The elliptic three-loop integrals of hadronic vacuum polarization in chiral perturbation theory
Laurent Lellouch, Alessandro Lupo, Mattias Sjö, and Pierre Vanhove · 2026
Read The elliptic three-loop integrals of hadronic vacuum polarization in chiral perturbation theory by Laurent Lellouch, Alessandro Lupo, Mattias Sjö, and Pierre Vanhove (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-15252.txt.
Introduction
Lellouch and co-authors detail the integrals behind a three-loop calculation of hadronic vacuum polarization. Each integral gets a full treatment. They explain how to compute every one and lay out the math behind the method. The work ends in a fast numerical code. That code gives accurate values for any complex photon virtuality. The account supports the companion physics result published earlier.
Catalog record
| CREATOR | Laurent Lellouch, Alessandro Lupo, Mattias Sjö, and Pierre Vanhove |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 291763 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.15252
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The elliptic three-loop integrals of hadronic vacuum polarization in chiral perturbation theory Laurent Lellouch, Alessandro Lupo, Mattias Sjö, and Pierre Vanhove Abstract This work presents a detailed account of the Feynman integrals required for the three-loop hadronic vacuum polarization calculation performed in 33 . We explain how to compute each of the three-loop integrals, and outline the mathematical framework underlying their evaluation. This culminates in a practical numerical implementation that enables fast and accurate evaluation of these integrals for arbitrary complex values of the photon virtuality. 1 Introduction In the perturbative calculation of any particle physics observable, nontrivial loop integrals will eventually appear at a sufficiently high order in the perturbative series. Here, that order is three-loop or N 3 LO in chiral perturbation theory 21 ; 22 , and the observable is the hadronic vacuum polarization, a key contribution to the very low-energy QCD corrections to the magnetic moments of leptons. We performed this calculation in a recent paper 33 , to which we defer the reader for a more complete picture of the physical background. However, several novel loop integrals that arise there are sufficiently complicated that we have chosen to dedicate a paper to them. As explained in ref. 33 , vacuum polarization receives contributions from six elliptic master integrals denoted E_{i}(d;t) , evaluated in their finite two-dimensional form E_{i}(2;t) , or in the finite part \bar{E}_{i}(4;t) of their four-dimensional form. These master integrals belong to the broader class of elliptic integrals, i.e., integrals that cannot be expressed in terms of logarithms, polylogarithms, or multiple polylogarithms (i.e., iterated integrals of logarithms and rational functions). Instead, they require elliptic functions. For any given loop integral, the ideal case is to have a closed-form solution, but for integrals as complicated as these, one normally has to resort to numerical methods. Although a complete overview of the state of the art is beyond the scope of this work, we benchmark our solutions against three existing codes: pySecDec 9 ; 10 ; 25 ; 26 , a mature general-purpose loop integrator based on sector decomposition; AMFlow 34 , a more novel approach leveraging integration-by-parts relations to obtain much higher precision than pySecDec at the cost of much slower calculations; and FeynTrop 8 ; 7 , another novel approach based on tropical decomposition, which is capable of very quickly obtaining a few digits of precision for finite integrals. There also exist methods based on numerical solutions to the differential equations satisfied by Feynman integrals 27 ; 1 ; 34 , which require initial conditions obtained via other numerical or analytical methods, although an alternative approach that bypasses this requirement has been proposed 14 . Lastly, some elliptic loop integrals are known in terms of elliptic polylogarithms, for which there exist implementations 44 ; 15 based on GiNaC 4 . However, these implementations lack analytic continuation beyond the radius of convergence of the defining power series, although this has been addressed through a representation in terms of iterated integrals of Eisenstein series 19 . In this work, we present essentially closed-form solutions for four of the six master integrals— E_{1}(2;t) , E_{2}(2;t) , E_{3}(2;t) and \bar{E}_{4}(4;t) —by extending and practically implementing the formulation of E_{1}(2;t) in terms of elliptic trilogarithms evaluated at sixths roots of unity 5 . While this still requires numerical calculation of this special case of elliptic polylogarithms, the speed and precision of this approach are orders of magnitude greater than those of an approach involving numerical integration. For the remaining two master integrals— \bar{E}_{5}(4;t) and \bar{E}_{6}(4;t) —we present an implementation in terms of contour integrals over E_{1}(2;t) . Although it falls short of a closed-form solution, this specialized implementation offers much higher precision and stability than general-purpose codes while maintaining speed comparable to pySecDec . All our results are valid for arbitrary complex values of the photon virtuality (with occasional exceptions near zero or at the two- and four-pion thresholds) and have numerical precision comfortably larger than that required by experiment. E_{1}(2;t) is the three-loop all-equal-masses sunset integral, and has been extensively studied. For real virtualities below threshold, it has a Bessel integral representation 24 ; 2 ; 5 , and for all virtualities, it has the aforementioned one in terms of elliptic trilogarithms 5 . These are natural functions of a nome q instead of the photon virtuality t , 1 1 1 Note the difference between the unphysical nome, for which q is the standard symbol, and the photon momentum, which is also commonly called q . To reduce the potential for confusion, here we call the photon momentum p and mainly express it in terms of the photon virtuality in units of the pion mass, t\coloneq p^{2}/M^{2} . so one major component of our extension of ref. 5 is the reliable mapping from t to q . Another is the straightforward implementation of E_{2}(2;t) and E_{3}(2;t) , which relate to E_{1}(2;t) via differentiation and therefore involve elliptic polylogarithms of lower weight. \bar{E}_{4}(4;t) is then simply a linear combination of these and non-elliptic integrals. Lastly, we express \bar{E}_{5}(4;t) and \bar{E}_{6}(4;t) as integrals of E_{1}(2;t) , indicating that these would be expressible in terms of elliptic polylogarithms of higher weight, although we are unable to find this expression and therefore resort to numerical integration. The remainder of this paper is organised as follows. In section 2 we recall the definition of the master integrals that we will evaluate numerically. In section 3 we give the expressions of the masters E_{1,2,3} in terms of elliptic polylogarithms as the sixth root of unity, and provide two alternative formulations of these polylogarithms in terms of Eisenstein series with a mod 6 character. We also discuss the properties of the map between the photon virtuality t and the nome q , needed for the numerical evaluation. Section 4 is devoted to the evaluation of the masters E_{5} and E_{6} , which involves integrals over the master E_{1} . Our results for the numerical evaluation of the master integrals and the vacuum polarization are given in section 5 , and we conclude with section 6 . In appendix A we give the small- t expansions of the elliptic master integrals. 2 Definitions Here, we briefly restate the notation used for the integrals in 33 , so that we may define the objects of interest. Letting p be the momentum of the external photon, with t\coloneq p^{2}/M^{2} its virtuality normalized to the pion mass M , 2 2 2 Since ref. 33 assumes the isospin limit, there is no mass difference between charged and neutral pions. The common pion mass is normally taken as M\approx 135\penalty\ \mathrm{MeV} . we choose the basis of loop momenta \{k_{j}\}_{j=1}^{9}=\left\{\ell_{1},\ell_{2},\ell_{3},\quad\ell_{1}-p,\ell_{2}-p,\ell_{3}-p,\quad\ell_{1}+\ell_{3},\ell_{2}+\ell_{3},\ell_{1}-\ell_{2}\right\} (1) so that all three-loop integrals in d dimensions (with d an integer minus 2\epsilon ) may be written in the form \big[e^{\gamma_{E}}M^{2}\big]^{-3\epsilon}I_{\vec{\nu}}^{(d)}(t;M^{2})=\int\frac{\mathrm{d}^{d}\ell_{1}\;\mathrm{d}^{d}\ell_{2}\;\mathrm{d}^{d}\ell_{3}}{(2\pi)^{3d/2}}\prod_{j=1}^{9}\frac{1}{\big(k_{j}^{2}-M^{2}+i\varepsilon\big)^{\nu_{j}}}\,, (2) where \vec{\nu}\in\mathbb{Z}^{9} determines which combination of propagators appears in the integral, and where the normalization factor (involving the Euler-Mascheroni constant \gamma_{E} ) ensures that the integral has integer mass dimension. Of the 11 three-loop master integrals appearing in 33 (as determined through the Laporta algorithm 29 implemented in LiteRed 2 31 ; 32 ), four can be written as products of one-loop integrals, allowing