Chen's double sieve, Goldbach's conjecture and the twin prime problem (Q4811277)

For an even integer \(N\geq4\), denote by \(D(N)\) the number of representations of~\(N\) as a sum of two primes. The Goldbach conjecture states that \(D(N)\geq1\), and if the Hardy-Littlewood programme can be completed then \(D(N)\sim2\Theta(N)\) as \(N\to\infty\), with \[ \Theta(N)={C_NN\over\log^2N},\qquad C_N=\prod_{p| N,p>2}{p-1\over p-2}\prod_{p>2}\biggl(1-{1\over (p-1)^2}\biggr). \] Meanwhile, an upper bound of the form \(D(N)\leq K\Theta(N)\) has been established; thus any value \(K>16\) being admissible was obtained by \textit{A.~Selberg} [11. Skand. Mat. Kongr., Trondheim 1949, 13--22 (1952; Zbl 0048.03101)], \(K>12\) by \textit{C.~D.~Pan} [Chinese Math. 5, 642--652 (1964; Zbl 0283.10026)], and \(K>8\) by \textit{E.~Bombieri} and \textit{H.~Davenport} [Proc. R. Soc. Lond., Ser. A 293, 1--18 (1966; Zbl 0151.04201)]. The main tools being used are the linear sieve and the Bombieri-Vinogradov theorem. In~1978 \textit{J.~R.~Chen} [Sci. Sin. 21, 701--739 (1978; Zbl 0399.10046)] injected a new idea in the application of Selberg's sieve, and by bringing in a complicated argument involving a `double sieve' he showed that \(K=7{\cdot}8342\) is an admissible value. Inspired by his own ideas in an earlier paper [Acta Arith. 55, 365--394 (1990; Zbl 0662.10033)] the author has given a more comprehensive treatment of Chen's double sieve, thereby showing that \(K=7{\cdot}8209\) is admissible; the new ingredient being the Rosser-Iwaniec linear sieve. With the current state of affair, the proof of Goldbach's conjecture almost amounts to trying to overcome the `parity obstacle', and thereby showing that \(D(N)\gg\Theta(N)\). If one is to bypass the parity obstacle, the best corresponding result is that \(D_{1,2}(N)\geq H\Theta(N)\), where \(D_{1,2}(N)\) is the number of representations of~\(N\) as the sum of a prime and a number with at most two prime divisors, and Chen was the first to achieve this, showing that \(H=0{\cdot}67\) is admissible. There have been several improvements on the numerical value for~\(H\), and here the author obtains \(H=0{\cdot}836\). The twin prime problem is of a similar nature and status, and corresponding improvements for this problem are also given by the author, including results in short intervals.