Processing math: 100%

Tuesday, May 19, 2015

Mertens, Ein Beitrag zur analytischen Zahlentheorie, 1874

I started reading the article in the title to see how Mertens proves his famous theorems.

Introduction

Mertens says he will prove the formulas
px1p=loglogx+C
and px111p=Clogx
The sum and the product are over primes p. Mertens will also calculate the constants C and C. Mertens also says that the formulas are already in a paper by Chebyshev, but with doubtful proof. These formulas are now known as Mertens theorems.



Section 1

Mertens proves that θ(x)<2x; he says that Chebyshev gives a sharper bound which he does not need. Along the way Mertens also proves that ψ(x)<2x. I found Mertens reasoning easy to follow because it is an easy version of the reasoning of a (later) paper by Ramanujan that I read. Ramanujan's reasoning is more involved because he proves a stronger result, namely Bertrand's postulate.

Section 2

In this section Mertens proves that pxlogpp=logx+R with |R|<2. This statement is now known as Mertens first theorem. The proof takes less than a page, and only uses elementary inequalities. The proof is easy to read; in the middle I found a bit of help in this paper from Mark Villarino.

Here are some graphs illustrating this theorem.


In graph above, the blue dots are the function  pxlogpp and the red line is the function logx.

In the graph below I illustrate that the difference is indeed less than 2.



No comments:

Post a Comment