Paperlane

no. 17 · Wednesday 18 November 2026

Estimates for the Number of Solutions of Algebraic Diophantine Equations: From the Circle Method to the Manin Conjecture

Victor Volfson

math.NT · v3 · 59 pp

Abstract

This work continues the research of the author on estimates for the number of solutions of algebraic diophantine equations by means of the circle method and develops this line in the geometric direction, towards counting integral points on log Fano varieties. It consists of two parts. Part 1 is the corrected version of the previous work. For diagonal equations $c_1x^{k_1} +...+c_sx^{k_s} = n$ with natural coefficients, upper asymptotic estimates for the number of natural solutions in the range $s \leq 2^k$, where the classical asymptotics is not applicable, are obtained by the circle method and the Hua lemma. The estimates are refined for $s = 2,3,4$ and various $k_i$; in the range $s > 2^k$ they agree with the classical ones. Part 2 develops the same line: from counting solutions of equations to counting integral points on log Fano varieties, where the circle method fails because of insuf ficient dimension. In the log case, the product in the Peyre constant diverges as a harmonic series. A method of analytic compensation is introduced that cancels the divergent term and returns the problem to the range of applicability of the standard scheme. The main result is a method for computing the Peyre log constant with guar anteed accuracy and certified intervals for the number of integral points whenever an estimate of the remainder of the asymptotics is known. The log constant is computable unconditionally, without the Manin conjecture; the conjecture is needed only to interpret it as the leading term of the asymptotics. An experimental verification for the algebraic torus and the Fermat cubic surface is presented.

comments
59 pages. v3: Part 1 has been edited; Part 2 consists of clarifications of statements
fields
Number Theory
pdf ↗ page arXiv:1512.07079 ↗