ConceptComplete

Goldbach and Ternary Goldbach Problems

The Goldbach conjecture (every even n>2n > 2 is a sum of two primes) remains open, but the ternary version (every odd n>5n > 5 is a sum of three primes) was proved by Helfgott in 2013, completing a line of research initiated by Vinogradov.


Vinogradov's Three-Primes Theorem

Definition8.3Three-Primes Theorem

Vinogradov's theorem (1937): Every sufficiently large odd integer NN is a sum of three primes. The number of representations is: r3(N)=βˆ‘p1+p2+p3=N(log⁑p1)(log⁑p2)(log⁑p3)=12S(N)N2+O(N2/(log⁑N)A)r_3(N) = \sum_{\substack{p_1 + p_2 + p_3 = N}} (\log p_1)(\log p_2)(\log p_3) = \frac{1}{2}\mathfrak{S}(N) N^2 + O(N^2/(\log N)^A) for any A>0A > 0, where S(N)=∏p∣N(pβˆ’1)2βˆ’1(pβˆ’1)3∏p∀N(1+1(pβˆ’1)3)>0\mathfrak{S}(N) = \prod_{p|N}\frac{(p-1)^2 - 1}{(p-1)^3}\prod_{p\nmid N}\left(1 + \frac{1}{(p-1)^3}\right) > 0 for odd N>5N > 5.

ExampleHelfgott's Complete Proof

Helfgott (2013) verified the ternary Goldbach conjecture for all odd N>5N > 5 by: (1) proving Vinogradov's theorem for N>1027N > 10^{27} with explicit error terms using detailed zero-free region computations; (2) verifying computationally that every odd integer 7≀N≀10277 \leq N \leq 10^{27} is a sum of three primes, using Oliveira e Silva's verification of binary Goldbach up to 4Γ—10184 \times 10^{18}.


The Binary Goldbach Problem

Definition8.4Binary Goldbach and Exceptional Set

The binary Goldbach conjecture: every even N>2N > 2 is a sum of two primes. The circle method gives: r2(N)=βˆ‘p1+p2=Nlog⁑p1log⁑p2=S2(N)N+E(N)r_2(N) = \sum_{p_1+p_2=N} \log p_1 \log p_2 = \mathfrak{S}_2(N) N + E(N) where the singular series S2(N)=2∏p>2p(pβˆ’2)(pβˆ’1)2∏p∣Np>2pβˆ’1pβˆ’2>0\mathfrak{S}_2(N) = 2\prod_{p>2}\frac{p(p-2)}{(p-1)^2}\prod_{\substack{p|N \\ p>2}}\frac{p-1}{p-2} > 0 for even NN. The error term E(N)E(N) is controlled by exceptional zeros of Dirichlet L-functions.

RemarkExceptional Set for Goldbach

Let E(X)E(X) denote the number of even integers 2<N≀X2 < N \leq X that are not sums of two primes. Current bounds: E(X)β‰ͺX0.879E(X) \ll X^{0.879} (Pintz), improving on Montgomery-Vaughan's X1βˆ’Ξ΄X^{1-\delta}. Under GRH, E(X)β‰ͺX1/2+Ξ΅E(X) \ll X^{1/2+\varepsilon}. The conjecture is E(X)=0E(X) = 0 for all XX. Computationally verified for N≀4Γ—1018N \leq 4 \times 10^{18} (Oliveira e Silva, Herzog, Pardi).


Techniques

ExampleCircle Method for Primes

The key difficulty for Goldbach (compared to Waring) is that the generating function βˆ‘p≀N(log⁑p)e2Ο€ipΞ±\sum_{p \leq N} (\log p) e^{2\pi ip\alpha} involves primes, not powers. On major arcs: βˆ‘p≀N(log⁑p)e2Ο€ipΞ±β‰ˆΞΌ(q)Ο•(q)βˆ‘n≀Ne2Ο€inΞ²\sum_{p \leq N} (\log p) e^{2\pi ip\alpha} \approx \frac{\mu(q)}{\phi(q)} \sum_{n \leq N} e^{2\pi in\beta} by the PNT in arithmetic progressions. On minor arcs: the bound βˆ£βˆ‘(log⁑p)e2Ο€ipα∣β‰ͺN/(log⁑N)A|\sum (\log p)e^{2\pi ip\alpha}| \ll N/(\log N)^A (Vinogradov) requires sophisticated type I/II estimates.