E. Kowalski's blog






         Comments on mathematics, mostly.

24.10.2013

James Maynard, auteur du théorème de l’année

How many times in a year is an analytic number theorist supposed to faint from admiration? We’ve learnt of the full three prime Vinogradov Theorem by Helfgott, then of Zhang’s proof of the bounded gap property for primes. Now, from Oberwolfach, comes the equally (or even more) amazing news that James Maynard has announced a proof of the bounded gap property that manages not only to ask merely for the Bombieri-Vinogradov theorem in terms of information concerning the distribution of primes in arithmetic progressions, but also obtains a gap smaller than 700 (in fact, even better when using optimal narrow k-tuples), where the efforts of the Polymath8 project only lead to 4680, using quite a bit of machinery.

(The preprint should be available soon, from what I understand, and thus a full independent verification of these results.)

Two remarks, one serious, one not (the reader can guess which is which):

(1) Again, from friends in Oberwolfach (teaching kept me, alas, from being able to attend the conference), I heard that Maynard’s method leads to the bounded gap property (with increasing bounds on the gaps) using as input any positive exponent of distribution for primes in arithmetic progressions (where Bombieri-Vinogradov means exponent 1/2; incidentally, this also means that the Generalized Riemann Hypothesis is strong enough to get bounded gaps, which did not follow from Zhang’s work). From the point of view of modern proofs, there is essentially no difference between positive exponent of distribution and exponent 1/2, since either property would be proved using the large sieve inequality and the Siegel-Walfisz theorem, and it makes little sense to prove a weaker large sieve inequality than the one that gives exponent 1/2. Question: could one conceivably even dispense with the large sieve inequality, i.e., prove the bounded gap property only using the Siegel-Walfisz theorem? This is a bit a rhetorical question, since the large sieve is nowadays rather easy, but maybe the following formulation is of some interest: do we know an example of an increasing sequence of integers n_k, not sparse, not weird, that satisfies the Siegel-Walfisz property, but has unbounded gaps, i.e., \liminf (n_{k+1}-n_k)=+\infty?

(2) There are still a bit more than two months to go before the end of the year; will a bright PhD student rise to the challenge, and prove the twin prime conjecture?

[P.S. Borgesian readers will understand the title of this post, although a spanish version might have been more appropriate...]

5 Comments

  1.   Greg — 30.10.2013 @ 1:54    Reply

    To my understanding from hearing James speak at Oberwolfach: he expects that his method will end up only requiring any positive exponent of distribution, but he did not announce that he had a proof of that yet.

    Another remarkable consequence of his work is that assuming the Elliott-Halberstam conjecture, one actually gets double-gaps p_{n+2} – p_n bounded by 700 infinitely often; no previous method could achieve that under any reasonable hypothesis, I believe.

  2.   Gil Kalai — 31.10.2013 @ 14:01    Reply

    Dear Emmanuel, Does Maynard’s method deal (or expected to) with a large number of primes (more than two) in bounded intervals?

  3.   Gergely Harcos — 05.11.2013 @ 1:17    Reply

    Dear Gil Kalai, my understanding (from what I heard of Maynard’s Oberwolfach talk) is that he can obtain a large number primes in bounded intervals. Basically (or hopefully), with a new way of weighting the translates of a tuple, he can show that the average number of primes in such a tuple is greater than 100, say. For the same reason he can do with any positive level of distribution: the average number of primes in the translated tuple is the available level of distribution times a large number, say a million.

  4.   Gil Kalai — 11.11.2013 @ 17:07    Reply

    Dear Gergely, many thanks. This is very impressive.

  5.   j. — 20.11.2013 @ 6:28    Reply

    In Spanish: James Maynard, autor del teorema del año.

RSS feed for comments on this post. TrackBack URI

Leave a comment

© 2014 E. Kowalski's blog   Hosted by uzh|ethz Blogs