I am currently trying to understand the architecture of various proofs of results in additive analytic number theory. In particular, I am studying the proof that every large odd integer is the sum of three primes. In the exposition I’m reading the author, K. Soundararajan, begins by making some guesses on the order of various quantities using arithmetic combinatorics. I want to learn this method and I'm looking for easier, toy problems, where I can practice this approach.
$\begingroup$
$\endgroup$
3
-
2$\begingroup$ What you want to look into is Cramer's model, and how it's used to make probabilistic heuristics. $\endgroup$Wojowu– Wojowu2026-01-31 16:02:03 +00:00Commented 15 hours ago
-
$\begingroup$ @Wojowu I have studied Cramer's model and I understand the heuristics behind the prime number theorem. What I want to do is run Vinogradov's method to different types of arithmetic progressions and understand the limitations of this method. $\endgroup$Mustafa Said– Mustafa Said2026-01-31 16:10:57 +00:00Commented 15 hours ago
-
2$\begingroup$ This is very different to what you seem to ask in the body of the question, which is explicitly about the heuristics. If you are asking about tools used to turn such heuristics into formal proofs, that's a completely different story $\endgroup$Wojowu– Wojowu2026-01-31 16:16:23 +00:00Commented 15 hours ago
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
3
Additive analytic number theory is a different field from arithmetic combinatorics, and it also differs from additive combinatorics. The goals and the methods are different.
The solution of the ternary Goldbach problem by Vinogradov belongs to additive analytic number theory. Vinogradov used the so-called circle method (or Hardy-Littlewood method). (Update: Anurag Sahay kindly pointed out in a comment that subsequently other methods were found for the same problem.) Applications of this method are always technical, because one needs to carefully analyze the underlying generating series (which are trigonometric series, also known as exponential sums), e.g. how it behaves around rational points of small denominators ("major arcs") and away from such points ("minor arcs"). I recommend that you study first the solution of Waring's problem with this method. A good source is Vaughan's book "The Hardy-Littlewood method", where Chapter 2 (pp. 8-25) discusses Waring's problem and Chapter 3 (pp. 27-36) discusses Goldbach's problem. A solid background in analytic number theory helps (good textbooks are Davenport: Multiplicative number theory and Montgomery-Vaughan: Multiplicative number theory I).
As a side remark, it might appear odd that one needs to study multiplicative number theory before additive number theory. The reason is that in additive number theory we add things that are defined in terms of multiplication (e.g. powers in Waring's problem and primes in Goldbach's problem). As the famous physicist Lev Landau once said: "Why add prime numbers? Prime numbers are made to be multiplied, not added."
-
1$\begingroup$ thank for the explanation and the references. I appreciate your kind tone and I will try to adapt the proof to Waring's problem. $\endgroup$Mustafa Said– Mustafa Said2026-01-31 20:47:55 +00:00Commented 10 hours ago
-
1$\begingroup$ Dear GH: I wonder the degree to which "(which is still the only method known for the problem)" is true. It depends, of course, on how widely you define the circle method, but there are proofs of Vinogradov's theorem in the literature which do not use the circle method in its classical form -- I have in mind MR3347954 and MR0834356. The latter especially does not seem to use anything I would consider a circle method on a quick skim. $\endgroup$Anurag Sahay– Anurag Sahay2026-01-31 20:53:13 +00:00Commented 10 hours ago
-
1$\begingroup$ @AnuragSahay I did not know about these references, thanks for pointing them out! I updated my post accordingly. $\endgroup$GH from MO– GH from MO2026-01-31 21:27:04 +00:00Commented 9 hours ago