Single-Mass Heavy-Flavor Contributions

• Single-mass heavy-flavor contributions are DIS corrections that retain one heavy-quark mass (e.g., charm or bottom) while excluding simultaneous two-mass effects, and they factorize into massless Wilson coefficients and massive operator matrix elements.
• Their three-loop computations provide key insights into anomalous dimensions and the structure of heavy-flavor operator matrix elements, employing advanced symbolic and summation techniques.
• These contributions are essential for precision DIS phenomenology at high Q², influencing both unpolarized and polarized structure functions as well as variable-flavor-number matching in QCD analyses.

Single-mass heavy-flavor contributions are the deep-inelastic scattering (DIS) corrections in which one heavy-quark mass scale is retained at a time—typically charm or bottom—while simultaneous two-mass effects are excluded. In the asymptotic regime $Q^2 \gg m^2$, these contributions admit an operator-product-expansion description in which heavy-flavor Wilson coefficients factorize into massless Wilson coefficients and massive operator matrix elements (OMEs). This framework governs the three-loop treatment of inclusive neutral-current and charged-current structure functions such as $F_2$, $F_L$, $F_1$, $g_1$, and $xF_3$, and it also provides the transition functions used in variable-flavor-number schemes (VFNS) (Ablinger et al., 2010).

1. Definition, observables, and kinematic regime

“Single-mass” denotes the situation in which the heavy-flavor sector contains one heavy-quark mass scale only, for example $m_c$ or $m_b$, and no simultaneous two-mass effects such as charm-bottom mixing. In the leading-twist treatment of DIS, the coefficient functions are decomposed into a massless part and a heavy-flavor part,

$$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$

so that the heavy contribution appears as a distinct perturbative sector in $F_2$, $F_2$0, and, in the polarized case, $F_2$1 (Ablinger et al., 2021).

For neutral-current DIS, the measured structure function is decomposed into light- and heavy-flavor pieces,

$F_2$2

with the heavy terms becoming particularly important at small Bjorken $F_2$3 (Ablinger et al., 19 Sep 2025). For charged-current DIS, the single-mass heavy sector has an additional feature: besides heavy-quark pair production, there is also single heavy-flavor excitation, specifically $F_2$4, which contributes already at parton level in the non-singlet charged-current structure functions (Behring et al., 2016).

The entire asymptotic construction is restricted to sufficiently large virtuality. Several papers state that the approximation is numerically reliable for $F_2$5 once $F_2$6, while it is not intended for the threshold region $F_2$7 (Ablinger et al., 2012). A recurrent caution in the literature is that asymptotic heavy-flavor expressions capture twist-2 contributions and omit power-suppressed terms $F_2$8, so they are high-scale descriptions rather than exact finite-$F_2$9 results (Ablinger et al., 2010).

2. Asymptotic factorization and the role of massive OMEs

The central structural statement is the asymptotic factorization of heavy-flavor Wilson coefficients into massless Wilson coefficients and massive OMEs. In Mellin space, one form used for the neutral-current case is

$F_L$0

where $F_L$1 contains the heavy-mass dependence and $F_L$2 is massless; the Mellin convolution in $F_L$3-space becomes ordinary multiplication in $F_L$4-space (Ablinger et al., 2010). Equivalent generic formulations appear throughout the three-loop literature,

$F_L$5

with $F_L$6 the Bjorken-$F_L$7 convolution (Ablinger et al., 2024).

The OMEs are local twist-2 matrix elements between on-shell partonic states in the presence of one heavy mass,

$F_L$8

and they encode the full asymptotic mass dependence after renormalization and mass factorization (Ablinger et al., 2010). For the unpolarized three-loop single-mass case, the relevant OMEs are

$F_L$9

with corresponding polarized OMEs in the helicity-dependent sector (Ablinger et al., 2024).

The renormalization logic is standard but highly constraining. The calculations use $F_1$0 for coupling and operator renormalization and on-shell renormalization for the heavy-quark mass; after ultraviolet and mass renormalization, the remaining collinear singularities are removed by mass factorization (Ablinger et al., 2010). A crucial consequence is that the pole and logarithmic terms of the three-loop OMEs are determined by anomalous dimensions up to three loops, $F_1$1-function coefficients up to two loops, and lower-order OMEs, while the constant term is the genuinely new three-loop contribution (Ablinger et al., 2010). This is why OME calculations simultaneously provide independent access to pieces of the three-loop anomalous dimensions.

3. Three-loop development and completion of the single-mass sector

The three-loop program developed in stages. Early NNLO asymptotic work established that massive three-loop OMEs account for all heavy-flavor terms except the power corrections $F_1$2 and reported fixed Mellin moments for all three-loop massive OMEs. That stage also yielded new all-$F_1$3 expressions for the $F_1$4 terms, including the constant term $F_1$5, confirmed the corresponding fermionic pieces of the three-loop anomalous dimensions, and presented the first genuine three-loop ladder-type results for general Mellin moment $F_1$6 (Ablinger et al., 2010).

A later status report stated that six out of eight relevant three-loop OMEs had been computed,

$F_1$7

while

$F_1$8

were still missing (Ablinger et al., 2014). In parallel, the pure-singlet channel was advanced by a complete calculation of the asymptotic three-loop heavy-flavor corrections to $F_1$9 in the $g_1$0 sector together with the transition matrix element $g_1$1 and an independent recalculation of the full three-loop pure-singlet anomalous dimension (Ablinger et al., 2014).

The gluonic transition sector was completed by the three-loop calculation of the unpolarized and polarized OMEs

$g_1$2

including their Mellin-space and $g_1$3-space forms and their analytic continuation from even or odd moments into the complex plane (Ablinger et al., 2022). A subsequent 2024 overview then stated that all single-mass OMEs and all asymptotic inclusive heavy-flavor Wilson coefficients had been completed to three-loop order for unpolarized and polarized DIS, together with the single-mass VFNS matching relations (Ablinger et al., 2024).

The next step was phenomenological rather than structural: the first numerical three-loop single-mass heavy-flavor corrections to the inclusive structure functions $g_1$4 and $g_1$5 were reported in 2025, providing a full NNLO numerical realization of the asymptotic single-mass program for both unpolarized and polarized inclusive DIS (Ablinger et al., 19 Sep 2025).

4. Analytic structures and computational technology

The analytic function space of single-mass heavy-flavor calculations is considerably richer than the one encountered in lower-order massless DIS. For the $g_1$6 three-loop terms, the Mellin-space results are expressed through harmonic sums up to weight 4, and the basis can be reduced to six basic sums,

$g_1$7

together with rational functions of $g_1$8 and zeta values (Ablinger et al., 2010). Ladder-type diagrams already generate more complicated nested sums, including generalized harmonic sums in intermediate expressions (Ablinger et al., 2010).

The pure-singlet sector introduced a qualitatively new feature in DIS: generalized harmonic sums in the final Mellin-space result for $g_1$9, and corresponding $xF_3$0-space structures involving not only ordinary harmonic polylogarithms in $xF_3$1 but also functions with argument $xF_3$2 (Ablinger et al., 2014). More broadly, later summaries emphasized that pure-singlet terms require generalized harmonic sums, whereas non-singlet and some $xF_3$3 sectors can be expressed in terms of ordinary harmonic sums or harmonic polylogarithms only (Ablinger et al., 2021).

Additional diagram classes generate cyclotomic harmonic sums and cyclotomic harmonic polylogarithms. These arise from alphabets built from cyclotomic polynomials and extend the standard harmonic-polylogarithm alphabet $xF_3$4 (Ablinger et al., 2012). The gluonic three-loop sector brings in finite binomial and inverse binomial sums in Mellin space and iterated integrals over square-root-valued alphabets in $xF_3$5-space, including plus distributions and denominator structures such as $xF_3$6 with $xF_3$7 (Ablinger et al., 2022). The 2025 numerical analysis further noted that $xF_3$8 contains binomial sums, square-root iterated integrals, and $xF_3$9-type solutions in $m_c$0-space (Ablinger et al., 19 Sep 2025).

The computational pipeline combines large-scale symbolic reduction with specialized summation technology. Reported tools include QGRAF for diagram generation, FORM for algebraic manipulation, integration-by-parts reduction with Reduze2 implementing Laporta’s algorithm, and master-integral evaluation via generalized hypergeometric functions, Mellin-Barnes representations, hyperlogarithms, and differential equations. The nested sums are handled with Sigma, HarmonicSums, EvaluateMultiSums, SumProduction, and OreSys; later surveys also highlight the method of arbitrarily large Mellin moments and the method of guessing for recurrence reconstruction (Ablinger et al., 2014). A common misconception addressed explicitly in this literature is that leading logarithms alone are sufficient: both the 2012 and 2024 analyses stress that leading small-$m_c$1 terms do not approximate the full result well by themselves, because subleading contributions and constant terms are numerically important (Ablinger et al., 2012).

5. Phenomenology, polarized observables, and flavor schemes

The phenomenological importance of single-mass heavy-flavor terms is greatest at small $m_c$2, where the heavy contribution can be a large fraction of the inclusive structure function and its scaling violations differ from the massless case (Ablinger et al., 19 Sep 2025). For $m_c$3, the asymptotic approximation is stated to be already accurate at the $m_c$4 level when

$m_c$5

at two loops, and this same high-$m_c$6 regime suppresses higher-twist effects (Ablinger et al., 19 Sep 2025). By contrast, the asymptotic description of $m_c$7 is much less accurate; a survey reports that power corrections remain important and that the pure asymptotic terms describe $m_c$8 only for much larger scales, around $m_c$9 (Ablinger et al., 2021).

The first complete three-loop numerical analysis for inclusive $m_b$0 used $m_b$1, PDFs from Alekhin et al., $m_b$2, and on-shell masses $m_b$3, $m_b$4. It found that, in the small-$m_b$5 region, the heavy-flavor fraction

$m_b$6

rises from about $m_b$7 at $m_b$8 to about $m_b$9 at $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$0; around $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$1 the heavy-flavor correction crosses through zero; and at very large $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$2, NNLO heavy-flavor corrections can be negative and as large as about $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$3, while the total $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$4 remains positive (Ablinger et al., 19 Sep 2025). In the same study, the gluonic coefficient $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$5 dominates the charm contribution, $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$6 is sizable and negative, $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$7 dominates at large $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$8, and $$\mathbb{C}_{2,L}^{(i)}\!\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right) = C_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2}\right) + H_{2,L}^{(i)}\left(x,\frac{Q^2}{\mu^2},\frac{m_q^2}{\mu^2}\right),$$9 and $F_2$0 remain relevant at the percent level (Ablinger et al., 19 Sep 2025). Earlier pure-singlet analyses likewise found that the heavy-flavor pure-singlet contribution to $F_2$1 is negative in the studied kinematics and significantly larger for charm than for bottom (Ablinger et al., 2014).

For polarized DIS, the same asymptotic single-mass machinery applies, but the three-loop calculation is carried out in the Larin scheme. The 2025 inclusive analysis reports that the gluonic coefficient $F_2$2 is dominant also in $F_2$3, and its first moment vanishes through three loops,

$F_2$4

This vanishing first moment drives an oscillatory structure in $F_2$5-space (Ablinger et al., 19 Sep 2025). In the region where $F_2$6 itself is small, the heavy-flavor ratio can be roughly $F_2$7, $F_2$8, and $F_2$9 for $F_2$00, respectively; around $F_2$01 the ratio changes sign, around $F_2$02 it becomes negative again, and yet the full $F_2$03 remains positive (Ablinger et al., 19 Sep 2025). A 2024 study further notes that, at NNLO, Larin- and $F_2$04-scheme polarized evolution differ noticeably, with quark distributions differing by about $F_2$05–$F_2$06 at small $F_2$07 and gluon distributions by about $F_2$08, so consistent polarized phenomenology requires PDFs evolved in the Larin scheme (Ablinger et al., 2024).

The same OMEs also determine matching across heavy-flavor thresholds in the VFNS, including the generation of the heavy-quark PDFs $F_2$09 (Ablinger et al., 2024). In this sense, single-mass asymptotic calculations are not only fixed-flavor predictions but also transition functions for an $F_2$10-flavor description (Ablinger et al., 2010). A separate NNLO implementation in the S-ACOT-$F_2$11 general-mass scheme illustrates how single-mass heavy-flavor structure functions such as $F_2$12 and $F_2$13 are assembled from massive and zero-mass ingredients using the rescaling variable

$F_2$14

which enforces threshold kinematics and yields smooth interpolation between fixed-flavor and zero-mass limits (Guzzi et al., 2011).

6. Charged-current sector, sum rules, and scope

Single-mass heavy-flavor contributions are not restricted to neutral-current $F_2$15 and $F_2$16. In charged-current non-singlet DIS, the asymptotic three-loop heavy-flavor corrections to

$F_2$17

and

$F_2$18

have been calculated for general Mellin moment $F_2$19 and in $F_2$20-space (Behring et al., 2016). These observables contain two distinct heavy-flavor mechanisms: heavy-quark pair production, encoded in $F_2$21-type Wilson coefficients, and single heavy-flavor excitation $F_2$22, encoded in $F_2$23-type coefficients (Behring et al., 2015).

The charged-current asymptotic results show that charm corrections to $F_2$24 and $F_2$25 are typically in the range $F_2$26, with the asymptotic approximation becoming reliable over a broader $F_2$27-range as $F_2$28 increases; at lower $F_2$29, power corrections remain visible, especially for $F_2$30 and therefore $F_2$31 (Behring et al., 2016). For $F_2$32, the charm effect is up to about $F_2$33 at small $F_2$34 and about $F_2$35 at large $F_2$36, while the $F_2$37 correction relative to the massless three-flavor result is typically at the $F_2$38–$F_2$39 level (Behring et al., 2015).

The first Mellin moment has a special status. For the Adler sum rule, the charged-current heavy-flavor analysis states that there are no QCD corrections and no quark-mass corrections; in the asymptotic heavy-flavor limit this follows because the relevant non-singlet OMEs vanish at $F_2$40 by fermion-number conservation, and the first moment of the corresponding massless Wilson coefficient also vanishes (Behring et al., 2016). For the Gross–Llewellyn Smith sum rule, the heavy-flavor effect in the asymptotic region reduces essentially to the replacement

$F_2$41

in the massless coefficient, with CKM weights included (Behring et al., 2015). The same pattern appears for the unpolarized Bjorken sum rule in the charged-current $F_2$42 sector (Behring et al., 2016).

The scope of the single-mass literature is also sharply delimited. The main three-loop results are asymptotic, twist-2, and usually formulated in the fixed-flavor-number scheme with $F_2$43 light flavors for phenomenological applications (Ablinger et al., 19 Sep 2025). They are not intended for threshold kinematics, where $F_2$44, and they exclude simultaneous two-mass effects; a 2024 overview states that two-mass corrections were nearly finished, while the single-mass sector was complete (Ablinger et al., 2024). A plausible implication is that “single-mass heavy-flavor contributions” should be understood not as a universal finite-$F_2$45 solution, but as the fully developed high-scale component of heavy-flavor QCD factorization, with direct relevance to precision determinations of $F_2$46, heavy-quark masses, twist-2 PDFs, and threshold matching in modern DIS analyses (Ablinger et al., 19 Sep 2025).