Froissart-Gribov Projections in Scattering & DVCS

• Froissart-Gribov projections are integral representations that extract partial-wave components from scattering amplitudes using Legendre functions.
• They convert subthreshold information into high-energy bounds by relating low-energy absorptive parts to crossed-channel observables in both hadronic scattering and DVCS.
• These projections underpin rigorous total-cross-section bounds and serve as a systematic tool in QCD analyses, linking experimental amplitudes to GPD models.

Froissart-Gribov projections are integral representations that isolate partial-wave components of scattering amplitudes and, in crossed-channel form, relate those components to absorptive parts in another channel. In analytic $S$-matrix theory they are one of the standard mechanisms by which crossing, analyticity, and partial-wave unitarity are converted into statements about high-energy asymptotics; in contemporary QCD they also serve as observables built from deeply virtual Compton scattering amplitudes and generalized parton distributions (GPDs), where they resolve the response of a hadron to non-local probes carrying definite $t$-channel angular momentum $J$ (Azimov, 2012, Semenov-Tian-Shansky et al., 2023).

1. Definitions and canonical forms

The expression “Froissart-Gribov projection” covers several closely related projections. In the most elementary form, a partial wave is extracted from an amplitude by projection onto a Legendre polynomial,

$$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$

with $z=\cos\theta$. In crossed-channel applications, the projection is recast in terms of Legendre functions of the second kind $Q_l$, which makes analyticity in the angular variable explicit and is particularly useful for high-$l$ estimates. For $\pi\pi$ scattering, the $t$-channel partial wave for $l\ge 2$ and $t$0 is written as

$t$1

so that a below-threshold $t$2-channel quantity is expressed directly through the physical $t$3-channel absorptive part. In DVCS, the same logic appears in the $t$4-channel expansion

$t$5

with projection coefficients

$t$6

These formulas differ in normalization and kinematic interpretation, but they all implement the same operation: projection onto definite angular momentum in a channel related to the original amplitude by analyticity and crossing (Azimov, 2012, Martin et al., 2013, Semenov-Tian-Shansky et al., 2023, Azimov, 2011).

Setting	Projection	Function class
$t$7-channel partial wave	$t$8	$t$9
Crossed-channel partial wave	$J$0	$J$1
DVCS $J$2-channel coefficient	$J$3	$J$4

A central feature of the crossed-channel form is that it converts low-energy or subthreshold information in one channel into weighted averages of absorptive parts in another. That mechanism is what makes the projections structurally important for both rigorous high-energy bounds and GPD phenomenology.

2. Analyticity, Legendre functions, and partial-wave suppression

The high-energy utility of Froissart-Gribov projections comes from the asymptotics of Legendre functions together with analyticity away from the physical interval. In the standard partial-wave expansion,

$J$5

unitarity imposes $J$6. A second, analyticity-based bound is obtained by rewriting $J$7 as a contour integral involving $J$8 and then deforming the contour as far from the cut $J$9 as analyticity allows. For large $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$0, $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$1 falls exponentially,

$$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$2

and this yields bounds of the form

$$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$3

depending on the precise normalization used. The parameter $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$4 or $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$5 is fixed by the nearest singularity in the crossed channel, so the absence of massless exchanges is essential: if a singularity reached the edge of the physical region, the entire mechanism would be altered (Azimov, 2011, Azimov, 2012).

This also explains why physical and nonphysical angular configurations behave differently. For $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$6, Legendre polynomials remain bounded; outside the physical interval they can grow exponentially with $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$7. The physical amplitude is therefore much more strongly constrained than its analytic continuation to nonphysical angles. The modern reinterpretations of the Froissart theorem stress exactly this point: the familiar $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$8 behavior is not generated by unitarity alone, but by unitarity together with analyticity, the absence of massless hadrons, and an assumption on the allowed growth of the amplitude in nonphysical configurations (Azimov, 2012, Azimov, 2011, Azimov, 2012).

3. Role in the Froissart theorem and in improved total-cross-section bounds

In the standard argument, the relevant partial waves are those below a critical angular momentum $$a_l(s)=\frac{1}{2}\int_{-1}^{+1}dz\,P_l(z)\,f(s,z),$$9, determined by the point where the unitarity bound and the analyticity bound intersect. Summing roughly up to $z=\cos\theta$0 gives a forward amplitude bounded by $z=\cos\theta$1, and via the optical theorem this controls $z=\cos\theta$2. If the amplitude in nonphysical regions grows no faster than a power of $z=\cos\theta$3, one recovers the canonical Froissart behavior $z=\cos\theta$4; if that assumption is relaxed, faster growth of $z=\cos\theta$5 need not contradict unitarity (Azimov, 2011).

The 2013 analysis of the Froissart bound without unknown constants makes the projection technology explicit. It defines

$z=\cos\theta$6

and uses Froissart-Gribov projections to relate $z=\cos\theta$7 near threshold to the $z=\cos\theta$8-channel D-wave. For $z=\cos\theta$9 scattering an absolute upper bound on the D-wave below threshold is obtained, and the paper derives an energy-averaged total-cross-section bound of the form

$Q_l$0

with the logarithmic scale made explicit in terms of pion quantities. The significance of the projections here is not merely technical: they provide the bridge from a low-energy crossed-channel partial wave to a rigorous asymptotic statement about high-energy cross sections (Martin et al., 2013).

A related refinement emphasized in later reanalyses is that even under the usual assumptions the original Froissart estimate can be strengthened, because the scale $Q_l$1 should itself slowly grow with $Q_l$2. This reinforces the conclusion that the familiar log-squared form is contingent, rather than automatic (Azimov, 2011, Azimov, 2012).

4. Deeply virtual Compton scattering and generalized parton distributions

In DVCS, Froissart-Gribov projections have been reformulated as quantities directly tied to generalized Compton form factors and GPDs. At leading order the charge-even Compton form factor is

$Q_l$3

The $Q_l$4-channel coefficients $Q_l$5 extracted from the partial-wave expansion of $Q_l$6 are the DVCS Froissart-Gribov projections. For a spinless target, the explicit expressions in terms of the GPD on the cross-over line $Q_l$7 are

$Q_l$8

and, for even $Q_l$9,

$l$0

For spin-$l$1 targets, the analysis introduces electric and magnetic combinations,

$l$2

and derives a magnetic projection involving the associated Legendre function $l$3. The paper identifies these projections as the hadron’s response to string-like QCD probes with definite $l$4-channel angular momentum $l$5, and states that the relation between FG projections and GPDs for spin-$l$6 targets is established there for the first time (Semenov-Tian-Shansky et al., 2023).

Within the dual parametrization framework, the projections become moments of the GPD quintessence function $l$7: $l$8 while

$l$9

The same work derives sum rules connecting specific $\pi\pi$0 to Mellin moments of GPDs; for example,

$\pi\pi$1

Phenomenologically, the projections were studied in the GK, MMS, and KM15 GPD models and compared with model-independent Compton form factor extractions using artificial neural networks. The reported findings are that higher-$\pi\pi$2 projections are especially sensitive to the valence region and to the skewness profile parameter $\pi\pi$3, while the $\pi\pi$4 projection contains and is dominated by the $\pi\pi$5-term. The same study also reports the first numerical estimates of FG projections from DVCS amplitudes directly extracted from experimental data (Semenov-Tian-Shansky et al., 2023).

5. Analogues and related developments in other high-energy frameworks

The logic behind Froissart-Gribov projections extends beyond the traditional hadron-hadron setting. In CFT Mellin amplitudes, a conformal partial-wave expansion plays the role of the ordinary partial-wave decomposition, and the flat-space limit reproduces the standard Froissart-Martin bound in $\pi\pi$6, including the coefficient $\pi\pi$7, with $\pi\pi$8 the lightest crossed-channel exchange. The same analysis finds that for $\pi\pi$9 two subtractions suffice for a dispersion relation in Mellin space, whereas for $t$0 the number of required subtractions is greater than two and diverges as $t$1 (Haldar et al., 2019).

In small-$t$2 deep inelastic scattering, an explicit Froissart-Gribov projection is not written, but the Mellin-space analysis is used in a closely related way. A Froissart-bounded parametrization of $t$3 is propagated through DGLAP-based Mellin methods to obtain $t$4, and the resulting structure function is shown to preserve the same $t$5 behavior at small $t$6. The paper explicitly states that the expansion around the Mellin singularity at $t$7 is the analytic mechanism by which the Froissart bound is enforced (Kaptari et al., 2019).

In Regge-Gribov field theory with pomerons and odderons, the “Froissart limitation” appears as a restriction on physically admissible high-energy behavior. The two-dimensional transverse-space model studied by renormalization-group methods yields

$t$8

which is well within the Froissart bound, while supercritical phases are judged unphysical because they violate projectile-target symmetry (Braun et al., 26 Jan 2026).

A distinct, but related, realization occurs in the Gribov-Zwanziger approach to high-energy QCD evolution. There the infrared-modified gluon propagator produces an exponential large-impact-parameter tail,

$t$9

and the solution of the non-linear equation gives a Froissart disc whose radius grows as

$l\ge 2$0

with finite width in $l\ge 2$1. This is not presented as a direct application of FG projection machinery, but it is another context in which the Froissart asymptotic regime is recovered through analyticity-sensitive infrared structure (Gotsman et al., 2020).

6. Conceptual status, misconceptions, and significance

A recurring misconception in the literature is that the familiar $l\ge 2$2 behavior follows from unitarity alone, so that any faster rise would automatically signal a breakdown of unitarity. The reanalyses summarized here reject that interpretation. Their common claim is that unitarity and the absence of massless exchanges are indispensable, but not sufficient for the canonical bound; one also needs control over the amplitude in nonphysical angular configurations. If the nonphysical amplitude grows faster than any fixed power of energy, then a total cross section growing faster than $l\ge 2$3 is not excluded by general principles (Azimov, 2012, Azimov, 2011, Azimov, 2012).

A second point of clarification concerns dispersion relations. Traditional derivations often use single or double dispersion relations with a finite number of subtractions, but one modern line of analysis argues that these assumptions are stronger than necessary. In that view, the truly essential ingredients are partial-wave unitarity, the location of crossed-channel singularities, and the asymptotics of Legendre functions entering the Froissart-Gribov representation (Azimov, 2011).

In present-day hadron structure studies, the significance of the projections is different but equally structural. In DVCS they are treated as observable, model-discriminating quantities that connect experimental amplitudes, GPD models, Mellin moments, and potentially lattice-QCD form factors. This suggests a broadening of their role: from a tool for bounding asymptotic cross sections to a systematic language for decomposing the response of hadrons to non-local QCD probes with definite $l\ge 2$4-channel quantum numbers (Semenov-Tian-Shansky et al., 2023).