IBIS : Inverse BInomial sum Solver
Paul A.J.W. van Hoegaerden, Coenraad B. Marinissen, Wouter J. Waalewijn · 2025
Read IBIS : Inverse BInomial sum Solver by Paul A.J.W. van Hoegaerden, Coenraad B. Marinissen, Wouter J. Waalewijn (2025) 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-2506-19904.txt.
Introduction
Van Hoegaerden, Marinissen, and Waalewijn build a fast solver for stubborn sums in particle physics. Their 2025 paper deals with inverse binomial sums, which pop up when loop integrals are expanded. Standard tools grind slowly on them. Their method rests on new recursion rules drawn from telescoping series and repeated syncing steps.
Answers come out as S-sums, with the sum's upper limit kept in analytic form. The solver, named IBIS, runs as a program in the FORM system. Sums up to weight six finish in under a second. Tests against two general packages show the speed edge. Worked cases show where such sums arise, though some lie past the solved classes. The authors offer IBIS as a starting point for wider study.
Catalog record
| CREATOR | Paul A.J.W. van Hoegaerden, Coenraad B. Marinissen, Wouter J. Waalewijn |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-19904
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
IBIS : Inverse BInomial sum Solver Paul A.J.W. van Hoegaerden, Coenraad B. Marinissen, Wouter J. Waalewijn Abstract In higher-loop calculations, Mellin-Barnes representations are used to simplify the denominators encountered in Feynman parameter integrals. The contour integral of these representations yield sums over residues. We extend the classes of such sums that can be calculated, to include those involving inverse binomials. Our results are expressed in terms of so-called S S -sums, where the dependence on the upper limit of the sum is analytic. This is accomplished by deriving several new recursion relations, obtained from telescoping series and repeated synchronization. We make our results available through IBIS (“Inverse BInomial sum Solver”), a Form program to perform such inverse binomial sums. To illustrate the efficiency of our code: sums up to weight 6 can be carried out in less than a second. We show in an example how inverse binomial sums arise, though in many instances these are nested with binomial sums, beyond the cases studied here. Our work thus provides a starting point for studying such sums, that will open up new avenues in higher-loop calculations. 1 Introduction Collider physics has entered a precision era, driven by the demands of increasingly precise experimental measurements and enabled by substantial advances in theoretical calculations. An essential component of these theoretical calculations are QCD corrections, given the large size of the strong coupling, the fact that the LHC collides protons (QCD bound states) and that many analysis involve jets (QCD final states). Higher order perturbative calculations give important corrections to predictions of the rates of scattering processes and substantially reduce the theory uncertainty. The past two decades have brought significant advances, from automation of next-to-leading order (NLO) calculations Ellis:2011cr with a range of automated tools Berger:2008sj ; GoSam:2014iqq ; Bevilacqua:2011xh ; Cascioli:2011va ; Badger:2012pg , to next-to-next-to-leading order (NNLO) becoming the standard for 2\to 2 processes Gehrmann-DeRidder:2019ibf ; Czakon:2019tmo ; Boughezal:2015dva ; Gehrmann-DeRidder:2015wbt ; Campbell:2016lzl ; Chen:2019zmr ; Czakon:2015owf ; Catani:2019hip . Even first NNLO results for 2\to 3 processes Czakon:2021mjy and next-to-next-to-next-to-leading order (N 3 LO) results for 2\to 1 processes Anastasiou:2015vya ; Dulat:2018bfe ; Duhr:2020seh ; Chen:2021vtu have been achieved. Methodological advances in the calculation of two-loop amplitudes Laporta:2000dsw ; Gehrmann:1999as ; Henn:2013pwa and in the treatment of IR singular real radiation Binoth:2004jv ; Anastasiou:2003gr ; Catani:2007vq ; Gehrmann-DeRidder:2005btv ; Currie:2013vh ; Czakon:2010td ; Boughezal:2011jf ; Boughezal:2015dva ; Boughezal:2015eha ; Gaunt:2015pea ; Somogyi:2008fc ; DelDuca:2016ily were essential to this success, enabling predictions that are differential in the kinematics. To obtain reliable predictions across the full kinematic range requires resummation, which can be achieved by matching to parton showers. A specific ingredient in the evaluation of multi-loop integrals is Feynman parametrization, which however shifts computational challenges to difficult Feynman parameter integrals. These Feynman parameter integrals can be addressed by using Mellin–Barnes (MB) representations, rewriting them as contour integrals along the imaginary axis of the complex plane, involving ratios of Gamma functions. By closing the contour at infinity, these integrals are converted into infinite sums of residues via the residue theorem. Subsequent operations, as outlined in Banik:2023rrz , include resolving 1/\epsilon singularities of dimensionally ( d=4-2\epsilon ) regularized Feynman integrals Smirnov:1999gc ; Tausk:1999vh ; Smirnov:2009up , finding their analytic expressions in terms of hypergeometric functions Kalmykov:2020cqz or multiple polylogarithms Vollinga:2004sn , performing numerical integration Dubovyk:2019krd , etc. This strategy has been successfully applied in the analytical evaluation of Feynman integrals Smirnov:2012gma ; Dubovyk:2022obc , and several computer codes have been developed to facilitate the construction of MB representations and the subsequent summation of residues Czakon:2005rk ; Gluza:2007rt ; Smirnov:2009up ; Ochman:2015fho , which are summarized in Belitsky:2022gba . In this paper, we focus on the evaluation of MB integrals in terms of nested sums. Over the years, general classes of nested sums have been investigated, and their symbolic summation has led to the development of algorithms suitable for computer algebra systems. In most cases, the building blocks of these algorithms are the so-called S - and Z -sums, which generalize harmonic sums and (multiple) zeta values. Harmonic sums are the fundamental components of the Summer Vermaseren:1998uu package in Form Vermaseren:2000nd ; Ruijl:2017dtg . These sums frequently appear in single scale Feynman integrals, for which a Mathematica implementation is also available Ablinger:2014rba . For Feynman integrals that depend on multiple scales, one can use the Form implementation XSummer Moch:2005uc , or an implementation within the GiNaC framework Bauer:2000cp ; Vollinga:2005pk . The availability of both the Summer and XSummer packages, coupled with Form ’s exceptional speed and efficiency in managing large expressions, are the primary reasons we selected Form for our implementation. A particularly interesting class of sums encountered in multi-loop calculations using the MB representation takes the following form \sim\sum_{j=1}^{n-1}\frac{1}{\binom{n}{j}}...\,\,, (1) where the dots denote additional terms, e.g. harmonic sums. In this paper, we refer to such sums as inverse binomial sums . In particular, the classical identity \displaystyle\sum_{j=0}^{n}\frac{1}{\binom{n}{j}}=\frac{n+1}{2^{n+1}}\sum_{j=1}^{n+1}\frac{2^{j}}{j}\,, (2) which the simplest example of the sums we consider here, has been rigorously proved in earlier works gould1972combinatorial ; Rockett01121981 ; pla1997sum ; Trif01022000 ; SURY1993351 . Further research has explored more intricate identities MANSOUR2002196 ; Sury2004 ; sym13112002 , including those incorporating harmonic sums 2014arXiv1411.6916J ; 2018arXiv180603022B . In this paper, we extend these results by rewriting the basis set of inverse binomial sums in tab. 1 in terms of so-called S -sums, which we make available through an accompanying Form packaged called IBIS (“Inverse BInomial sum Solver”). The parameters defining the structure of these inverse binomial sums are summarized in tab. 2 . Specifically, tab. 2 provides the numerical or symbolic ranges and constraints for these variables. Fixed sign 0 harmonic sums \sum_{j=\text{}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{1}{(j+c)^{k}} 1 harmonic sum \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{1}{(j+c)^{k}}S_{p_{1},...,p_{s}}(n-j) and \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{1}{(j+c)^{k}}S_{q_{1},...,q_{r}}(j) 2 harmonic sums \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{1}{(j+c)^{k}}S_{p_{1},...,p_{s}}(n-j)S_{q_{1},...,q_{r}}(j) Sign altering 0 harmonic sums \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{(-1)^{j}}{(j+c)^{k}} 1 harmonic sum \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{(-1)^{j}}{(j+c)^{k}}S_{p_{1},...,p_{s}}(n-j) and \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{(-1)^{j}}{(j+c)^{k}}S_{q_{1},...,q_{r}}(j) 2 harmonic sums \sum_{j=\text{max}\{1,-c+1\}}^{n-1}\frac{j!(n-j)!}{n!}\frac{(-1)^{j}}{(j+c)^{k}}S_{p_{1},...,p_{s}}(n-j)S_{q_{1},...,q_{r}}(j) Table 1 : The basis set of inverse binomial sums solvable using IBIS . The S -sums appearing here are defined in eq. ( 4 ). Using procedures provided by IBIS and Summer a larger set can be solved, as discussed in the IBIS manual. Variable Set of possible values c \mathbb{Z} k \mathbb{N}_{0} p_{i},q_{j} \mathbb{Z}_{\neq 0} s , r \mathbb{N}_{0} n Symbolic: n Numerical: \mathbb{N}_{>1} Table 2 : Possible values for the parameters in tab. 1 that are supported by IBIS . To illustrate the relevance of these sums, we provide a concrete example in app. A , which leads to the following sum of residues \displaystyle-\sum_{k_{3}=1}^{\infty}\frac{1}{1+k_{3}}\sum_{k_{1}=1}^{k_{3}-1}\frac{1}{\binom{k_{3}}{k_{1}}}\frac{(-1)^{k_{1}}}{1+k_{1}}\sum_{k_{2}=1}^{\infty}\frac{\binom{k_{1}+k_{2}}{k_{2}}}{(2+k_{2}+k_{3})(1+k_{2})}S_{2}(1+k_{2}+k_{3})\,. (3) In this expression, one must first evaluate an infinite sum over k_{2} that includes a binomial coefficient before the inverse binomial sum in k_{1} can be computed. Although such infinite binomial sums have been studied before Ablinger:2014wca ; Ablinger:2014bra ; Weinzierl:2004bn ; Jegerlehner:2002em ; Fleischer:1998nb , they have not been solved in a general form. The sum in eq. ( 3 ) is more complicated than what IBIS can currently handle. Nevertheless, we believe that IBIS provides an important first step towards addressing such sums and a baseline for future exploration. In the outlook (sec. 6 ), we describe how IBIS can be extended to accommodate these binomial coefficients. In addition to the IBIS code, a detailed manual and concrete examples are available at