Go to Harold Diamond's Home Page
" Elementary" methods in prime number theory involve the use only of real variable techniques; methods of complex function theory or harmonic analysis or use of the properties of the Riemann zeta function in C are not allowed. This article contains an elementary proof of the prime number theorem in the form
pi(x) - li(x) = O{x exp(-c log^alpha x)}.
The proof uses a generalization of the Selberg formula and a tauberian argument.
This is a survey article on elementary methods.
Beurling's numbers are a generalization of the integers in which only multiplicative structure is preserved. This article is one of a series that I have written on this theme. Under a weak hypothesis upon the distribution of the Beurling integers, an analogue of Chebyshev's O bound for pi(x)/(x/log x) is established.
Methods of E. Landau from a hundred years ago show that Beurling generalized numbers having an integer counting function N(x) = cx + O(x^a) with c > 0 and 1/2 < a < 1, satisfy the prime number theorem with the error bound of de la Vallee Poussin. Here we show by a probabilistic construction that this error bound is in fact optimal for a discrete Beurling number system.
A tauberian theorem is a kind of " unaveraging" result. For technical reasons the " kernel" function which creates the average is usually assumed to be positive. Here it shown that one can convert a class of kernels which change sign into positive ones for which tauberian methods succeed.
For f a periodic function mod q, let f^ denote the finite Fourier transform of f. A criterion is given for deciding when f can be recovered from the restriction of f^ to the integers relatively prime to q. Applications are given when this representation applies.
In sieve theory estimates are made of the number of integers remaining in a given collection when those falling into certain residue classes are removed. The authors have devised a new sieve whose estimates are expressed in terms of a pair of functions defined by a coupled system of difference differential equations. This article is the conclusion of a series in which the authors establish the existence of the required functions.