# Volume 3 - 1980

### 1. Some remarks on the mean value of the riemann zeta-function and other Dirichlet series-II

This is a sequel (Part II) to an earlier article with the same title. There are reasons to expect that the estimates proved in Part I without the factor $(\log\log H)^{-C}$ represent the real truth, and this is indeed proved in part II on the assumption that in the first estimate $2k$ is an integer. %This is of great interest, for little has been known on the mean value of $\vert\zeta(\frac{1}{2}+it)\vert^k$ for odd $k$, say $k=1$; for even $k$, see the book by E. C. Titchmarsh [The theory of the Riemann zeta function, Clarendon Press, Oxford, 1951, Theorem 7.19]. The proofs are based on applications of classical function-theoretic theorems, together with mean value theorems for Dirichlet polynomials or series. %In the case of the zeta function, the principle is to write $\vert\zeta(s)\vert^k=\vert\zeta(s)^{k/2}\vert^2$, where $\zeta(s)^{k/2}$ is related to a rapidly convergent series which is essentially a partial sum of the Dirichlet series of $\zeta(s)^{k/2}$, convergent in the half-plane $\sigma>1$.

### 2. One more proof of Siegel's theorem

This paper gives a new elementary proof of the version of Siegel's theorem on $L(1,\chi)=\sum_{n=1}^{\infty}\chi(n)n^{-1}$ for a real character $\chi(\!\!\!\!\mod k)$. The main result of this paper is the theorem: If $3\leq k_1\leq k_2$ are integers, $\chi_1(\!\!\!\!\mod k_1)$ and $\chi_2(\!\!\!\!\mod k_2)$ are two real non-principal characters such that there exists an integer $n>0$ for which $\chi_1(n)\cdot\chi_2(n)=-1$ and, moreover, if $L(1,\chi_1)\leq10^{-40}(\log k_1)^{-1}$, then $L(1,\chi_2)>10^{-4} (\log k_2){-1}\cdot(\log k_1)^{-2}k_2^{-40000L(1,\chi_1)}$. From this the result of T. Tatuzawa on Siegel's theorem follows.

### 3. On a theorem of Erdos and Szemeredi

K. F. Roth proved in 1957 that if $1 = q_1 < q_2 <q_3\ldots$ is the sequence of square-free integers, then $q_{n+1} - q_n = O(n^{\frac{3}{13}}(\log n)^{\frac{4}{13}})$. P. ErdÃ¶s set the problem in more general conditions. If $2 \leq b_1 < b_2 < b_3\ldots$ is a sequence of mutually coprime integers, and $\sum\frac{1}{b_i} < \infty$, he proved that if $1 = q_1 < q_2 <q_3\ldots$ is the sequence of integers not divisible by any $b_i$, then $q_{n+1} - q_n = O\left(q_n^{\theta}\right),$ where $0<\theta<1.$ C. Szemeredi made an important progress and proved that if $Q(x)=\sum_{q_i\leq x} 1$, then $Q(x+h)-Q(x) >\!\!> h,$ where $h \geq x^{\theta}.$ Refining some of Szemeredi's ideas, it is proved in this paper that %if 0 < < 1, and $\sum\frac{1}{b_i}<\infty$, then $Q(x+h) - Q(x) >\!\!> h,$ where $x\geq h \geq x^{\theta}$ and $\theta >\frac{1}{2}$ is any constant. %In the later part, using the ideas of Jutila, Brun and I. M. Vinogradov, a stronger version (Theorem 2) is proved.