A New Way to Count Primes: The Green–Sawhney Proof Explained
Take two prime numbers, p and q. Square the first, add four times the square of the second, and see what you get. With p = 5 and q = 2 you get 25 + 16 = 41, which is prime. With p = 5 and q = 3 you get 61, also prime. With p = 7 and q = 2 you get 65, which is not.
The question is whether that recipe keeps producing primes forever. In late 2024 Ben Green of Oxford and Mehtaab Sawhney of Columbia proved that it does. Their paper, "Primes of the form p² + nq²", settled a problem posed by John Friedlander and Henryk Iwaniec. Quanta Magazine's report on the result tells the story of how the collaboration came about. This piece is about why the problem is hard and what was new in the solution.

Illustration: a spiral of numbers where primes, shown brighter, form unexpected patterns.
Why "infinitely many" is the hard part
Euclid proved around 300 BCE that the primes never run out. But once you ask for primes of a particular shape, the problem changes completely. It's easy to show that primes of the form 4k + 1 are infinite. Nobody knows whether there are infinitely many primes of the form n² + 1, even though a computer finds them everywhere it looks.
The difficulty is thinness. Among numbers up to a million, about 78,000 are prime. Numbers of the form n² + 1 are far rarer, about a thousand of them below a million. The rarer the set, the harder it is to show that primes don't eventually vanish from it. The standard tools for counting primes, known as sieve methods, have a well-known blind spot here: on thin sets they cannot tell numbers with exactly one prime factor apart from numbers with two.
A short history of thin sets
In 1882 Heinrich Weber showed that for any fixed n, x² + ny² is prime infinitely often when x and y can be any whole numbers. Then the sets got thinner. In 1998 Friedlander and Iwaniec proved, in a famous paper, that infinitely many primes have the form a² + b⁴. In 2001 Roger Heath-Brown did the same for x³ + 2y³.
Requiring x and y to be primes themselves makes the set thinner again, and adds a second layer of difficulty: you are now counting primes built out of primes. Friedlander and Iwaniec conjectured that p² + 4q² still produces infinitely many. That is the case Green and Sawhney settled. They also proved the same for p² + nq² whenever n leaves a remainder of 0 or 4 when divided by 6, and they gave an estimate of how many such primes appear below any bound.
A quick check shows why p has to be odd. If p = 2, the result is 4 + 4q², which is even and so never prime. With p odd, p² + 4q² is always odd, and it is a sum of two squares, p² + (2q)². By a theorem of Fermat, an odd prime is a sum of two squares exactly when it leaves remainder 1 after dividing by 4. So every prime this recipe makes, such as 41, 61, 109, 137 and 149, has that property.
The detour through "rough" numbers
Instead of attacking primes p and q head on, Green and Sawhney first worked with rough numbers: numbers with no small prime factors. Rough numbers behave like primes in many statistical ways but are far easier to handle. The plan was to prove the result for rough numbers and then show that, for this particular problem, swapping in genuine primes changes nothing important.
That second step is where the work was. In the language of the field, the standard sieve machinery reduces the problem to estimating two kinds of sums, called Type I and Type II. For this problem, the Type II sums were the obstacle.
A tool borrowed from combinatorics
The key was the Gowers norm, a measure of how much hidden structure a set of numbers contains. Timothy Gowers introduced it around 2000 while giving a new proof of Szemerédi's theorem on arithmetic progressions. Green himself used Gowers norms with Terence Tao in 2004 to prove that the primes contain arithmetic progressions of every length.
Gowers norms were not expected to help with sieve problems like this one. Green and Sawhney showed that, in their setting, the troublesome sums could be controlled by a Gowers norm. Then they leaned on a 2018 theorem of Tao and Tamar Ziegler about polynomial patterns in the primes to show that the primes and the rough numbers look the same through that lens. Once that was done, the rough-number result carried over to actual primes.
Why mathematicians care
Nobody needs a list of primes of the form p² + 4q². The interest is in the method. Problems about primes in thin sets are some of the oldest open questions in number theory, from n² + 1 to the twin prime conjecture. A route that links sieve theory to Gowers norms gives researchers a new tool for those problems. Whether it will reach any of the famous ones is still unknown, but it has opened a door nobody had seen.
Sources
- arXiv:2410.04189, "Primes of the form $p^2 + nq^2$". https://arxiv.org/abs/2410.04189
- Quanta Magazine, "Mathematicians Uncover a New Way to Count Prime Numbers". https://www.quantamagazine.org/mathematicians-uncover-a-new-way-to-count-prime-numbers-20241211/
- Friedlander J, Iwaniec H, "The Polynomial X 2 + Y 4 Captures Its Primes", The Annals of Mathematics 148 (1998): 945. https://doi.org/10.2307/121034
Comments (0)
Leave a Reply
Comments are moderated. Please read our Comment Policy before posting.
Log in to post a comment.