Join short-distance coefficients to long-distance inputs

For a sufficiently inclusive high-scale hadron collision, the leading-power collinear formula convolves two incoming PDFs with a partonic hard coefficient. The incoming momentum fractions are integrated over the kinematically allowed region. The coefficient includes flux, spin and color averaging, hard phase space and the specified measurement. PDFs are dimensionless densities per unit momentum fraction; an integrated hard coefficient carries cross-section units. The convolution therefore retains the correct dimensions without interpreting each PDF as a normalized single-particle probability.

The theorem applies to defined classes of observables and scale hierarchies, rather than every hadronic measurement. Power corrections arise from long-distance physics relative to the hard scale. Multiple collisions, restrictive gaps, spectator interactions and transverse-momentum measurements may demand other factorization structures or invalidate a naive formula. Even when leading-power factorization holds, logarithms of additional small scales can spoil fixed-order convergence and need resummation. Calling a process hard is a starting condition, not a complete accuracy argument.

A fragmentation function describes production of a named hadron carrying a fraction of an outgoing parton’s longitudinal momentum within a factorized inclusive description. It is not the inverse of a PDF and cannot generally be computed by letting a free perturbative quark evolve to zero energy. Its boundary dependence on the hadron species is nonperturbative; perturbative evolution then controls scale dependence. Fragmentation-function universality enables comparisons among suitable processes while their fits retain experimental and parametrization uncertainties.

For an inclusive jet, an infrared-safe jet algorithm combines the partonic final state into the measured object, so an identified-hadron fragmentation function is not automatically inserted. Hadronization still gives power corrections. Conversely a single identified hadron is not collinear safe by itself in a purely partonic calculation: collinear singular structure must be assigned to fragmentation functions. Keeping the measured object explicit prevents an otherwise reasonable convolution from omitting its essential long-distance input or double counting a correction.

Renormalization scale controls the definition of alpha s, while factorization scales split collinear contributions between coefficients and nonperturbative functions. Their dependence cancels from the all-order physical prediction within the factorization approximation. At fixed order the cancellation is incomplete by terms beyond that order. A PDF set evolved at one accuracy should therefore be combined with a coefficient and coupling at a compatible accuracy and scheme; changing only one ingredient is not a consistent scale comparison.

A scale envelope is a diagnostic of missing perturbative terms, not a confidence interval with guaranteed statistical coverage. It does not automatically include PDF errors, parametrization bias, power corrections, electroweak effects or a factorization failure. Accidental cancellations can make an envelope misleadingly narrow. The illustrative calculation below integrates simple supplied functions exactly, while real predictions require multidimensional kinematics and fitted inputs. The probability instrument checks normalization arithmetic only and has neither momentum-fraction convolutions nor correlated PDF uncertainty eigenvectors.