ConceptComplete

Modern Sieve Applications

Modern sieve theory combines classical upper/lower bound sieves with additional input from algebraic geometry, harmonic analysis, and L-functions to achieve results beyond the parity barrier. The breakthroughs on gaps between primes exemplify this synthesis.


GPY Sieve and Bounded Gaps

Definition7.5GPY Method

The Goldston-Pintz-Yildirim (GPY) method uses a modified Selberg sieve with weights concentrated on integers nn such that several translates n+h1,…,n+hkn + h_1, \ldots, n + h_k are simultaneously prime. The key quantity is Sk=βˆ‘x≀n≀2xw(n)(βˆ‘i=1k1n+hiΒ primeβˆ’m)S_k = \sum_{x \leq n \leq 2x} w(n) \left(\sum_{i=1}^k \mathbf{1}_{n+h_i \text{ prime}} - m\right) where w(n)w(n) is a smooth Selberg-type weight and mm is a threshold. If Sk>0S_k > 0, then some tuple (n+h1,…,n+hk)(n+h_1, \ldots, n+h_k) contains at least m+1m + 1 primes.

ExampleMaynard-Tao Theorem on Bounded Gaps

Theorem (Maynard, 2013; Tao, polymath8b): lim inf⁑nβ†’βˆž(pn+1βˆ’pn)≀246\liminf_{n\to\infty}(p_{n+1} - p_n) \leq 246. More generally, lim inf⁑(pn+mβˆ’pn)<∞\liminf(p_{n+m} - p_n) < \infty for every fixed mm, i.e., there are bounded gaps containing m+1m+1 primes. The method uses multidimensional Selberg sieve weights w(n)=(βˆ‘d1∣n+h1,…,dk∣n+hkdi≀RΞ»d1,…,dk)2w(n) = (\sum_{\substack{d_1 | n+h_1, \ldots, d_k|n+h_k \\ d_i \leq R}} \lambda_{d_1,\ldots,d_k})^2 optimized over a smooth function of kk variables.


Sieve and Algebraic Input

Definition7.6Friedlander-Iwaniec and Algebraic Sieves

Friedlander-Iwaniec (1998): There are infinitely many primes of the form a2+b4a^2 + b^4. The proof combines the half-dimensional sieve (sieving a thin sequence with density x3/4x^{3/4}) with bilinear form estimates using algebraic geometry over Z[i]\mathbb{Z}[i]. This broke the parity barrier for a specific thin sequence by exploiting the algebraic structure of Gaussian integers.

RemarkGreen-Tao and Primes in Arithmetic Progressions

The Green-Tao theorem (2004): the primes contain arbitrarily long arithmetic progressions. The proof uses SzemerΓ©di's theorem relativized to a pseudorandom set, with the pseudorandomness verified via sieve theory (the WW-trick and Goldston-Yildirim type estimates). This shows that sieve methods, combined with additive combinatorics, can prove results far beyond the classical sieve framework.


Numerical Sieve Applications

ExampleChen's Theorem

Chen's theorem (1966/1973): Every sufficiently large even number NN is the sum of a prime and a number that is either prime or a product of two primes (P2P_2). The proof uses the weighted sieve (Rosser-Iwaniec linear sieve) with Bombieri-Vinogradov as input: ∣{p≀N:Nβˆ’p∈P2}βˆ£β‰«N(log⁑N)2|\{p \leq N : N - p \in P_2\}| \gg \frac{N}{(\log N)^2}. The sieve lower bound barely beats zero thanks to the level of distribution x1/2βˆ’Ξ΅x^{1/2-\varepsilon} from BV.