Loop Contribution in QCD & Beyond

• Loop contribution is a perturbative element arising from loop-level structures, with definitions varying across QCD, superstring theory, and effective gravity.
• In inclusive Higgs production at N³LO, the RVV term combines two-loop amplitudes with real emission, reducing theory uncertainties in collider predictions.
• Advanced techniques like reverse unitarity and integration-by-parts reduce complex integrals to master forms expressed via multiple and harmonic polylogarithms.

Searching arXiv for the target paper and closely related work on N3LO Higgs production and loop contributions.

arXiv search query: (Dulat et al., 2014)

“Loop contribution” denotes a contribution to a perturbative observable that is associated with loop-level structures, but its concrete meaning depends on the framework. In perturbative QCD, a particularly explicit example is the real‑virtual‑virtual contribution to inclusive Higgs production at next‑to‑next‑to‑next‑to‑leading order, defined by the interference of the two‑loop Higgs‑plus‑one‑parton amplitude with the tree amplitude and integrated over the one‑parton phase space (Dulat et al., 2014). Taken together with related usages in superstring theory, post‑Newtonian gravity, and precision phenomenology, the term is best understood as context-dependent: in superstring theory loop order corresponds geometrically to the genus $g$ of the worldsheet, whereas in post‑Newtonian effective field theory loop diagrams can encode classical multi‑graviton exchanges rather than quantum corrections (Bettadapura et al., 2020, Blümlein et al., 2019).

1. Definition and domain dependence

In the heavy–top effective theory for Higgs production, the partonic cross section is expanded as

$$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$

with $a_s=\alpha_s/\pi$ and $z=m_H^2/\hat s$. In this setting, a loop contribution is a definite piece of $\hat{\sigma}^{(n)}$ associated with virtual amplitudes of specified loop order. At $N^3LO$, the inclusive Higgs cross section receives triple‑real, double‑real‑virtual, real‑virtual‑virtual, and triple‑virtual contributions, and the loop contribution emphasized in the cited work is the real‑virtual‑virtual, or RVV, term (Dulat et al., 2014).

This usage is not universal across theoretical physics. In perturbative closed superstring theory, the genus‑$g$ contribution to an amplitude is the $g$-loop contribution, and the relevant geometry is the supermoduli space $\mathfrak M_g$ of genus‑$g$ super Riemann surfaces (Bettadapura et al., 2020). In the post‑Newtonian effective field theory of binary dynamics, loop diagrams arise from integrating out gravitational fields and represent higher‑order classical corrections to the conservative interaction potential; at 5PN in the static sector this becomes a genuine five‑loop calculation in a purely classical field theory (Blümlein et al., 2019). This suggests that “loop contribution” names a structural role in a perturbative expansion rather than a single universal mathematical object.

2. Real‑virtual‑virtual loop contributions in inclusive Higgs production

The concrete QCD example is inclusive Higgs boson production in gluon fusion at hadron colliders,

$$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$0

computed in QCD in the heavy–top effective theory with effective interaction

$$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$1

The RVV contribution is one building block of $$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$2 and comes from the interference of the two‑loop amplitude for Higgs plus one parton with the tree‑level amplitude, integrated over the one‑parton phase space (Dulat et al., 2014).

The relevant partonic channels are

$$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$3

With

$$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$4

the loop contribution under discussion is

$$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$5

where $$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$6 is the two‑body phase space of a massive Higgs plus one massless parton, and $$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$7 are color–spin averaging factors (Dulat et al., 2014).

A common misconception is to equate a loop contribution with a purely virtual correction. The RVV term is not purely virtual in that sense: it contains a two‑loop amplitude, but it also contains one real parton in the final state and therefore carries both loop and phase‑space singularities. The paper explicitly identifies it as “real‑virtual‑virtual” for that reason (Dulat et al., 2014).

The phenomenological motivation is precision Higgs physics. NNLO predictions still carry theory uncertainties of order $$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$8, whereas a complete $$\hat{\sigma}(z) = \hat{\sigma}_0(z)\Big(1 + a_s\, \hat{\sigma}^{(1)}(z) + a_s^2\,\hat{\sigma}^{(2)}(z) + a_s^3\,\hat{\sigma}^{(3)}(z) + \dots\Big),$$9 calculation is expected to reduce this to about $a_s=\alpha_s/\pi$0, comparable to or below current experimental systematics. The RVV term is therefore not ancillary; it is one of the ingredients required to assemble the final finite $a_s=\alpha_s/\pi$1 prediction (Dulat et al., 2014).

3. Singular structure, master integrals, and analytic form

The RVV cross section is regulated in conventional dimensional regularization with

$a_s=\alpha_s/\pi$2

Its structure near threshold is written as

$a_s=\alpha_s/\pi$3

and then decomposed into singular and regular pieces,

$a_s=\alpha_s/\pi$4

The coefficients admit Laurent expansions in $a_s=\alpha_s/\pi$5,

$a_s=\alpha_s/\pi$6

with the negative powers of $a_s=\alpha_s/\pi$7 encoding soft, collinear, and UV singularities (Dulat et al., 2014).

The computation uses reverse unitarity, replacing phase‑space delta functions by cut propagators,

$a_s=\alpha_s/\pi$8

so that loop and phase‑space integrations can be treated on the same footing. Integration‑by‑parts identities and a Laporta reduction reduce all required integrals to a basis of 72 master integrals $a_s=\alpha_s/\pi$9 (Dulat et al., 2014).

With

$z=m_H^2/\hat s$0

the masters satisfy a coupled first‑order differential system

$z=m_H^2/\hat s$1

where the only singularities in $z=m_H^2/\hat s$2 are at $z=m_H^2/\hat s$3 and $z=m_H^2/\hat s$4. For 56 of the 72 masters, the system is transformed to a canonical basis $z=m_H^2/\hat s$5 in Henn’s sense,

$z=m_H^2/\hat s$6

while the remaining 16 masters are solved in a non‑canonical but still tractable form by expanding in $z=m_H^2/\hat s$7 and iteratively decoupling the system (Dulat et al., 2014).

The resulting analytic functions are multiple polylogarithms and, in the final RVV cross sections, harmonic polylogarithms with indices in $z=m_H^2/\hat s$8. Symbolically,

$z=m_H^2/\hat s$9

with rational functions $\hat{\sigma}^{(n)}$0. The paper’s central analytic statement is therefore that the coefficients of the Laurent expansion are linear combinations of harmonic polylogarithms of the ratio of the Higgs boson mass and the partonic center of mass energy (Dulat et al., 2014).

4. Boundary conditions and expansion by regions

The differential equations determine the master integrals only up to boundary constants. The paper develops a specific method for decomposing these boundary conditions into physical contributions, and this is one of its distinctive technical results (Dulat et al., 2014).

Near threshold, $\hat{\sigma}^{(n)}$1, the system is approximated by

$\hat{\sigma}^{(n)}$2

A Jordan decomposition,

$\hat{\sigma}^{(n)}$3

leads to a basis in which each integration constant is associated with a definite integrating factor $\hat{\sigma}^{(n)}$4 multiplied, when required, by powers of $\hat{\sigma}^{(n)}$5. This isolates the asymptotic building blocks that can actually occur in the physical integral family (Dulat et al., 2014).

Those integrating factors are then matched to regions found by expansion by regions in the sense of Beneke–Smirnov, aided by the code asy. For the RVV system, only regions with integrating factors of the form

$\hat{\sigma}^{(n)}$6

occur. Consequently, only eigenvalues $\hat{\sigma}^{(n)}$7 can contribute, and all other boundary constants vanish. This reduces the number of boundary conditions that must be computed explicitly from 72 to 19 (Dulat et al., 2014).

The remaining nonzero boundary constants are then evaluated by direct momentum‑space expansion by regions or by parametric integrals with region scaling from asy. The paper illustrates this on a simple one‑loop example and on an actual two‑loop RVV boundary integral, the double cut of the tennis‑court topology. A plausible implication is that the method is not merely a threshold‑expansion device; it is a systematic way of assigning integration constants to momentum regions in multi‑scale loop systems (Dulat et al., 2014).

5. Infrared structure and assembly into the physical cross section

The RVV contribution is not finite by itself. Its Laurent expansion contains poles in $\hat{\sigma}^{(n)}$8 originating from soft divergences, collinear divergences, and virtual infrared divergences from loop diagrams. The threshold factors

$\hat{\sigma}^{(n)}$9

make the threshold logarithms explicit (Dulat et al., 2014).

These singularities are not pathological; they are one element of the standard QCD cancellation pattern. The paper states that the leading soft limit of the RVV cross section agrees with the two‑loop soft current results of Duhr–Gehrmann and Li–Zhu, and that the singularity structure is compatible with the known $N^3LO$0 soft expansion of the inclusive cross section (Dulat et al., 2014). When the RVV term is combined with triple‑real, double‑real‑virtual, and triple‑virtual pieces, together with renormalization and mass factorization counterterms, the infrared and collinear poles cancel or are absorbed into PDFs, leaving a finite physical $N^3LO$1 cross section (Dulat et al., 2014).

The explicit analytic expressions are lengthy and are supplied as Mathematica files. The paper defines sggRVV, sgqRVV, and sqqbarRVV for the bare RVV cross sections, normalized by

$N^3LO$2

For the singular pieces, the factors $N^3LO$3 are kept unexpanded and $N^3LO$4 is expanded up to $N^3LO$5; for the regular pieces, the product $N^3LO$6 is expanded up to $N^3LO$7 (Dulat et al., 2014).

6. Broader meanings of loop contribution in adjacent fields

The phrase “loop contribution” acquires different technical meanings in other theories represented in the literature block. In type II superstring theory, the genus‑3 vacuum amplitude can receive boundary contributions from the compactified supermoduli space, and the paper on three‑loop superstring amplitudes shows that these boundary contributions vanish in closed oriented type II superstring theory with unbroken spacetime supersymmetry (Bettadapura et al., 2020). Here the loop contribution is a genus contribution, not a Feynman momentum loop.

In post‑Newtonian effective field theory, the five‑loop static contribution to the gravitational interaction potential of two point masses is a classical contribution represented by multi‑graviton exchange diagrams in momentum space. At 5PN in the static sector, the potential is determined by five‑loop diagrams, and the main result is a rational $N^3LO$8 correction to the static Lagrangian (Blümlein et al., 2019). This directly contradicts the common assumption that loop diagrams are inherently quantum.

In strongly background‑dependent QED, loop contributions can also be one‑particle reducible. The scalar and spinor propagator papers show that, in a constant electromagnetic field, the one‑loop propagator receives a finite 1PR contribution in addition to the familiar 1PI term, and that this reducible piece is expressible in terms of derivatives of the propagator and the one‑loop Euler–Heisenberg Lagrangian (Edwards et al., 2017, Ahmadiniaz et al., 2017).

In precision phenomenology, the term often denotes loop‑induced corrections to observables rather than exclusive diagram classes. Examples in the supplied corpus include the one‑loop excited lepton triplet contribution to $N^3LO$9, which is small for degenerate triplets but can be large for non‑degenerate masses (Rehman et al., 2020); the two‑loop rainbowlike and Barr–Zee contributions to fermion EDMs in $g$0-parity violating supersymmetry (Yamanaka, 2012, Yamanaka, 2012); and the one‑loop vacuum polarization contribution to the magnetic field power spectrum in de Sitter space, which after dynamical renormalization group resummation changes the power from $g$1 to $g$2 with $g$3 (Motohashi et al., 2014).

7. Conceptual clarifications

Several clarifications follow from these examples. First, a loop contribution is not necessarily synonymous with a fully virtual correction: the RVV Higgs contribution contains a two‑loop amplitude but also one real parton in the final state (Dulat et al., 2014). Second, loop order is not defined uniformly across all frameworks: in superstring theory it is the genus of the worldsheet (Bettadapura et al., 2020), whereas in post‑Newtonian gravity it counts classical multi‑field exchanges in an effective field theory (Blümlein et al., 2019).

Third, reducibility is not the same as irrelevance. The one‑particle reducible contributions to the Euler–Heisenberg Lagrangian and to propagators in constant fields were previously neglected, yet the cited works show that they are finite and nonzero (Edwards et al., 2017, Ahmadiniaz et al., 2017). Fourth, loop contributions do not have a fixed phenomenological sign or size. In RPV supersymmetric EDMs, sfermion‑loop Barr–Zee diagrams act destructively relative to the known fermion‑loop contribution (Yamanaka, 2012), while in inflationary magnetogenesis the one‑loop vacuum polarization effect always enhances the infrared magnetic field strength because $g$4 is positive irrespective of the fermion mass (Motohashi et al., 2014).

Taken together, these results support a precise but nonuniform definition: a loop contribution is a perturbative sector identified by the topology or order of virtual integration variables, yet its physical interpretation depends on the observable, the background, and the perturbative framework. In the specific and technically mature example of inclusive Higgs production at $g$5, the RVV contribution exemplifies how such loop sectors are isolated, reduced to master integrals, solved analytically, and finally combined with the rest of the perturbative expansion to produce a finite collider prediction (Dulat et al., 2014).