The sunset endpoint constants as random-walk averages, with the seven-loop vacuum moment

Abstract. For the equal-mass sunset integral in two dimensions, we write every endpoint coordinate as an explicit signed constant times a positive average over a planar random walk. For n propagators and 1 ≤ j < n/2, the coordinate is (−2)j/j! times the mean of Λj(Wn−j), where Wk is the distance from the origin after k unit steps in independent uniform directions and Λj is the logarithmic kernel of j fused Bessel factors. For arbitrary positive masses there is one such average for each subset, positive exactly when the subset is lighter than its complement. The proof combines the spectral representation of Groote, Körner and Pivovarov with Jensen’s formula. It gives signs and strict growth, the endpoint constants as polynomials in walk averages and odd zeta values, a complex coordinate formula through a continued Domb series, and a local reflection law for the endpoint densities. For eight propagators we prove 70 FPnote[I02K06] − 45π2 FPnote[I04K04] = 11π2ζ(5). It implies the identity ∫0∞ t(1 − 72t2)K0(t)8 dt = 77ζ(5), which Desbois and Ouvry found numerically. With Zhou’s evaluations, the seven-loop vacuum moment is expressed by m(1 + x1 + ⋯ + x7), ζ(5), a walk average and two critical L-values of the weight-six, level-six newform. The eight-Bessel conjecture of Broadhurst and Mellit becomes equivalent to an evaluation of that Mahler measure. All the work was done by AI. Generative-AI disclosure and author responsibility. The statements of the main theorems and outlines of their proofs were first obtained with Claude Fable 5.2 and Claude Opus 5.5, which also found the two zeta-value identities numerically, carried out computations and audited the proofs. OpenAI ChatGPT-6 Pro wrote the complete proofs; it reduced the eight-propagator identity to a single contour integral, showed that it is equivalent to the all-K0 identity of Theorem 7(e), found the monotonicity of the coordinates in the number of propagators, the bound in the right half-plane and the elliptic form of the three-line density, and corrected several statements. It also checked the sources, produced the certified numerical verification and drafted this paper and its supplement. The author directed and audited the work throughout and takes full responsibility for the mathematics and the final text. Changes from version 1.0.0. Version 1.1.0, dated 15 November 2026, adds part (e) of Theorem 7, the identity ∫0∞ t(1 − 72t2)K0(t)8 dt = 77ζ(5), which Desbois and Ouvry found numerically; given Zhou’s sum rules, it is equivalent to the eight-factor identity of part (b). The verification package adds the exact reduction of part (e) and a numerical check, at 60 and 80 digits, of the eight all-K0 moments against part (e) and Zhou’s three sum rules. The proof supplement is unchanged. This record holds the paper (23 pages) and its verification package Matuzas_Sunset_Endpoint_Constants_Random_Walk_Averages_verification.zip (version 1.1.0): the supplement proof_details.pdf (55 pages, with its LaTeX source), which contains the complete analytic proofs; an exact SymPy program for the algebra (136 recorded checks, including all twenty admissible coefficients for three through ten propagators, the eight-factor elimination and the all-K0 reduction of Theorem 7(e)); exact rational interval arithmetic for the eight-propagator comparisons, propagated from the certified balls of the every-loop note's verification package; mpmath programs that compare independent numerical representations of the fusion kernels, the unequal-mass coefficients, the coordinates and vacuum moments through ten propagators, the eight-propagator finite parts, the all-K0 moments of Theorem 7(e) and the endpoint densities; the data, output log and recorded times of every run; and a SHA-256 manifest. README.md in the package is the entry point and maps every statement of the paper to its proof section and to the programs that check it. A numerical comparison is never a premise of a theorem, and Conjecture 8 remains a conjecture. The package was checked with Python 3.12.14, mpmath 1.3.0 and SymPy 1.14.0. Companion notes: J. Matuzas, The sunset endpoint constants at every loop order: coordinates of the vacuum Bessel moments, doi:10.5281/zenodo.24179055; J. Matuzas, The four-loop sunset endpoint constants: a Mahler measure and a Domb series, doi:10.5281/zenodo.24122507; J. Matuzas, The five-loop sunset endpoint constants: a Mahler measure and a second Domb series, doi:10.5281/zenodo.24139789. 2020 Mathematics Subject Classification: Primary 33C10; Secondary 11F67, 11R06, 31A05, 60G50, 81T18, 81Q30. Preprint. External peer review is not claimed.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-15
DOI
https://doi.org/10.5281/zenodo.24237373
Primary Topic
Advanced Mathematical Identities
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

The sunset endpoint constants as random-walk averages, with the seven-loop vacuum moment

Jonas Matuzas
Zenodo (CERN European Organization for Nuclear Research)
Advanced Mathematical Identities
preprint

The sunset endpoint constants as random-walk averages, with the seven-loop vacuum moment

Jonas Matuzas
preprint en

Abstract

Abstract. For the equal-mass sunset integral in two dimensions, we write every endpoint coordinate as an explicit signed constant times a positive average over a planar random walk. For n propagators and 1 ≤ j < n/2, the coordinate is (−2)j/j! times the mean of Λj(Wn−j), where Wk is the distance from the origin after k unit steps in independent uniform directions and Λj is the logarithmic kernel of j fused Bessel factors. For arbitrary positive masses there is one such average for each subset, positive exactly when the subset is lighter than its complement. The proof combines the spectral representation of Groote, Körner and Pivovarov with Jensen’s formula. It gives signs and strict growth, the endpoint constants as polynomials in walk averages and odd zeta values, a complex coordinate formula through a continued Domb series, and a local reflection law for the endpoint densities. For eight propagators we prove 70 FPnote[I02K06] − 45π2 FPnote[I04K04] = 11π2ζ(5). It implies the identity ∫0∞ t(1 − 72t2)K0(t)8 dt = 77ζ(5), which Desbois and Ouvry found numerically. With Zhou’s evaluations, the seven-loop vacuum moment is expressed by m(1 + x1 + ⋯ + x7), ζ(5), a walk average and two critical L-values of the weight-six, level-six newform. The eight-Bessel conjecture of Broadhurst and Mellit becomes equivalent to an evaluation of that Mahler measure. All the work was done by AI. Generative-AI disclosure and author responsibility. The statements of the main theorems and outlines of their proofs were first obtained with Claude Fable 5.2 and Claude Opus 5.5, which also found the two zeta-value identities numerically, carried out computations and audited the proofs. OpenAI ChatGPT-6 Pro wrote the complete proofs; it reduced the eight-propagator identity to a single contour integral, showed that it is equivalent to the all-K0 identity of Theorem 7(e), found the monotonicity of the coordinates in the number of propagators, the bound in the right half-plane and the elliptic form of the three-line density, and corrected several statements. It also checked the sources, produced the certified numerical verification and drafted this paper and its supplement. The author directed and audited the work throughout and takes full responsibility for the mathematics and the final text. Changes from version 1.0.0. Version 1.1.0, dated 15 November 2026, adds part (e) of Theorem 7, the identity ∫0∞ t(1 − 72t2)K0(t)8 dt = 77ζ(5), which Desbois and Ouvry found numerically; given Zhou’s sum rules, it is equivalent to the eight-factor identity of part (b). The verification package adds the exact reduction of part (e) and a numerical check, at 60 and 80 digits, of the eight all-K0 moments against part (e) and Zhou’s three sum rules. The proof supplement is unchanged. This record holds the paper (23 pages) and its verification package Matuzas_Sunset_Endpoint_Constants_Random_Walk_Averages_verification.zip (version 1.1.0): the supplement proof_details.pdf (55 pages, with its LaTeX source), which contains the complete analytic proofs; an exact SymPy program for the algebra (136 recorded checks, including all twenty admissible coefficients for three through ten propagators, the eight-factor elimination and the all-K0 reduction of Theorem 7(e)); exact rational interval arithmetic for the eight-propagator comparisons, propagated from the certified balls of the every-loop note's verification package; mpmath programs that compare independent numerical representations of the fusion kernels, the unequal-mass coefficients, the coordinates and vacuum moments through ten propagators, the eight-propagator finite parts, the all-K0 moments of Theorem 7(e) and the endpoint densities; the data, output log and recorded times of every run; and a SHA-256 manifest. README.md in the package is the entry point and maps every statement of the paper to its proof section and to the programs that check it. A numerical comparison is never a premise of a theorem, and Conjecture 8 remains a conjecture. The package was checked with Python 3.12.14, mpmath 1.3.0 and SymPy 1.14.0. Companion notes: J. Matuzas, The sunset endpoint constants at every loop order: coordinates of the vacuum Bessel moments, doi:10.5281/zenodo.24179055; J. Matuzas, The four-loop sunset endpoint constants: a Mahler measure and a Domb series, doi:10.5281/zenodo.24122507; J. Matuzas, The five-loop sunset endpoint constants: a Mahler measure and a second Domb series, doi:10.5281/zenodo.24139789. 2020 Mathematics Subject Classification: Primary 33C10; Secondary 11F67, 11R06, 31A05, 60G50, 81T18, 81Q30. Preprint. External peer review is not claimed.

Zenodo (CERN European Organization for Nuclear Research)
Vilnius University (LT)
Advanced Mathematical Identities
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon