Fragmentation Functions and Timelike Evolution

A fragmentation function $D_i^h(z,\mu)$ describes how an identified hadron $h$ carries a fraction $z$ of a short-distance parton channel $i$ after unobserved final-state radiation and hadronization. It is a nonperturbative, renormalized matrix element. Timelike evolution predicts its scale dependence, but it is not the spacelike PDF equation with labels casually exchanged.

Required background. DGLAP evolution and scaling violation supplies convolutions, plus distributions, moments, and sum-rule validation.

Helpful background. Jets and event-shape observables supplies the distinction between an identified hadron and an inclusive energy-flow observable.

Identified-hadron factorization

For single-inclusive hadron production in $e^+e^-$ annihilation at hard scale $Q$, define a measured energy fraction such as $z=2P_h\!\cdot q/Q^2$ in the massless limit. A leading-power factorization has the form

$$\frac{d\sigma^h}{dz}(z,Q) =\sum_i\left[C_i^{\mathrm T}\!\left(Q,\mu_R,\mu_F;\mathcal S\right) \otimes D_i^h(\mu_F;\mathcal S)\right](z) +O\!\left(\frac{m_h^2}{z^2Q^2}, \frac{\Lambda_{\mathrm{QCD}}^p}{Q^p}\right),$$ $$[C\otimes D](z)=\int_z^1\frac{d\xi}{\xi}\,C(\xi)D\!\left(\frac{z}{\xi}\right).$$

The coefficient creates a short-distance parton; the fragmentation function describes the inclusive sum over all unobserved states that contain $h$. Both depend on the factorization scheme and scale, and that dependence cancels in the cross section through the calculated order.

A schematic quark definition makes its character clear:

$$D_q^h(z)\ \sim\ \sum_X\int d\lambda\,e^{-i\lambda P_h^+/z} \operatorname{tr}\!\left[ \gamma^+\langle0|W\psi_q(0)|hX\rangle \langle hX|\bar\psi_q(\lambda n)W|0\rangle \right]_{\mathrm{ren}}.$$

The exact prefactor and link geometry depend on convention, but three features are essential: a cut sum over $X$, future-directed final-state Wilson lines, and ultraviolet renormalization. Operator definitions and their momentum sum rule are developed in Collins and Soper 1982, §§5–7, pp. 467–88.

Timelike DGLAP evolution

In the convention where $i$ labels the fragmenting parent,

$$\frac{\partial D_i^h(z,\mu)}{\partial\ln\mu^2} =\sum_j\left[P^{\mathrm T}_{ji}(\alpha_s(\mu)) \otimes D_j^h(\mu)\right](z).$$

The index order encodes the branching $i\to j+cdots$ followed by fragmentation of $j$. At leading order, spacelike and timelike kernels obey a transpose relationship after conventions are aligned. Beyond leading order, analytic continuation, phase space, and scheme choices prevent one from replacing $P^{\mathrm T}$ by a naive transpose of $P^{\mathrm S}$.

Mellin moments again turn convolutions into products:

$$D_{i,N}^h(\mu)=\int_0^1dz\,z^{N-1}D_i^h(z,\mu), \qquad \frac{dD_{i,N}^h}{d\ln\mu^2} =\sum_j\gamma^{\mathrm T}_{ji,N}D_{j,N}^h.$$

This is a matrix evolution problem in quark-singlet and gluon channels. Nonsinglet flavor combinations can evolve separately, but an identified hadron does not obey a PDF-like valence-number normalization.

Momentum conservation and a decisive check

If the sum includes all hadron species and unobserved quantum numbers, energy–momentum conservation implies

$$\sum_h\int_0^1dz\,zD_i^h(z,\mu)=1$$

for each parent parton $i$ in the standard normalization. Differentiate this identity and insert timelike DGLAP. The result requires the corresponding $N=2$ moment of the full splitting matrix to conserve momentum. This check tests simultaneously the kernel orientation, species sum, endpoint terms, and numerical integration.

For a restricted set—one charged hadron species, only a detector acceptance, or a flavor-tag category—the integral need not equal one. Applying the full momentum sum rule to an incomplete set is a normalization error.

Universality and factorization limits

Question	Correct qualification
Can the same collinear $D_i^h$ enter $e^+e^-$, DIS, and hadron collisions?	Yes where a leading-power collinear theorem applies, using a common scheme, scale, and hadron definition.
Is a fragmentation function a hadronization event generator?	No. It is an inclusive one-hadron matrix element and does not specify exclusive multiplicities or correlations.
Does it describe transverse momentum inside a jet?	Not by itself. A TMD fragmentation function or a more differential fragmenting-jet object is then needed.
Is the endpoint $z\to1$ ordinary fixed order?	Not necessarily. Threshold logarithms and nonperturbative power corrections are enhanced.
Is very small $z$ automatically controlled?	No. Hadron-mass corrections, multiplicity logarithms, and soft physics can invalidate a simple truncation.

Universality is always tied to the operator and factorization statement, not to a numerical fit detached from its release, data selection, and covariance. A broad account of perturbative evolution, hadron-mass effects, and fit methodology is given in Albino 2010, §§2–5, pp. 2489–2556.

Checks and failure modes

Space–time label check. Carry an explicit superscript $\mathrm S$ or $\mathrm T$ on kernels and evolution maps. Agreement at leading order is not authorization to reuse a higher-order spacelike table.

Scale cancellation. Evolve $D_i^h$ and the coefficient at a common timelike accuracy; verify that the physical spectrum has only omitted-order $\mu_F$ dependence.

Support and mass check. Enforce the support of the stated $z$ definition. At small $z$, the nominal $m_h^2/(z^2Q^2)$ term can cease to be small even for a large $Q$.

Flavor-tag check. A tag can mix production, decay, and detector definitions with the theoretical hadron label. State precisely which final states are summed before invoking universality or a sum rule.

Common pitfalls

Treating $D_i^h$ as the inverse of a PDF. Initial-state and final-state cut matrix elements have different Wilson-line and analytic structures. Crossing intuition does not replace a timelike factorization theorem.

Summing one hadron species to unity. The momentum sum rule requires the complete hadronic final-state sum. A restricted species carries only part of the parent momentum.

Ignoring fit covariance. Quark flavors and the gluon are correlated by data and evolution. Varying one fitted curve at a time is not generally a faithful uncertainty propagation.

Handoff

The portable timelike object is

$$\left\{D_i^h(z,\mu_0;\mathcal S),\ C_i^{\mathrm T}, P_{ji}^{\mathrm T},\ U^{\mathrm T}(\mu,\mu_0), \text{species sum},\ \text{covariance},\ \text{power domain}\right\}.$$

For observables that sum energy flow into jets rather than tag a hadron, continue to QCD radiation, jets, and event shapes. For transverse-momentum-resolved fragmentation, the rapidity and soft-subtraction logic of TMD factorization is also required.

References

• Albino, Simon. “The Hadronization of Partons.” Reviews of Modern Physics 82, no. 3 (2010): 2489–2556. DOI. Open PDF.
• Collins, John C., and Davison E. Soper. “Parton Distribution and Decay Functions.” Nuclear Physics B 194, no. 3 (1982): 445–92. DOI.