It has been known since the pioneering work of Postnikov (1956) that character sums modulo prime powers pkpk with small pp and large kk admit bounds of much shorter length than for generic or prime moduli. The result of Postnikov was consecutively improved by Gallagher (1972), Iwaniec (1974) and Chang (2014).


Igor Shparlinski

Research Area

UNSW Sydney


Wed, 19/09/2018 - 2:00pm


RC-4082, The Red Centre, UNSW

In a joint work with Bill Banks we modify the traditional scheme and reduce the problem to estimating bivariate exponential sums which can be treated much more efficiently than previously used univariate sums. In turn, we improve the bounds of  Iwaniec (1974) on  the zero-free region of LL-functions and the error term for the counting function of primes in progressions modulo pkpk as well as the bound of Green (2012) on non-correlation of the M{\"o}bius function
with multiplicative characters.

Jointly   with Kui Liu and Tianping Zhang, we used a similar approach to prove power cancellations among very short sums of Kloosterman sums modulo pkpk. As an application, we break Selberg's 2/32/3-barrier for the Dirichlet divisor problem in arithmetic progressions modulo pkpk and move it all the way up to 11. That is, we give an asymptotic formula for the intervals of length xx with pk≤x1−εpk≤x1−ε rather than q≤x2/3−εq≤x2/3−ε as in the case of an arbitrary modulus qq (for any  ε>0ε>0).

We will also discuss (rather bright) perspectives of further extensions of these results and   (rather bleak) perspectives for their further improvements.

Joint work with Bill Banks, Kui Liu, Tianping Zhang