QCD Radiation, Jets, and Event Shapes

Soft and collinear emissions are individually singular in perturbative QCD, yet suitably defined energy-flow observables are finite. Jets and event shapes achieve this by assigning unresolved states the same measured value. Their calculability is therefore a property of the measurement definition plus real–virtual cancellation, not of a visual resemblance between a parton and a spray of hadrons.

Required background. Inclusive annihilation and the emergence of jets supplies the real–virtual origin of multijet events. Jets and event-shape observables supplies measurement functions and infrared safety.

Helpful background. Sudakov logarithms and resummation supplies the all-order organization required near exclusive limits.

Singular radiation and color

For a resolved quark branching $q\to qg$, with $z$ the daughter-quark energy fraction and $\theta$ the opening angle, the leading collinear probability has the form

$$d\mathcal P_{q\to qg} \simeq\frac{\alpha_s}{2\pi} C_F\frac{1+z^2}{1-z}\,dz\, \frac{d\theta^2}{\theta^2}.$$

The limits $z\to1$ and $\theta\to0$ expose the soft and collinear enhancements. In an inclusive splitting kernel, plus prescriptions and virtual endpoint terms complete this expression. Gluon radiation carries $C_A$ rather than $C_F$ in the corresponding soft limit, so quark and gluon jets have different radiation patterns, subject to observable definition and nonperturbative corrections.

At angles unable to resolve individual charges in a branching, soft radiation couples coherently to their total color charge. Angular ordering in a leading parton-shower picture is one implementation of this coherence, but a shower’s ordering variable and recoil prescription are algorithmic choices whose accuracy must be validated against the targeted logarithms.

Infrared and collinear safety

Let $V_n(p_1,\ldots,p_n)$ be an observable on an $n$-parton state. The defining unresolved limits are

$$V_{n+1}(p_1,\ldots,p_n,k\to0)=V_n(p_1,\ldots,p_n)$$

and

$$V_{n+1}(\ldots,zp,(1-z)p,\ldots)=V_n(\ldots,p,\ldots).$$

When these limits hold smoothly enough, the observable does not distinguish states responsible for a real–virtual singularity, so the cancellation survives differentially. IRC safety is necessary for an ordinary fixed-order partonic prediction; it does not guarantee the absence of large logarithms, hadronization effects, non-global structure, or Glauber sensitivity.

The sequential-recombination anti-$k_t$ algorithm provides a concrete jet test. For objects with transverse momenta $k_{ti}$,

$$d_{ij}=\min(k_{ti}^{-2},k_{tj}^{-2}) \frac{\Delta R_{ij}^2}{R^2}, \qquad d_{iB}=k_{ti}^{-2}.$$

Repeatedly combine the pair with smallest $d_{ij}$, or declare a jet when a beam distance is smallest, using a stated recombination scheme. Soft particles preferentially cluster into nearby hard jets, producing stable cone-like boundaries while preserving IRC safety Cacciari, Salam, and Soyez 2008, §§2–4. The radius $R$, recombination prescription, constituent definition, and treatment of overlapping objects are part of the observable.

Thrust as an auditable event shape

In $e^+e^-$ annihilation, thrust is

$$T=\max_{\boldsymbol n} \frac{\sum_i|\boldsymbol p_i\!\cdot\boldsymbol n|} {\sum_i|\boldsymbol p_i|}, \qquad \tau=1-T.$$

Two pencil-like back-to-back jets have $\tau\to0$, while a more isotropic event has larger $\tau$. Add a soft particle of energy $\omega$ at angle $\theta$ to the thrust axis. To leading order in $\omega/Q$,

$$\delta\tau\simeq\frac{\omega}{Q} \left(1-|\cos\theta|\right).$$

This vanishes when $\omega\to0$ and also in the collinear limit $\theta\to0$ or $\pi$, directly verifying both safety conditions. The original thrust construction and its two-jet interpretation were introduced by Farhi 1977, pp. 1587–88.

Safety does not make the $\tau\to0$ fixed-order expansion uniform. Terms appear as distributions containing

$$\alpha_s^n\left[\frac{\ln^m\tau}{\tau}\right]_+, \qquad m\le2n-1.$$

At leading power in the dijet region, a factorization has the schematic form

$$\frac{d\sigma}{d\tau} =H(Q,\mu) \int ds_a\,ds_b\,dk\, J(s_a,\mu)J(s_b,\mu)S(k,\mu) \delta\!\left(\tau-\frac{s_a+s_b}{Q^2}-\frac{k}{Q}\right) +R(\tau).$$

The natural scales are $\mu_H\sim Q$, $\mu_J\sim Q\sqrt\tau$, and $\mu_S\sim Q\tau$. Evolving the hard, jet, and soft functions to common scales resums the logarithms; matching adds the nonsingular remainder $R$. A precision factorization and resummed thrust distribution are developed in Becher and Schwartz 2008, §§2–5.

Observable and regime classification

Feature	Consequence	Required response
small jet radius $R$	logarithms of $R$ and enhanced boundary sensitivity	state radius counting and resum if parametrically large
jet veto or small event shape	Sudakov logarithms of the veto-to-hard ratio	hard–jet/beam–soft factorization and matching
restriction to part of angular phase space	non-global logarithms	identify the non-global evolution or limit the accuracy claim
recoil-sensitive axis	soft recoil moves the measured direction	use a recoil-aware factorization and axis definition
soft scale near $\Lambda_{\mathrm{QCD}}$	leading power corrections or shape functions	supply nonperturbative input and matching
hadron collision	underlying event, pileup, beam remnants, and possible Glauber issues	declare subtraction, grooming, fiducial, and factorization conditions

Jet substructure introduces further scales and is not automatically covered by an inclusive jet theorem. Grooming can remove some soft sensitivity while creating transition regions and new logarithms; its parameters belong in the measurement definition.

Checks and failure modes

Unresolved-emission test. Add a particle with decreasing energy, then split a particle into increasingly collinear daughters. The implemented observable must approach the unsplit value numerically as well as analytically.

Fixed-order check. Expand any resummed or shower result to the fixed order available and compare singular coefficients and color factors. Matching must subtract, not duplicate, the common terms.

Scale hierarchy check. Keep profile scales in their natural regions and merge them smoothly where the hierarchy ends. Independent arbitrary variations can violate cancellations; use correlated variations tied to the factorization relation.

Hadronization check. Compare the induced soft scale with $\Lambda_{\mathrm{QCD}}$. A small correction in an inclusive rate can become leading near an event-shape endpoint.

Definition check. Quote algorithm, radius, recombination, constituent inputs, grooming, axis, cuts, and binning. “The jet cross section” is not a unique observable.

Common pitfalls

Equating IRC safety with small uncertainty. Safety removes uncancelled singularities. It says nothing by itself about convergence, large logarithms, nonperturbative size, or detector corrections.

Using a shower as an all-order proof. A shower implements selected logarithms with modeling choices. Its accuracy is established by expansion, analytic comparisons, and observable-specific validation.

Treating every soft effect as hadronization. Perturbative soft functions, underlying event, pileup, and nonperturbative fragmentation have different origins and correlations.

Handoff

An analyzable radiation observable is the tuple

$$\left\{V\text{ or jet algorithm},\ \mathcal M,\ Q_i, \text{IRC limits},\ \text{factorization theorem}, \text{logarithmic accuracy},\ \text{matching}, \text{nonperturbative and beam corrections}\right\}.$$

For operator-based energy flow in a conformal setting, continue to event shapes and energy correlators. For the complete error propagation of a QCD observable, continue to the QCD prediction and uncertainty record.

References

• Becher, Thomas, and Matthew D. Schwartz. “A Precise Determination of $\alpha_s$ from LEP Thrust Data Using Effective Field Theory.” Journal of High Energy Physics 2008, no. 7 (2008): 034. DOI. Open PDF.
• Cacciari, Matteo, Gavin P. Salam, and Gregory Soyez. “The Anti-$k_t$ Jet Clustering Algorithm.” Journal of High Energy Physics 2008, no. 4 (2008): 063. DOI. Open PDF.
• Farhi, Edward. “A QCD Test for Jets.” Physical Review Letters 39, no. 25 (1977): 1587–88. DOI.