The closing unit of the masterclass studies the distances between consecutive primes. It derives their average ln p from the prime number theorem, reads the tables of record gaps against Cramér's conjecture, builds long runs of composite numbers and explains why they stay below the average, and states the theorems of Westzynthius, Rankin, and the two proofs of 2014 on large gaps. It then explains the method of Goldston, Pintz, and Yıldırım, admissible tuples, and the level of distribution, and tells how Zhang's 70 million became Maynard's 600 and the Polymath bound 246 within a year, and why the twin prime conjecture is still out of reach.
Preview: the beginning of the unit
The prime number theorem of Unit 12 says how many primes there are up to \(x\), and nothing about where they stand. Two primes can be as close as \(11\) and \(13\), and they can be as far apart as \(370{,}261\) and \(370{,}373\), with \(111\) composite numbers in between. This unit is about the distances between consecutive primes, and it asks what they are on average, how far below and above the average they go, what is conjectured about the records, and what was proved in 2013 and 2014, when the question whether some bounded distance between primes occurs infinitely often was answered within a year by a lecturer nobody knew, a postdoctoral researcher, and an online collaboration of dozens of mathematicians. The unit uses the prime number theorem and Cramér’s model of Unit 12 and the Selberg sieve, the Bombieri-Vinogradov theorem, and the parity problem of Unit 15. As in Units 13 to 15, \(\ln\) is the natural logarithm; research papers on this subject write \(\log\) for it.
1 The average gap
Let \(p_n\) denote the \(n\)-th prime, so that \(p_1 = 2\), \(p_2 = 3\), \(p_3 = 5\), and let
\begin{equation}
g_n = p_{n+1} - p_n \label{eq:gap}
\end{equation}
be the gap after \(p_n\). The first gaps are \(1, 2, 2, 4, 2, 4, 2, 4, 6, 2, 6, \ldots\), and \(g_1 = 1\) is the only odd gap, because every prime after \(2\) is odd, so every later gap is even. A gap of \(2\) is a pair of twin primes (Unit 8). The sum of the first \(N - 1\) gaps is \(p_N - 2\), because every prime except the first and the last cancels when the differences are added (the sum telescopes, as the proof box shows), and this simple fact, combined with the prime number theorem, fixes the average.