Furstenberg and Katznelson[6] used ergodic theory to prove a multidimensional version and Leibman and Bergelson[7] extended it to polynomial progressions as well. Most recently, Green and Tao proved the Green–Tao theorem which says that the prime numbers contain arbitrarily long arithmetic progressions. Since the prime numbers are a subset of density 0, they introduced a "relative" Szemerédi theorem which applies to subsets with density 0 that satisfy certain pseudorandomness conditions. Later on Conlon, Fox, and Zhao[8][9] strengthened this theorem by weakening the necessary pseudorandomness condition. In 2020, Bloom and Sisask[10] proved that any set A such that \sum_{n \in A} \frac{1}{n} diverges must contain arithmetic progressions of length 3; this is the first non-trivial case of another conjecture of Erdős postulating that any such set must in fact contain arbitrarily long arithmetic progressions.

Improving bounds

See also: Szemerédi's theorem#Quantitative bounds, Erdős conjecture on arithmetic progressions#Progress and related results and Salem–Spencer set#Size

There has also been work done on improving the bound in Roth's theorem. The bound from the original proof of Roth's theorem showed that

r_3([N]) \leq c\cdot\frac{N}{\log\log N}

for some constant c. Over the years this bound has been continually lowered by Szemerédi,[11] Heath-Brown,[12] Bourgain,[13][14] and Sanders.[15][16] The current (July 2020) best bound is due to Bloom and Sisask[10] who have shown the existence of an absolute constant c>0 such that

r_3([N]) \leq \frac{N}{(\log N)^{1+c}}.

In February 2023 a preprint[17][18] (later published[19]) by Kelley and Meka gave a new bound of:

r_3([N]) \leq 2^{-\Omega((\log N)^{1/12})} \cdot N.

Four days later, Bloom and Sisask published a preprint giving an exposition of the result[20] (later published[21]), simplifying the argument and yielding some additional applications. Several months later, Bloom and Sisask obtained a further improvement to r_3([N]) \leq \exp(-c(\log N)^{1/9})N, and stated (without proof) that their techniques can be used to show r_3([N]) \leq \exp(-c(\log N)^{5/41})N.[22]

In a 2026 preprint,[23] Raghavan reported further improved bounds, proving:

|A| \leq \exp(-c\log(N)^{1/6}\log\log(N)^{-1/6})N.

There has also been work done on the other end, constructing the largest set with no 3-term arithmetic progressions. The best construction has barely been improved since 1946 when Behrend[24] improved on the initial construction by Salem and Spencer and proved

r_3([N]) \geq N\exp(-c\sqrt{\log N}).

Due to no improvements in over 70 years, it is conjectured that Behrend's set is asymptotically very close in size to the largest possible set with no 3-term progressions.[10] If correct, the Kelley-Meka bound will prove this conjecture.

Roth's theorem in finite fields

As a variation, we can consider the analogous problem over finite fields. Consider the finite field \mathbb{F}_3^n, and let r_3(\mathbb{F}_3^n) be the size of the largest subset of \mathbb{F}_3^n which contains no 3-term arithmetic progression. This problem is actually equivalent to the cap set problem, which asks for the largest subset of \mathbb{F}_3^n such that no 3 points lie on a line. The cap set problem can be seen as a generalization of the card game Set.

In 1982, Brown and Buhler[25] were the first to show that r_3(\mathbb{F}_3^n) = o(3^n). In 1995, Roy Mesuhlam[26] used a similar technique to the Fourier-analytic proof of Roth's theorem to show that r_3(\mathbb{F}_3^n) = O\left(\frac{3^n}{n}\right). This bound was improved to O(3^n/n^{1 + \epsilon}) in 2012 by Bateman and Katz.[27]

In 2016, Ernie Croot, Vsevolod Lev, Péter Pál Pach, Jordan Ellenberg and Dion Gijswijt developed a new technique based on the polynomial method to prove that r_3(\mathbb{F}_3^n) = O(2.756^n).[28][29][30]

The best known lower bound is 2.2202^{n}, discovered in December 2023 by Google DeepMind researchers using a large language model (LLM).[31]

Another generalization of Roth's theorem shows that for positive density subsets, there not only exists a 3-term arithmetic progression, but that there exist many 3-APs all with the same common difference.

Roth's theorem with popular differences: For all \epsilon > 0, there exists some n_0 = n_0(\epsilon) such that for every n > n_0 and A \subset \mathbb{F}_3^n with |A| = \alpha3^n, there exists some y \neq 0 such that |\{x : x, x + y, x + 2y \in A\}| \ge (\alpha^3 - \epsilon)3^n.

If A is chosen randomly from \mathbb{F}_3^n, then we would expect there to be \alpha^33^n progressions for each value of y. The popular differences theorem thus states that for each |A| with positive density, there is some y such that the number of 3-APs with common difference y is close to what we would expect.

This theorem was first proven by Green in 2005,[32] who gave a bound of n_0 = \text{tow}((1/\epsilon)^{O(1)}), where \text{tow} is the tower function. In 2019, Fox and Pham recently improved the bound to n_0 = \text{tow}(O(\log\frac{1}{\epsilon})).[33]

A corresponding statement is also true in \mathbb{Z} for both 3-APs and 4-APs.[34] However, the claim has been shown to be false for 5-APs.[35]

References

  1. ^ Roth, Klaus (1953). "On certain sets of integers". Journal of the London Mathematical Society. 28 (1): 104–109. doi:10.1112/jlms/s1-28.1.104
  2. ^ van der Waerden, B. L. (1927). "Beweis einer Baudetschen Vermutung". Nieuw. Arch. Wisk.. 15: 212–216.
  3. ^ Salem, Raphaël & Spencer, Donald C. (1942). "On sets of integers which contain no three terms in arithmetical progression". Proceedings of the National Academy of Sciences of the United States of America. 28 (12): 561–563. Bibcode:1942PNAS...28..561S. doi:10.1073/pnas.28.12.561. MR 0007405. PMC 1078539. PMID 16588588
  4. ^ Erdös, Paul & Turán, Paul (1936). "On Some Sequences of Integers". Journal of the London Mathematical Society. 4 (4): 261–264. doi:10.1112/jlms/s1-11.4.261. MR 1574918.
  5. ^ Gowers, W. T. (1998). "A new proof of Szemerédi's theorem for arithmetic progressions of length four". Geometric and Functional Analysis. 8 (3): 529–551. doi:10.1007/s000390050065
  6. ^ Furstenberg, Hillel & Katznelson, Yitzhak (1978). "An ergodic Szemerédi theorem for commuting transformations". Journal d'Analyse Mathématique. 38 (1): 275–291. doi:10.1007/BF02790016. MR 531279. S2CID 123386017
  7. ^ Bergelson, Vitaly & Leibman, Alexander (1996). "Polynomial extensions of van der Waerden's and Szemerédi's theorems". Journal of the American Mathematical Society. 9 (3): 725–753. doi:10.1090/S0894-0347-96-00194-4. MR 1325795.
  8. ^ Conlon, David; Fox, Jacob; Zhao, Yufei (2015). "A relative Szemerédi theorem". Geometric and Functional Analysis. 25 (3): 733–762. arXiv:1305.5440. doi:10.1007/s00039-015-0324-9. MR 3361771.
  9. ^ Zhao, Yufei (2014). "An arithmetic transference proof of a relative Szemerédi theorem". Mathematical Proceedings of the Cambridge Philosophical Society. 156 (2): 255–261. arXiv:1307.4959. Bibcode:2014MPCPS.156..255Z. doi:10.1017/S0305004113000662. MR 3177868. S2CID 119673319
  10. ^ Thomas F. Bloom, Olof Sisask, Breaking the logarithmic barrier in Roth's theorem on arithmetic progressions, arXiv:2007.03528, 2020
  11. ^ Szemerédi, Endre (1990). "Integer sets containing no arithmetic progressions". Acta Mathematica Hungarica. 56 (1–2): 155–158. doi:10.1007/BF01903717. MR 1100788.
  12. ^ Heath-Brown, Roger (1987). "Integer sets containing no arithmetic progressions". Journal of the London Mathematical Society. 35 (3): 385–394. doi:10.1112/jlms/s2-35.3.385. MR 889362.
  13. ^ Bourgain, Jean (1999). "On triples in arithmetic progression". Geometric and Functional Analysis. 9 (5): 968–984. doi:10.1007/s000390050105. MR 1726234. S2CID 392820
  14. ^ Bourgain, Jean (2008). "Roth's theorem on progressions revisited". Journal d'Analyse Mathématique. 104 (1): 155–192. doi:10.1007/s11854-008-0020-x. MR 2403433. S2CID 16985451
  15. ^ Sanders, Tom (2012). "On certain other sets of integers". Annals of Mathematics. 185 (1): 53–82. arXiv:1007.5444. doi:10.1007/s11854-012-0003-9. MR 2892617. S2CID 119727492
  16. ^ Sanders, Tom (2011). "On Roth's theorem on progressions". Annals of Mathematics. 174 (1): 619–636. arXiv:1011.0104. doi:10.4007/annals.2011.174.1.20. MR 2811612. S2CID 53331882
  17. ^ Kelley, Zander & Meka, Raghu (2023-02-10). "Strong Bounds for 3-Progressions". arXiv:2302.05537
  18. ^ Sloman, Leila (2023-03-21). "Surprise Computer Science Proof Stuns Mathematicians". Quanta Magazine
  19. ^ Kelley, Zander & Meka, Raghu (2023-11-06). "Strong Bounds for 3-Progressions". 2023 IEEE 64th Annual Symposium on Foundations of Computer Science (FOCS). IEEE. pp. 933–973. arXiv:2302.05537. doi:10.1109/FOCS57990.2023.00059. ISBN 979-8-3503-1894-4.
  20. ^ Bloom, Thomas F. & Sisask, Olof (2023-02-14). "The Kelley–Meka bounds for sets free of three-term arithmetic progressions". Essential Number Theory. 2: 15–44. arXiv:2302.07211. doi:10.2140/ent.2023.2.15
  21. ^ Bloom, Thomas F. & Sisask, Olof (2023-12-31). "The Kelley–Meka bounds for sets free of three-term arithmetic progressions". Essential Number Theory. 2 (1): 15–44. arXiv:2302.07211. doi:10.2140/ent.2023.2.15. ISSN 2834-4634
  22. ^ Bloom, Thomas F. & Sisask, Olof (2023-09-05). "An improvement to the Kelley-Meka bounds on three-term arithmetic progressions". arXiv:2309.02353
  23. ^ Raghavan, Rushil (2026-05-15), "Improved Bounds for 3-Progressions", arXiv, doi:10.48550/arXiv.2603.27045. arXiv:2603.27045, retrieved 2026-05-19
  24. ^ Behrend, F. A. (1946). "On sets of integers which contain no three terms in arithmetical progression". Proceedings of the National Academy of Sciences of the United States of America. 32 (12): 331–332. Bibcode:1946PNAS...32..331B. doi:10.1073/pnas.32.12.331. PMC 1078964. PMID 16578230
  25. ^ Brown, T. C. & Buhler, J. P. (1982). "A density version of a geometric Ramsey theorem". Journal of Combinatorial Theory. 32 (1): 20–34. Series A. doi:10.1016/0097-3165(82)90062-0
  26. ^ Mesuhlam, Roy (1995). "On subsets of finite abelian groups with no 3-term arithmetic progressions". Journal of Combinatorial Theory. 71 (1): 168–172. Series A. doi:10.1016/0097-3165(95)90024-1
  27. ^ Bateman, M. & Katz, N. (2012). "New bounds on cap sets". Journal of the American Mathematical Society. 25 (2): 585–613. arXiv:1101.5851. doi:10.1090/S0894-0347-2011-00725-X. hdl:2022/19057
  28. ^ Ellenberg, Jordan S. & Gijswijt, Dion (2016). "On large subsets of \mathbb{F}_q^n with no three-term arithmetic progression". Annals of Mathematics, Second Series. 185 (1): 339–343. arXiv:1605.09223. doi:10.4007/annals.2017.185.1.8. S2CID 119683140
  29. ^ Croot, Ernie; Lev, Vsevolod F.; Pach, Péter Pál (2017). "Progression-free sets in \mathbb{Z}_{4}^{n} are exponentially small". Annals of Mathematics. 185 (1): 331–337. 2nd series. arXiv:1605.01506. doi:10.4007/annals.2017.185.1.7
  30. ^ Klarreich, Erica (May 31, 2016). "Simple Set Game Proof Stuns Mathematicians". Quanta
  31. ^ Romera-Paredes, Bernardino; Barekatain, Mohammadamin; Novikov, Alexander; Balog, Matej; Kumar, M. Pawan; Dupont, Emilien; Ruiz, Francisco J. R.; Ellenberg, Jordan S.; Wang, Pengming; Fawzi, Omar; Kohli, Pushmeet; Fawzi, Alhussein (2023-12-14). "Mathematical discoveries from program search with large language models". Nature. 625 (7995): 468–475. doi:10.1038/s41586-023-06924-6. ISSN 1476-4687. PMC 10794145. PMID 38096900
  32. ^ Green, Ben (2005). "A Szemerédi-type regularity lemma in abelian groups, with applications". Geometric and Functional Analysis. 15 (2): 340–376. doi:10.1007/s00039-005-0509-8. MR 2153903.
  33. ^ Fox, Jacob & Pham, Huy Tuan (April 2021). "Popular progression differences in vector spaces". International Mathematics Research Notices. 2021 (7): 5261–5289. arXiv:1708.08482. Bibcode:2017arXiv170808482F. doi:10.1093/imrn/rny240
  34. ^ Green, Ben & Tao, Terrence (2010). "An Arithmetic Regularity Lemma, an Associated Counting Lemma, and Applications". An Irregular Mind. Vol. 21. Bolyai Society Mathematical Studies. Bolyai Society Mathematical Studies. pp. 261–334. arXiv:1002.2028. Bibcode:2010arXiv1002.2028G. doi:10.1007/978-3-642-14444-8_7. ISBN 978-3-642-14443-1. S2CID 115174575
  35. ^ Bergelson, Vitaly; Host, Bernard; Kra, Bryna (2005). "Multiple recurrence and nilsequences. With an appendix by Imre Ruzsa". Inventiones Mathematicae. 160 (2): 261–303. doi:10.1007/s00222-004-0428-6. S2CID 1380361