arXiv subjects

David Broadhurst

Publications and source records attributed to David Broadhurst.

At least 19 recordsLinked to original sources

arXiv2026

Resurgent Lambert series from Feynman and beyond

Lambert series of the form $\sum_{n>0}a(n)q^n/(1-q^n)$ are ubiquitous in mathematical physics. In particular, 2-loop sunrise and 3-loop banana Feynman diagrams yield Lambert series with $a(n)$ of the form $\chi(n)/n^s$ where $\chi(n)$ is a Dirichlet character. Resurgence concerns the singular limit as $|q|$ approaches 1. In the Feynman cases we can control this limit, obtaining rapidly convergent expressions, since the Lambert series are iterated integrals of holomorphic Eisenstein series twisted by a character. We generalize this result, to include modular resurgent structures found in topological-string observables.

hep-th↗

arXiv2025

Polynomials and asymptotic constants in a resurgent problem from 't Hooft

In a recent study of the quantum theory of harmonic oscillators, Gerard 't Hooft proposed the following problem: given $G(z)=\sum_{n=1}^\infty\sqrt{n}\,z^n$ for $|z|<1$, find its analytic continuation for $|z|\ge1$, excluding a branch-cut $z\in[1,\,\infty)$. A solution is provided by the bilateral convergent sum $G(z)=\frac12\sqrt{\pi}\sum_{n=-\infty}^\infty(2\pi{\rm i}n-\log(z))^{-3/2}$. On the negative real axis, $G(-{\rm e}^u)$ has a sign-constant asymptotic expansion in $1/u^2$, for large positive $u$. Optimal truncation leaves exponentially suppressed terms in an asymptotic expansion ${\rm e}^{-u}\sum_{k=0}^\infty P_k(x)/u^k$, with $P_0(x)=x-\frac23$ and $P_k(x)$ of degree $2k+1$ evaluated at $x=u/2-\lfloor u/2\rfloor$. At large $k$, these polynomials become excellent approximations to sinusoids. The amplitude of $P_k(x)$ increases factorially with $k$ and its phase increases linearly, with $P_k(x)\sim\sin((2k+1)C-2\pi x)R^{2k+1}\Gamma(k+\frac12)/\sqrt{2\pi}$, where $C\approx1.0688539158679530121571$ and $R\approx0.5181839789815558726739$ are asymptotic constants satisfying $R\exp({\rm i}\,C)=\sqrt{-1/(2+\pi{\rm i})}$.

math.NT↗

arXiv2025

Resurgent Lambert series with characters

We consider certain Lambert series as generating functions of divisor sums twisted by Dirichlet characters and compute their exact resurgent transseries expansion near $q=1^-$. For special values of the parameters, these Lambert series are expressible in terms of iterated integrals of holomorphic Eisenstein series twisted by the same characters and the transseries representation is a direct consequence of the action of Fricke involution on such twisted Eisenstein series. When the parameters of the Lambert series are generic the transseries representation provides for a quantum-modular version of Fricke involution which for a particular example we show being equivalent to modular resurgent structures found in topological strings observables.

math.NT↗

arXiv2025

Number of partitions of modular integers (with an Appendix by P. Deligne)

For integers $n,k,s$, we give a formula for the number $T(n,k,s)$ of order $k$ subsets of the ring $\mathbb{Z}/n\mathbb{Z}$ whose sum of elements is $s$ modulo $n$. To do so, we describe explicitly a sequence of matrices $M(k)$, for positive integers $k$, such that the size of $M(k)$ is the number of divisors of $k$, and for two coprime integers $k_{1},k_{2}$, the matrix $M(k_{1}k_{2})$ is the Kronecker product of $M(k_{1})$ and $M(k_{2})$. For $s=0, 1, 2$, and for $s=k/2$ when $k$ is even, the sequences $T(n,k,s)$ are related to the number of necklaces with $k$ black beads and $n-k$ white beads, and to Lyndon words. This work begins with empirical determinations of $M(k)$ up to $k=10000$, from which we infer a closed formula that encompasses many entries in the Encyclopedia of Integer Sequences. Its proof comes from work on Ramanujan sums, by Ramanathan, with a generalization to wider problems linked to representation theory and recently described by Deligne.

math.NT↗

arXiv2024

Five families of rapidly convergent evaluations of zeta values

This work derives 5 methods to evaluate families of odd zeta values by combining a power of $\pi$ with Lambert series whose ratios of successive terms tend to $e^{-\pi\sqrt{a}}$ with integers $a\ge7$, outperforming Ramanujan's results with merit $a=4$. Families with $a=7$ and $a=8$ evaluate $\zeta(2n+1)$. Families with $a=9$ and $a=16$ evaluate $\zeta(4n+1)$ with faster convergence. A fifth family with $a=12$ evaluates $\zeta(6n+1)$ and gives the fastest convergence for $\zeta(6n+7)$. Members of three of the families were discovered empirically by Simon Plouffe. An intensive new search strongly suggests that there are no more than 5 families with integers $a\ge7$. There are at least 20 families that involve Lambert series with rational $a>4$. Quasi-modular transformations of Lambert series resolve rational sequences that were discovered empirically. Expansions of Lambert series in polylogarithms, familiar from quantum field theory, provide proofs of all known evaluations with rational merit $a>4$.

math.NT↗

arXiv2023

Richard Feynman's talent for finding things out

This article is a summary of a talk about Richard Feynman, given at a conference Polymaths across the Eras organized in November 2023 by the St Cross Centre for the History and Philosophy of Physics (HAPP) in Oxford. It describes Feynman as an unconventional polymath, primarily concerned with his own understanding of nature, having little regard for previous authorities. His curiosity, respect for experiment, ability to calculate and notable ingenuity led to important developments in quantum mechanics, quantum electrodynamics, and strong interactions. In particle physics, Feynman diagrams have been a mainstay of calculation for the last 75 years. His relish for the spoken word led to renown as an educator and raconteur. His perceptions were influential in a wide range of other scientific fields, including weak interactions, gravity, superconductivity, biology, nanotechnology, algorithmic computation and quantum computers. In all of this, he held to the idea that nature has an intrinsic simplicity and beauty that permit us to find things out, if only we will try hard enough.

physics.hist-ph↗

arXiv2023

Dickman multiple polylogarithms and the Lindemann-Furry letters

The Dickman function $\rho(u)$ gives the asymptotic probability that a large integer $N$ has no prime divisor exceeding $N^{1/u}$. We expand it in terms of rapidly computable multiple polylogarithms, as defined by Goncharov and intensively used for evaluations of Feynman integrals in quantum field theory. In parallel, we solve Buchstab's differential-delay equation, which concerns large integers $N$ divisible by no prime less than $N^{1/u}$. Discussion of the latter problem occurred in letters to the journal Nature during the second world war, from the physicists Frederick Lindemann and Wendell Furry. We recount how Furry evaluated a dilogarithm in reply to a puzzle resulting from Mertens' third theorem, raised by Lindemann. We refine Furry's analysis to include multiple polylogarithms of weights up to 200.

math.NT↗

arXiv2022

Multivariate elliptic kites and tetrahedral tadpoles

This work deals with two types of Feynman integrals in perturbative quantum field theory: the 2-loop 2-point kite, with 5 arbitrary internal masses, and its completion by a sixth propagator, to give a 3-loop tetrahedral tadpole, with 6 arbitrary masses. These general-mass cases cover broken and unbroken gauge theories, based on the Lie algebras U(1), SU(2) and SU(3), for the electromagnetic, weak and strong interactions. The elliptic substructure of these integrals should not be regarded as an obstruction. Rather, it is a bonus, thanks to the arithmetic-geometric mean of Gauss. Compact formulae are given, to handle all cases. Zero-mass limits are carefully considered. Anomalous thresholds of triangles in the kite pose no problem. The number theory of tadpoles is investigated, with intriguing results.

hep-th↗

arXiv2022

Taming a resurgent ultra-violet renormalon

Perturbative expansions in quantum field theory diverge for at least two reasons: the number of Feynman diagrams increases dramatically with the loop number and the process of renormalization may make the contribution of some diagrams large. We give an example of the second problem, from an ultra-violent renormalon of $\phi^3$ theory in 6 dimensions, where we can compute to very high loop-order. Taming this renormalon involves recent work on resurgence. This challenge is much more demanding than the corresponding problem for Yukawa theory in 4 dimensions.

hep-th↗

arXiv2022

Resonant resurgent asymptotics from quantum field theory

We perform an all-order resurgence analysis of a quantum field theory renormalon that contributes to an anomalous dimension in six-dimensional scalar $\phi^3$ theory and is governed by a third-order nonlinear differential equation. We augment the factorially divergent perturbative expansion associated to the renormalon by asymptotic expansions to all instanton orders, in a conjectured and well-tested formula. A distinctive feature of this renormalon singularity is the appearance of logarithmic terms, starting at second-instanton order in the trans-series. To highlight this and to illustrate our methods, we also analyze the trans-series for a closely related second-order nonlinear differential equation that exhibits a similarly resonant structure but lacks logarithmic contributions.

hep-th↗

arXiv2021

Large Tate--Shafarevich orders from good $abc$ triples

Record values are determined for the order $|\Sha|$ of the Tate--Shafarevich group of an elliptic curve $E$, computed analytically by the Birch--Swinnerton-Dyer conjecture, and for the Goldfeld--Szpiro ratio $G=|\Sha|/\sqrt{N}$, where $N$ is the conductor of $E$. The curves have rank zero and are isogenous to quadratic twists of Frey curves constructed from coprime positive integers $(a,b,c)$ with $a+b=c$ and $c>r^{1.4}$, where the radical $r$ is the product of the primes dividing $abc$. Curves with $|\Sha|>250000^2$ and $G>12$ are found in 20 isogeny classes. Three curves have $G>150$. The largest value of $|\Sha|$ is $1937832^2>3.755\times10^{12}$. This is more than 3.5 times the previous record, which had been computed at a cost about 600 times greater than that for the new record. The primes 25913, 27457, 36929 and 49253 are identified as divisors of $|\Sha|$ values.

math.NT↗

arXiv2021

Empirical determinations of Feynman integrals using integer relation algorithms

Integer relation algorithms can convert numerical results for Feynman integrals to exact evaluations, when one has reason to suspect the existence of reductions to linear combinations of a basis, with rational or algebraic coefficients. Once a tentative reduction is obtained, confidence in its validity is greatly increased by computing more decimal digits of the terms and verifying the stability of the result. Here we give examples of how the PSLQ and LLL algorithms have yielded remarkable reductions of Feynman integrals to multiple polylogarithms and to the periods and quasi-periods of modular forms. Moreover, these algorithms have revealed quadratic relations between Feynman integrals. A recent application concerning black holes involves quadratic relations between combinations of Feynman integrals with algebraic coefficients.

hep-ph↗

arXiv2018

Eta quotients and Rademacher sums

Eta quotients on $\Gamma_0(6)$ yield evaluations of sunrise integrals at 2, 3, 4 and 6 loops. At 2 and 3 loops, they provide modular parametrizations of inhomogeneous differential equations whose solutions are readily obtained by expanding in the nome $q$. Atkin-Lehner transformations that permute cusps ensure fast convergence for all external momenta. At 4 and 6 loops, on-shell integrals are periods of modular forms of weights 4 and 6 given by Eichler integrals of eta quotients. Weakly holomorphic eta quotients determine quasi-periods. A Rademacher sum formula is given for Fourier coefficients of an eta quotient that is a Hauptmodul for $\Gamma_0(6)$ and its generalization is found for all levels with genus 0, namely for $N = 1,2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 16, 18, 25$. There are elliptic obstructions at $N = 11, 14, 15, 17, 19, 20, 21, 24, 27, 32, 36, 49,$ with genus 1. We surmount these, finding explicit formulas for Fourier coefficients of eta quotients in thousands of cases. We show how to handle the levels $N=22, 23, 26, 28, 29, 31, 37, 50$, with genus 2, and the levels $N=30,33,34,35,39,40,41,43,45,48,64$, with genus 3. We also solve examples with genera $4,5,6,7,8,13$.

math.NT↗

arXiv2017

A magnetic double integral

In a recent study of how the output voltage of a Hall plate is affected by the shape of the plate and the size of its contacts, Udo Ausserlechner has come up with a remarkable double integral that can be viewed as a generalization of the classical elliptic "AGM" integral. Here we discuss transformation properties of the integral, which were experimentally observed by Ausserlechner, as well as its analytical and arithmetic features including connections with modular forms.

math.NT↗

arXiv2016

Feynman integrals, L-series and Kloosterman moments

This work lies at an intersection of three subjects: quantum field theory, algebraic geometry and number theory, in a situation where dialogue between practitioners has revealed rich structure. It contains a theorem and 7 conjectures, tested deeply by 3 optimized algorithms, on relations between Feynman integrals and L-series defined by products, over the primes, of data determined by moments of Kloosterman sums in finite fields. There is an extended introduction, for readers who may not be familiar with all three of these subjects. Notable new results include conjectural evaluations of non-critical L-series of modular forms of weights 3, 4 and 6, by determinants of Feynman integrals, an evaluation for the weight 5 problem, at a critical integer, and formulas for determinants of arbitrary size, tested up to 30 loops. It is shown that the functional equation for Kloosterman moments determines much but not all of the structure of the L-series. In particular, for problems with odd numbers of Bessel functions, it misses a crucial feature captured in this work by novel and intensively tested conjectures. For the 9-Bessel problem, these lead to an astounding compression of data at the primes.

physics.gen-ph↗

arXiv2015

Tests of conjectures on multiple Watson values

I define multiple Watson values (MWVs) as iterated integrals, on the interval $x\in[0,1]$, of the 6 differential forms $A=d\log(x)$, $B=-d\log(1-x)$, $T=-d\log(1-z_1x)$, $U=-d\log(1-z_2x)$, $V=-d\log(1-z_3x)$ and $W=-d\log(1-z_4x)$, where $z_1=\gamma^2$, $z_2=\gamma/(1+\gamma)$, $z_3=\gamma^2/(1-\gamma)$ and $z_4=\gamma=2\sin(\pi/14)$ solves the cubic $(1-\gamma^2)(1-\gamma)=\gamma$. Following a suggestion by Pierre Deligne, I conjecture that the dimension of the space of ${\mathbb Z}$-linearly independent MWVs of weight $w$ is the number $D_w$ generated by $1/(1-2x-x^2-x^3)=1+\sum_{w>0}D_w x^w$. This agrees with 6639 integer relation searches, of dimensions up to $D_5+1=85$, performed at 2000-digit precision, for $w<6$.

hep-th↗

arXiv2015

Multiple Landen values and the tribonacci numbers

Multiple Landen values (MLVs) are defined as iterated integrals on the interval $x\in[0,1]$ of the differential forms $A=d\log(x)$, $B=-d\log(1-x)$, $F=-d\log(1-\rho^2x)$ and $G=-d\log(1-\rho x)$, where $\rho=(\sqrt{5}-1)/2$ is the golden section. I conjecture that the dimension of the space of ${\mathbb Z}$-linearly independent MLVs of weight $w$ is a tribonacci number $T_w$, generated by $1/(1-x-x^2- x^3)=1+\sum_{w>0}T_w x^w$, and that a basis is provided by all the words in the $\{A,G\}$ sub-alphabet that neither end in $A$ nor contain $A^3$. For $w<9$, I construct a much more efficient basis, for a MLV datamine, where no prime greater than 11 occurs in the denominators of 3,357,257 coefficients of rational reduction of 49,151 MLVs. Numerical data for 40 primitives then enable fast evaluation of all of these MLVs to 20,000 digits. The datamine provides reductions of Ap\'ery-type sums $A_w=\sum_{n>0}(-1)^{n+1}n^{-w}/{2n\choose n}$ and 6 ladder-combinations of depth-1 polylogarithms ${\rm Li}_w(\rho^p)=\sum_{n>0}\rho^{pn}n^{-w}$ with $p\in\{1,2,3,4,6,8,10,12,20,24\}$ and coefficients given by Landen, Coxeter and Lewin at $w=2$. I prove that the former evaluate to MLVs and conjecture that the latter do. Comparison is made between the properties of MLVs and multiple polylogarithms at roots of unity, encountered in the quantum field theory of the standard model of particle physics.

hep-th↗

arXiv2014

Multiple Deligne values: a data mine with empirically tamed denominators

Multiple Deligne values (MDVs) are iterated integrals on the interval $x\in[0,1]$ of the differential forms $A=d\log(x)$, $B=-d\log(1-x)$ and $D=-d\log(1-\lambda x)$, where $\lambda$ is a primitive sixth root of unity. MDVs of weight 11 enter the renormalization of the standard model of particle physics at 7 loops, via a counterterm for the self-coupling of the Higgs boson. A recent evaluation by Erik Panzer exhibited the alarming primes 50909 and 121577 in the denominators of rational coefficients that reduce this counterterm to a Lyndon basis suggested by ideas from Pierre Deligne. Oliver Schnetz has studied this problem, using a method from Francis Brown. This gave 2111, 14929, 24137, 50909 and 121577 as factors of the denominator of the coefficient of $\pi^{11}/\sqrt{3}$. Here I construct a basis such that no denominator prime greater than 3 appears in the result. This is achieved by building a datamine of 13,369,520 rational coefficients, with tame denominators, for the the reductions of 118,097 MDVs with weights up to 11. Then numerical data for merely 53 primitives enables very fast evaluation of all of these MDVs to 20000 digits. In the course of this Aufbau, six conjectures for MDVs are formulated and stringently tested.

hep-th↗