math.NT — Number Theory
Cross-Energy of Dirichlet Partial Sums. Part 18: The Euler-Caceres bound
Gonek, Hughes and Keating factored , an Euler product times a product over the zeros. We use an exact form , a smoothed Euler factor, and for $\re s_0>\frac12$ study the Euler--Caceres bound , the Poisson kernel. It is the off-line Blaschke term of the Poisson--Jensen formula at , independent of , plus a positive Jensen gap. Under RH, , with universal constants for multiplicity (). Hence, for any single , RH holds if and only if , and RH with simple zeros if and only if tends to an explicit number ( at ). If $\Theta=\sup\re\rho>\frac12$, Landau's theorem in the variable gives for $\re s_0>1$, attained or not, so RH holds iff at one point. Unconditionally, for $\re s_0>1$, and the off-line term at is below . Under RH we identify the second-order term up to an interaction of the wells (known only to be ), related heuristically to Montgomery's pair correlation; it is governed by the gaps between zeros, and under a small-gap hypothesis the law holds with error . We give a function-field model where the bound holds, and controls (a Beurling system with the zeros of and a zero at , the Davenport--Heilbronn function) suggesting that a proof must couple the Euler product with the functional equation. Numerically the bound follows the law, stays below the floor of Dirichlet-polynomial mollifiers at the centre , and grows for the Davenport--Heilbronn function and for a planted off-line zero. We do not prove RH.