Non-planar hexa-box with two massive legs
The two-loop non-planar hexa-box with two off-shell legs of equal virtuality on non-adjacent corners and three massless legs, a master-integral family for the production of two vector bosons and a jet. Thirteen of its master integrals are new and are given along kinematic paths, with the three top-sector masters in closed form as iterated integrals through the finite part.
The integral
The family consists of the two-loop five-point integrals with two external legs off shell at one common virtuality, $p_1^2=p_3^2=m^2$, and three massless legs,
$$I[\nu_1,\dots,\nu_{11}] \;=\; \int d^d k_1\, d^d k_2\; \frac{D_9^{-\nu_9}\, D_{10}^{-\nu_{10}}\, D_{11}^{-\nu_{11}}}{D_1^{\nu_1}\, D_2^{\nu_2}\, D_3^{\nu_3}\, D_4^{\nu_4}\, D_5^{\nu_5}\, D_6^{\nu_6}\, D_7^{\nu_7}\, D_8^{\nu_8}}\,, \qquad d = 4-2\varepsilon,$$
up to an overall normalization, with loop momenta $k_1, k_2$, the eight massless propagators
$$D_1 = k_1^2,\qquad D_2 = k_2^2,\qquad D_3 = (k_1-k_2+p_4)^2,\qquad D_4 = (k_2-p_4)^2,$$
$$D_5 = (k_1+p_1+p_2+p_3+p_4)^2,\qquad D_6 = (k_1-k_2+p_1+p_2+p_3+p_4)^2,$$
$$D_7 = (k_1-k_2+p_2+p_3+p_4)^2,\qquad D_8 = (k_1-k_2+p_3+p_4)^2,$$
and the three irreducible numerators $D_9=(k_1+p_1)^2$, $D_{10}=(k_2+p_2)^2$, $D_{11}=(k_2+p_3)^2$, which appear only with $\nu_i\le 0$. The external momenta satisfy $p_1+p_2+p_3+p_4+p_5=0$ with $p_2^2=p_4^2=p_5^2=0$, and the kinematics has six scales, the five cyclic invariants $s_{ij}=(p_i+p_j)^2$, namely $s_{12}, s_{23}, s_{34}, s_{45}, s_{15}$, and the common virtuality $m^2$; equal virtuality covers $W^+W^-$ and $ZZ$ plus a jet, while the $WZ$ case has seven scales. The graph is non-planarno redrawing puts all eight lines in a plane without a crossing while keeping the five external legs on the outside, and at equal virtuality the reflection $\sigma$ that exchanges $p_1\leftrightarrow p_3$ is a symmetry of the family.
The family has 98 master integralsthe finite set of integrals to which every member of the family reduces by integration-by-parts identities in 62 sectors. Of these, 76 belong to known one-mass five-point and two-mass four-point sub-topologies and enter only as boundary data. The remaining 22 masters fall under $\sigma$ into six sector orbits, the top sectorthe sector in which all eight propagators of the diagram are present and five subsector orbits, with 13 representatives: $m_0, m_1, m_2$ in the top sector and ten in the subsector orbits. The quantities computed are these masters as Laurent series in $\varepsilon$ along straight segments in the six-scale space: the ten subsector representatives from their leading pole through order $\varepsilon^{2}$, and $m_0, m_1, m_2$ from $\varepsilon^{-4}$ through the finite part $\varepsilon^{0}$.
At a glance
- Process or family: two-loop non-planar hexa-box with off-shell legs $p_1^2=p_3^2=m^2$ on non-adjacent corners and $p_2, p_4, p_5$ massless (two vector bosons plus a jet)
- Loops and legs: two loops; five-point, six scales $(s_{12},s_{23},s_{34},s_{45},s_{15};m^2)$
- Master integrals: 98 (top sector: 3)
- Function class: polylog; genus-zero maximal cut, square root of a quartic $Q_4(t)$ on segment I
- Singular points or alphabet: 437 letters (404 rational, 33 algebraic in the five-point Gram root $\sqrt{\mathrm{Gram}_5}$) plus sixteen new letters outside that set; along the path, the zeros of $\mathrm{Gram}_5$
- Status: new
- Paper: Polylogarithmic, elliptic, and Calabi–Yau Feynman integrals from a hybrid bootstrap, section 4.8
Two-loop five-point integrals with two massive external legs are the master integrals for the production of two vector bosons and a jet at next-to-next-to-leading order in quantum chromodynamics (QCD), a correction that the 2023 Les Houches wishlist lists as wanted and not yet available; the two-loop integrals are the missing ingredient. Three neighboring sets of integrals are known: the complete planar two-mass five-point set of Abreu, Chicherin, Sotnikov and Zoia (2024), with one planar two-mass penta-box family also solved from its differential equations by Jiang, Liu, Xu and Yang (2024); the non-planar hexa-boxes with one off-shell leg of Abreu, Ita, Page and Tschernow (2022) and of Kardos, Papadopoulos, Smirnov, Syrrakos and Wever (2022), completed into the one-mass pentagon functions by Abreu et al. (2024); and the massless pentagon functions of Chicherin, Henn and Mitev (2018). No two-mass non-planar family appears to have been computed before. The hexa-box graph admits inequivalent placements of the two massive legs; the one solved here, equal masses on the non-adjacent legs $p_1$ and $p_3$, is the simplest.
A Landau analysis of the two-mass hexa-box families by the principal $A$-determinant of Mizera and Telen (2022) gives exactly two square-root letters: the parity-odd Gram root $\sqrt{\mathrm{Gram}_5}$ of the external momenta, and a two-mass threshold root in the channel that contains both massive legs, present only when those legs are adjacent and therefore absent here. Both roots rationalize, which makes the maximal cutthe integral with every propagator put on shell, which isolates the geometry of the top sector of the top sector a genus-zero curve and the family polylogarithmic. At $m^2=0$ the graph reduces to one of the known one-mass non-planar hexa-boxes, which supplies boundary data.
Closed form
Along each path segment every master is an iterated integrala repeated integral of one-forms along the path, labeled by the ordered list of the forms, the word of d-log formslogarithmic differential forms $d\log(\text{letter})$, whose iterated integrals are the polylogarithms of multi-loop calculations and algebraic one-forms, with letters drawn from an alphabet in the six scales. The alphabet starts from the letters of the one-mass non-planar hexa-box, the massless pentagon and the two-mass four-point integrals together with twelve new two-mass five-point letters, 437 in all and invariant under $\sigma$; the 33 algebraic letters contain $\sqrt{\mathrm{Gram}_5}$. Sixteen further letters lie outside these 437. Writing $P_{abcd}=\mathrm{tr}_4(p_a p_b p_c p_d)$ for the Dirac trace of four external momenta, the fifteen new parity-odd letters are the ratios
$$\frac{P_{abcd}-\sqrt{\mathrm{Gram}_5}}{P_{abcd}+\sqrt{\mathrm{Gram}_5}}\,,$$
each with $P_{abcd}^2-\mathrm{Gram}_5$ factoring over the rational alphabet and each reducing at $m^2\to 0$ to a known massless odd pentagon letter. The sixteenth, called $X_8$ in the expression file, is parity-even,
$$X_8 \;=\; -\,\frac{4\,\mathrm{Gram}_3(p_3,p_4,p_5)}{s_{45}}\,,$$
with $\mathrm{Gram}_3$ the Gram determinant of the three momenta listed. Five more irreducible quadratic polynomials that appear in the differential equation along the path segments are candidates for further letters; whether the alphabet is complete for the whole family remains open.
The masters are solved along three one-dimensional segments through the six-scale space, not over the whole space. Segment I runs from the Euclidean base point $(s_{12},s_{23},s_{34},s_{45},s_{15};m^2)=(-3,-5,-7,-2,-11;1)$ to the end point $(-3,-5,-7,-11,-17;2)$, linear in a parameter $t\in[0,1]$ and passing above the poles of the connection (the matrix of the differential equation) in the complex $t$ plane, and carries all 98 masters. The top sector couples to all 98, and the rational connection along segment I reaches polynomial degree 83 in $t$. Segment II varies all six scales, also detouring around poles of the connection, and carries only the five subsector orbits. Segment III starts at the same base point, varies $s_{12}$, $s_{23}$, $s_{34}$ at fixed $s_{45}$, $s_{15}$, $m^2$ and carries all 98 masters at a fixed numerical dimension rather than as a series in $\varepsilon$. At the base point all 98 masters are evaluated numerically with the auxiliary-mass-flow method of Liu and Ma (2022); the masters along each segment are obtained by transportsolution of the differential equation along a path, starting from a point where the integrals are known from the values at one point of the segment, with the connection an exact rational function of the segment parameter. On segment I no rational change of basis brings the 98-master system to a canonical $\varepsilon$-form in the sense of Henn (2013): half-integer exponents appear at irreducible quadratic letters, and an $\varepsilon$-factorized basis would carry the square roots of those letters as prefactors. Instead of rotating to such a basis, the connection and the masters are expanded in powers of $\varepsilon$ in the rational basis and the system is solved order by order in $\varepsilon$ by variation of parameters, the solution at each order entering the next as a source.
The top sector
On segment I each Laurent coefficient of the three top-sector masters is a finite sum
$$m_i[\varepsilon^{k}](t) \;=\; \sum C\,\sqrt{Q_4(t)}\;R(t)\,G_{\text{word}}(t)\,,$$
where $m_i[\varepsilon^{k}]$ denotes the coefficient of $\varepsilon^{k}$ in $m_i$, each $C$ is a boundary constant, each $R(t)$ an exact rational function, each $G_{\text{word}}(t)$ an iterated integral over d-log, rational and algebraic kernels in $t$ labeled by a word in the path letters, and $\sqrt{Q_4(t)}$, the algebraic prefactor, is the square root of the quartic
$$Q_4(t) \;=\; 3721\,t^4+6724\,t^3+11824\,t^2+16400\,t+7040\,,$$
whose zeros govern the homogeneous solutions of the top block: near each zero of $Q_4$ they scale with exponents $(\tfrac12,\tfrac12,0)$, so two of the three are proportional to $\sqrt{Q_4}$ there. The quartic is the five-point Gram determinant restricted to the segment, $Q_4(t)=\mathrm{Gram}_5(s_{ij}(t))$ identically in $t$ with $s_{ij}(t)$ the invariants along segment I, and its zero set is the second-type (Gram) Landau locus of the eight-propagator sector. Because $\sqrt{\mathrm{Gram}_5}$ restricts on the straight line to the square root of an irreducible quartic, a genus-one radical, no representation of the coefficients by Goncharov polylogarithms in $t$ is found; this is a property of the straight-line parametrization, and a polylogarithmic representation in the six invariants, where the Gram root rationalizes, may still exist.
At order $\varepsilon^{-4}$ the coefficients of the three top-sector masters are rational functions of the kinematics. Along segment I they are
$$m_0[\varepsilon^{-4}](t) \;=\; -\,\frac{12840\,t^3+40093\,t^2+34960\,t+9177}{105\,D(t)}\,,$$
$$m_1[\varepsilon^{-4}](t) \;=\; \frac{183840\,t^4+915198\,t^3+1564973\,t^2+1055732\,t+240492}{420\,D(t)}\,,$$
$$m_2[\varepsilon^{-4}](t) \;=\; -\,\frac{1227360\,t^5+8249742\,t^4+21107169\,t^3+25290010\,t^2+13922056\,t+2808708}{420\,D(t)}\,,$$
where
$$D(t) \;=\; m^2\,(-s_{15})\,s_{45}^2\,s_{35}\,s_{14} \;=\; (t+1)(6t+11)(9t+2)^2(10t+7)(16t+9)$$
on the segment, with $s_{35}=7+10t$, $s_{14}=9+16t$ and $105=-s_{12}s_{23}s_{34}$. They satisfy the top rows of the path differential equation identically in $t$. At the end point $t=1$ they give
$$m_0[\varepsilon^{-4}] = -\tfrac{571}{1079925}, \qquad m_1[\varepsilon^{-4}] = \tfrac{46591}{8639400}, \qquad m_2[\varepsilon^{-4}] = -\tfrac{854177}{8639400}.$$
The $\varepsilon$-factorized basis with square-root prefactors, which is not used for the transport, can still be written at a single point. At the end point it is $\hat g = (\sqrt{45709}\,g_0,\ \sqrt{45709}\,g_1,\ g_2)$, with $(g_0,g_1,g_2)$ the rotation of $(m_0,m_1,m_2)$ into a basis whose $\varepsilon$ expansion has uniform transcendental weight at each order and $\sqrt{45709}=\sqrt{Q_4(1)}$ the path radical at the end point, and its leading coefficients are
$$\hat g_0[\varepsilon^{-4}] = \tfrac{123164227}{3427200}, \qquad \hat g_1[\varepsilon^{-4}] = \tfrac{203357363}{3427200}, \qquad \hat g_2[\varepsilon^{-4}] = -\tfrac{940511}{1713600},$$
with denominator $3427200 = 2^7\, 3^2\, 5^2\, 7\, 17$, every prime factor of which occurs in the letters of the alphabet at the end point. At order $\varepsilon^{-3}$ the imaginary parts are rational multiples of $\pi$,
$$\frac{\mathrm{Im}\,\hat g_0[\varepsilon^{-3}]}{\pi} = \tfrac{33402719852867}{1005262876800}, \qquad \frac{\mathrm{Im}\,\hat g_1[\varepsilon^{-3}]}{\pi} = \tfrac{75672918098083}{1005262876800}, \qquad \frac{\mathrm{Im}\,\hat g_2[\varepsilon^{-3}]}{\pi} = -\tfrac{375001639771}{502631438400},$$
with denominator $1005262876800 = 2^7\, 3^4\, 5^2\, 7\, 13\, 17\, 23\, 109$, which has the same property. The real parts at order $\varepsilon^{-3}$ and the $\varepsilon^{-2}$, $\varepsilon^{-1}$ and $\varepsilon^{0}$ coefficients are given by the same sum with numerical boundary constants, with one more iterated integration at each order, exact rational kernels and the prefactor $\sqrt{Q_4}$; the part odd in $\sqrt{Q_4}$ is of order one at $t=1$, so the root cannot be removed by rescaling the three masters. Beyond order $\varepsilon^{-4}$ no symbol is given: on the segment the kernels carry the genus-one radical $\sqrt{Q_4(t)}$, and most of the boundary constants are known only numerically.
Boundary constants
The order-by-order top-sector solution carries 447 boundary constants $C$, each defined at the base point of segment I as a transport integral in an auxiliary mass starting from vacuum integrals known in closed form in terms of $\Gamma$ functions. Six of the 447 have closed forms, three rationals and three rational multiples of $\pi$; they are the constants behind the exact order-$\varepsilon^{-4}$ functions and the order-$\varepsilon^{-3}$ imaginary parts given above. For the rest an integer-relation (PSLQ) search finds no combination of the expected transcendental constants. Ten of the 447 are also fixed, without the auxiliary-mass transport, by requiring the solution to be regular at the five apparent singular points that the segment-I contour passes above, imposed as residue conditions.
The massless limit
At $m^2=0$ the representative master of the first subsector orbit, written $m_{219}$ (its label in the expression file and in the script output), collapses onto an integer combination of the massless non-planar pentagon basis of Chicherin, Gehrmann, Henn, Lo Presti, Mitev and Wasser, divided by $\sqrt{\Delta}$:
$$\varepsilon^4 e^{2\varepsilon\gamma_E}\, m_{219}\Big|_{m^2=0} \;=\; \frac{1}{\sqrt{\Delta}}\Big({-}6\,I_{55} - 3\,I_{3} - 3\,I_{18} + 3\,I_{32} - 3\,I_{44} + 3\,I_{51} - 3\,I_{52}\Big),$$
where $\gamma_E$ is the Euler–Mascheroni constant, the $I_k$ are elements of that d-log basis in its authors' numbering, and $\Delta$ is the massless five-point Gram determinant, $\mathrm{Gram}_5|_{m^2=0}=\Delta$. One subsector orbit vanishes identically at $m^2=0$, so the massless limit gives no normalization for it; the normalization comes instead from requiring its diagonal block to be $\varepsilon$-factorized and from the relation $\sigma$ imposes between the two sectors of the orbit.
The complete expression, with the rational $\varepsilon^{-4}$ functions, the exact kernels, the named constants and the transport representations of the five subsector orbits, is in the expression file inside the bundle under Evaluator below.
Checks
Checks
| master, order | value at $t=1$ | digits |
|---|---|---|
| $m_0$, $\varepsilon^{-4}$ | $-571/1079925 = -0.0005287404217\ldots$ | 70 |
| $m_0$, $\varepsilon^{0}$ | $-0.001651851923\ldots - 0.03336786498\ldots\,i$ | 56 |
| $m_1$, $\varepsilon^{0}$ | $-0.1330911915\ldots + 0.5446086713\ldots\,i$ | 56 |
| $m_2$, $\varepsilon^{0}$ | $3.081679200\ldots - 12.56699474\ldots\,i$ | 56 |
| $m_{219}$, $\varepsilon^{0}$ | $0.2101893763\ldots + 0.6174089231\ldots\,i$ | 49 |
| $m_{219}$, $\varepsilon^{2}$ | $0.1816567917\ldots - 0.08890601826\ldots\,i$ | 40 |
All rows are at $t=1$, the end point of segment I, a point used neither in a fit nor to fix a boundary value. The value is the closed form; the last column is the number of leading digits to which it agrees with an independent auxiliary-mass-flow evaluation (AMFlow) of the same master integral there.
Evaluator
Evaluator
- hexabox-evaluate.py: evaluates the three top-sector masters and the first subsector orbit along segment I from the exact path differential equation;
python3 hexabox-evaluate.pyprints the comparison of the Checks table at the end point. - The two-mass hexa-box bundle (zip, 14.3 MB) — the expression file, the path differential equation and boundary values the evaluator reads, its helper modules, the reference values at the end point and at a second point, the regeneration records of the 447 boundary constants and checksums; the zip also contains the script. The inputs of the higher-order and boundary-regeneration options (
--kmax 4,--kmax 5,--boundary-regenerate) and the Path-C self-check are a second archive (zip, 5.6 MB) that unpacks into the same folder.
Python 3 with mpmath, sympy and python-flint; the script reads its data files and helper modules from the unzipped bundle folder, so run it there.
Tools
| tool | role |
|---|---|
| Kira | integration-by-parts reduction to the master integrals and their differential equations along the path segments |
| Ratfit | reconstruction of the path connection as exact rational functions from numerical samples |
| Vopclose | the variation-of-parameters solution of the top sector, order by order in $\varepsilon$, and the transport of the subsector orbits |
| Leviathan | Euler-characteristic computation of the Landau singularities; identifies $Q_4$ with the Gram-determinant locus |
| Lockpick | integer-relation searches on the boundary constants |
| AMFlow | the boundary vector at the base point and the independent numerical evaluations in the Checks table |
Same family
The hexa-box is the only two-loop five-point family among the thirty integrals; the placements with the two massive legs adjacent, whose alphabet carries the two-mass threshold root, and with generic masses have not been computed in closed form. The closest collider relative among the thirty is the non-planar branch of two-loop W-pair production, a four-point family with internal top-quark lines and an elliptic sector.
The paper
- Polylogarithmic, elliptic, and Calabi–Yau Feynman integrals from a hybrid bootstrap (PDF) — Matthew D. Schwartz.
References
| S. Abreu, D. Chicherin, V. Sotnikov and S. Zoia, Two-loop five-point two-mass planar integrals and double Lagrangian insertions in a Wilson loop (2024) [arXiv:2408.05201] | the complete planar two-mass five-point integrals, the planar neighbors of this family |
| S. Abreu, H. Ita, B. Page and W. Tschernow, Two-loop hexa-box integrals for non-planar five-point one-mass processes, JHEP 03 (2022) 182 [arXiv:2107.14180] | the non-planar hexa-boxes with one off-shell leg, whose alphabet supplies most of the letters and whose $m^2=0$ limit supplies boundary data |
| A. Kardos, C. G. Papadopoulos, A. V. Smirnov, N. Syrrakos and C. Wever, Two-loop non-planar hexa-box integrals with one massive leg, JHEP 05 (2022) 033 [arXiv:2201.07509] | the one-mass non-planar hexa-boxes from the simplified differential-equation approach |
| D. Chicherin, J. Henn and V. Mitev, Bootstrapping pentagon functions, JHEP 05 (2018) 164 [arXiv:1712.09610] | the massless pentagon functions, whose alphabet is part of the union used here |
| S. Mizera and S. Telen, Landau discriminants, JHEP 08 (2022) 200 [arXiv:2109.08036] | the principal $A$-determinant, which gives the two square-root letters of the two-mass hexa-box families |
| J. M. Henn, Multiloop integrals in dimensional regularization made simple, Phys. Rev. Lett. 110 (2013) 251601 [arXiv:1304.1806] | the canonical $\varepsilon$-form of differential equations, obstructed here on the straight path by half-integer exponents |
| X. Liu and Y.-Q. Ma, AMFlow: a Mathematica package for Feynman integrals computation via auxiliary mass flow (2022) [arXiv:2201.11669] | the auxiliary-mass-flow method, source of the boundary vector at the base point and of the numerical values in the Checks table |