Legendre duality for certain summations over the Farey pairs

Abstract

Each irreducible fraction p/q>0 corresponds to a primitive vector (p,q)∈ Z2 with positive coordinates. Such a vector (p,q) can be uniquely written as the sum of two primitive vectors (a,b),(c,d)∈ Z≥ 02 spanning a parallelogram of oriented area one. We present new summation formulas over the set of such parallelograms. These formulas depend explicitly on a,b,c,d and thus define a summation over primitive vectors (p,q)=(a+c,b+d) indirectly. Equivalently, these sums may be interpreted as running over Farey pairs, i.e. pairs of fractions 0≤ c/d<a/b≤ 1 satisfying ad-bc=1. The input for our formulas is the graph of a strictly concave function g. The terms are the areas of certain triangles formed by tangents to the graph of g. Several of these formulas for different g yield values involving π. For g a parabola we recover the classical Mordell-Tornheim series (also called the Witten series). As a nice application we also discuss formulas for continued fractions for an arbitrary real number α involving coefficients of the continued fraction and the differences between the convergents and α. Using Hata's work, we show that the terms in the above formulas are the coefficients of the Legendre transform of g in a certain Schauder basis, allowing us to interpret our formulas as Parseval-type identities. We hope that the Legendre duality sheds new light on Hata's approach. Raising the terms in the above summation formula to the power s we obtain a function Fg(s). We prove that for a strictly concave g, the function Fg(s) converges for s>2/3 and diverges at s=2/3.

0

Turn this paper into a full lesson

ArcXiv compiles a staged curriculum from this paper: 8-12 lessons across beginner → advanced, synthesised section guides, visuals, flashcards, a quiz, exercises, and on-demand deep dives per section. Grounded in the abstract, never invented.

Discussion (0)

Sign in to join the discussion.

Loading comments…