QED light-by-light graphs with a kite self-energy

The content on this page was written by AI under human supervision.

Three-loop light-by-light scattering in QED with the electron mass kept: one electron line of the one-loop box carries the two-loop crossed-photon self-energy (the kite topology), and the electron propagator adjoining the insertion appears either once or squared. Both integrals are new and are given as single dispersion integrals over the squared mass of the dressed line.

The integral

Feynman diagram: a rectangular electron loop drawn as red double lines with an orange wavy photon attached at each of its four corners; the bottom side of the rectangle carries four extra vertices, joined in interleaved pairs by two orange photon arcs, one arching above the line and one hanging below it, so that the bottom side is cut into five electron segments
Red double lines are electron propagators of mass $m$, orange wavy lines are photons and dots are vertices. The bottom line of the one-loop box carries the crossed-photon self-energy: the electron emits two photons and reabsorbs them in the same order. The two outer segments of that line, the stubs, carry the same loop momentum and count as one propagator squared. The graph has ten lines and eight vertices, each vertex joining two electron lines and one photon.

Light-by-light scattering, $\gamma\gamma\to\gamma\gamma$, is the interaction of four massless on-shell photons ($p_i^2=0$) through loops of virtual electrons of mass $m$. With all momenta incoming the Mandelstam invariantsthe Lorentz-invariant combinations of external momenta that a scattering amplitude can depend on are $s=(p_1+p_2)^2$, $t=(p_1+p_3)^2$ and $u=-s-t$; the two planar channels of the box are $s$ and $u$. The electron mass sets the scale, and $m^2=1$ in the numerical work. Every integral in the family of this graph has the form

$$I[\nu_1,\ldots,\nu_n](s,t;m^2)\;=\;\int d^dl_1\,d^dl_2\,d^dl_3\;\prod_{j=1}^{n}\frac{1}{D_j^{\nu_j}},\qquad d=4-2\varepsilon,$$

up to an overall normalization convention. For this graph $n=9$; writing the loop momentum of the box as $\ell=l_1$ and those of the insertion as $k_1=l_2$, $k_2=l_3$, the propagators are

$$ \begin{aligned} D_1&=(\ell+p_1)^2-m^2, & D_2&=(\ell+p_1+p_2)^2-m^2, & D_3&=(\ell+p_1+p_2+p_3)^2-m^2,\\ D_4&=\ell^2-m^2, & D_5&=k_1^2-m^2, & D_6&=k_2^2-m^2,\\ D_7&=(\ell-k_1-k_2)^2-m^2, & D_8&=(\ell-k_1)^2, & D_9&=(\ell-k_2)^2. \end{aligned} $$

Here $D_1,D_2,D_3$ are the undressed sides of the box, $D_5,\ldots,D_9$ form the crossed-photon self-energythe electron emits photon 1, emits photon 2, absorbs photon 1, absorbs photon 2: two interleaved photon arcs on one fermion line, the same five-propagator topology as the two-loop kite, and $D_4$ is the electron line on either side of the insertion, whose two segments are the stubs. Both stubs carry the momentum $\ell$, so the QED graph in the figure is $I_{\rm KP}=I[1,1,1,2,1,1,1,1,1]$, with $D_4$ squared, while the single-stub member $I_{K_1}=I[1,1,1,1,1,1,1,1,1]$ has one power of it. Both $I_{\rm KP}$ and $I_{K_1}$ are ultraviolet finite. The quantities computed are their values at $d=4$ as functions of $s$ and $t$ at Euclidean points, $s,t\lt0$ with $u$ below its two-particle threshold $4m^2$, where the integrals are real.

At a glance

A self-energy insertion on one line of a loop is a two-particle-reduciblethe graph falls into two pieces when the two stub lines on either side of the insertion are cut piece of the amplitude, and such a graph is a dispersion integral over the squared mass $w$ of the dressed line: the one-loop box with that line given mass squared $w$, weighted by the spectral densitythe imaginary part of the self-energy across its cut, as a function of the invariant mass flowing through it of the insertion, as in the two-loop self-energy methods of Bauberger, Berends, Böhm and Buza (1995) and the dispersion relations of Remiddi and Tancredi (2016) for the sunrise and the kite. Sabry (1962) computed the spectral functions of the two-loop electron propagator, and the kite's master integralsa finite basis of integrals to which every integral of the family reduces by integration-by-parts identities are known as iterated integrals of modular forms for $\Gamma_1(6)$functions on the upper half-plane with a prescribed transformation law under $\Gamma_1(6)$, the subgroup of $SL(2,\mathbb{Z})$ that is the symmetry group of the equal-mass sunrise curve from Hönemann, Tempest and Weinzierl (2018); the density used here is derived instead from the differential equation of the kite. Bargiela, Chakraborty, Gambuti and Ozcelik (2026) computed the three-loop amplitudes with massless fermions and Ajjath, Chaubey and Shao (2023) the two-loop amplitudes with the fermion mass kept; at three loops with the electron mass kept, $I_{\rm KP}$ and $I_{K_1}$ do not appear to have been computed before.

The elliptic curve enters only through the insertion. On the maximal cutthe integral with every propagator put on shell, which isolates the intrinsic geometry of the graph the integral factorizes into the one-loop massive box, which is polylogarithmic, times the cut of the self-energy, and above $w=9m^2$ the three-particle cut of the kite is a phase-space integral on the elliptic curve of the equal-mass sunrise (Adams and Weinzierl, 2017), the $\Gamma_1(6)$ family with singular fibers at $w\in\{0,m^2,9m^2,\infty\}$. Below that threshold the density is a weight-two combination of logarithms and dilogarithms.

A second ten-line QED graph of three-loop light-by-light scattering, the non-planar fermion octagon whose two internal photons join opposite sides of the loop and cross, has not been computed; the crossed light-by-light box is its pinch with the four corner lines contracted. The octagon's maximal cut nevertheless fixes its geometry: a genus-one curve carrying thirty master integrals on the cut, whose modulus (the $j$-invariant) varies with $s$ and $t$. The curve of the graph computed here is inherited from the sunrise inside the self-energy, but the octagon's curve belongs to its own topology, so the massive three-loop $\gamma\gamma\to\gamma\gamma$ amplitude is not polylogarithmic even apart from its self-energy insertions.

Closed form

With the stubs kept, partial fractions in $\ell^2$ between the dispersion denominator $\ell^2-w$ and the stub propagator $\ell^2-m^2$ replace the one-loop box by a subtracted kernel and leave the density unchanged:

$$I_{\rm KP}=\frac1\pi\int_{m^2}^\infty\!\rho(w)\,K_P(w)\,\mathrm{d}w,\qquad K_P(w)=\frac{K(w)-K(m^2)}{(w-m^2)^2}-\frac{K'(m^2)}{w-m^2},$$

where the kernel $K(w)$ is the scalar one-loop box of Passarino and Veltman (1979) with three lines of mass $m$ and the dressed line of mass squared $w$, a closed-form combination of dilogarithms; $K(m^2)$ and $K'(m^2)$ are its value and its first derivative in $w$ at $w=m^2$, both taken from the closed form. The density $\rho(w)=-\,\mathrm{Im}\,J_{\rm top}(w+i0)$ is the spectral density of the kite, the discontinuity of its top-sector masterthe master integral with every propagator of the self-energy present once $J_{\rm top}$ across the cut. The squared stub would place a double pole at the lower endpoint $w=m^2$ of the dispersion integral; the two subtractions remove it, and $K_P$ is regular there. The single-stub member $I_{K_1}$ is the same integral with $K_1(w)=[K(w)-K(m^2)]/(w-m^2)$ in place of $K_P$. With the bare kernel $K(w)$ and the same density the integral is the box with a kite insertion, the no-stub member in the Checks table. The identity is general: any graph with a self-energy on one line of a loop reduces in this way to divided differences of the bare-line kernel integrated against the unchanged density of the insertion. At the reference point $(s,t,m^2)=(-1,-\tfrac13,1)$ the two kernel constants are $K(m^2)=0.1780502267\ldots$ and $K'(m^2)=-0.08182371131\ldots$.

The density is fixed by exact data alone. It is the discontinuity of the differential system of the nine kite master integrals in $w$, and regularity at the two-particle threshold $w=m^2$, where it behaves as $-2\pi\,(w-m^2)\ln(w-m^2)$, is the only boundary condition. Below the three-particle threshold the density is the weight-two combination of logarithms and dilogarithms given on the kite-insertion page, and above it the three-particle cut of the equal-mass sunrise inside the kite adds a single integral over that cut. The evaluation script uses both forms: inside the dispersion integral it generates the density as the power-series solution in $w$ of the kite system, started at $w=5m^2$ from values obtained by solving that system outward from $w=0$, where the master integrals have closed forms, and at fixed $w$, as in the density rows of the Checks table, it evaluates the closed form and compares the two. All dependence on $s$ and $t$ sits in the kernel $K$; an iterated-integral form in $s$ and $t$ themselves was not constructed. The complete expression, with the conventions, the density below and above the three-particle threshold and the kernel constants, is in the expression file inside the bundle under Evaluator.

Checks

Checks
pointclosed formindependent valuedigits
$I_{\rm KP}$ at $(s,t,m^2)=(-1,-\tfrac13,1)$$\;\;0.1120624814\ldots$$\;\;0.1120624814\ldots$38
$I_{K_1}$ at $(s,t,m^2)=(-1,-\tfrac13,1)$$-0.2162864879\ldots$$-0.2162864879\ldots$38
no-stub member at $(s,t,m^2)=(-1,-\tfrac13,1)$$\;\;0.5249577767\ldots$$\;\;0.5249577767\ldots$38
spectral density at $w=5\,m^2$$\;\;1.663479584\ldots$$\;\;1.663479584\ldots$75
spectral density at $w=12\,m^2$$\;\;0.3911816090\ldots$$\;\;0.3911816090\ldots$69
spectral density at $w=100\,m^2$$\;\;0.007791679551\ldots$$\;\;0.007791679551\ldots$65

The independent values are auxiliary-mass-flow evaluations (AMFlow) of the integrals of the nine-propagator family for the first three rows and of the kite master integrals at fixed $w$ for the density rows, none of which enters the closed form or the boundary vector at $w=5m^2$. The closed form agrees with them to the number of leading digits in the last column.

Evaluator

Evaluator

Python 3 with mpmath; the script reads its data files and helper modules from the unzipped bundle folder, so run it there. A full recomputation of the stored values takes about fifteen minutes on a laptop.

Tools
toolrole
Kiraintegration-by-parts reduction of the kite self-energy to nine master integrals and their differential system in $w$
Wayfinderpower-series solution of that system along $w\gt m^2$, which gives the spectral density
Nestorthe dispersion integral over $w$, with the threshold behavior of the density subtracted and restored in closed form
AMFlowthe independent numerical evaluations in the Checks table

Same family

The paper

References

E. Remiddi and L. Tancredi, Differential equations and dispersion relations for Feynman amplitudes. The two-loop massive sunrise and the kite integral, Nucl. Phys. B 907 (2016) 400–444 [arXiv:1602.01481]dispersion relations and the differential equations of the sunrise and the kite, from which the spectral density is derived
S. Bauberger, F. A. Berends, M. Böhm and M. Buza, Analytical and numerical methods for massive two-loop self-energy diagrams, Nucl. Phys. B 434 (1995) 383–407 [arXiv:hep-ph/9409388]the dispersive treatment of a line dressed by a massive self-energy
A. Sabry, Fourth order spectral functions for the electron propagator, Nucl. Phys. 33 (1962) 401–430the original spectral functions of the two-loop electron propagator
I. Hönemann, K. Tempest and S. Weinzierl, Electron self-energy in QED at two loops revisited, Phys. Rev. D 98 (2018) 113008; erratum Phys. Rev. D 110 (2024) 059901 [arXiv:1811.09308]the kite self-energy in iterated integrals of modular forms for $\Gamma_1(6)$, the known form of the insertion
L. Adams and S. Weinzierl, Feynman integrals and iterated integrals of modular forms (2017) [arXiv:1704.08895]the elliptic curve of the equal-mass sunrise, on which the three-particle cut of the kite is a phase-space integral
G. Passarino and M. J. G. Veltman, One-loop corrections for $e^+e^-$ annihilation into $\mu^+\mu^-$ in the Weinberg model, Nucl. Phys. B 160 (1979) 151–207the scalar one-loop box used as the dispersion kernel
P. Bargiela, A. Chakraborty, G. Gambuti and M. A. Ozcelik, Light-by-light scattering at three loops in massless QCD and QED: amplitudes and cross sections (2026) [arXiv:2603.22423]the three-loop light-by-light amplitudes with massless fermions
Ajjath A H, E. Chaubey and H.-S. Shao, Two-loop massive QCD and QED helicity amplitudes for light-by-light scattering (2023) [arXiv:2312.16966]the two-loop light-by-light amplitudes with the fermion mass kept

← back to Feynman diagrams