Parton distributions and factorization

In a collinear high-energy approximation, parton momenta are $x_1P_1$ and $x_2P_2$. Neglecting proton masses gives $\hat s=x_1x_2s$. A high machine center-of-mass energy therefore permits, but does not guarantee, a high partonic invariant mass. The probability weight depends strongly on which parton species and momentum fractions are required. A rare large-x configuration can suppress production even before the short-distance amplitude is considered.

A parton distribution function $f_{i/p}(x,\mu_F)$ encodes scale- and scheme-dependent proton matrix elements. At leading order it admits a useful density interpretation, but beyond leading order it is not an ordinary independent-event probability for a permanent constituent. Integrals of individual sea and gluon densities need not count a finite number of objects. Valence-number and total momentum sum rules instead constrain specific combinations, with all relevant parton species included at the stated scale.

The renormalization scale $\mu_R$ specifies the short-distance coupling and renormalized parameters. The factorization scale $\mu_F$ separates collinear contributions assigned to proton structure from those assigned to the hard coefficient. Evolving PDFs with their splitting kernels compensates the coefficient's factorization-scale dependence order by order. The two scales may be chosen equal for convenience, but their conceptual jobs remain different and their numerical values are not directly measured event properties.

At all orders, a consistent physical prediction is independent of those arbitrary separations. At finite order, residual variation probes some missing higher-order terms. A scale-variation envelope is useful evidence, not a statistical confidence interval or a guarantee against every omitted effect. PDF-fit uncertainties, parameter uncertainties, missing perturbative orders, and nonperturbative corrections are distinct sources. Combining them requires an explicit convention, particularly when two measurements share the same fitted proton inputs.

An inclusive hard observable with a scale well above hadronic scales is a favorable setting for collinear factorization. Restricting radiation, resolving very small transverse momentum, or probing multiple hard scatterings can demand more detailed factorization structures or additional nonperturbative functions. A formula written for an inclusive cross section cannot automatically predict an arbitrarily exclusive detector-level event pattern. Identify the observable and its scale hierarchy before deciding which proton information is sufficient.

The predictive test is to fit allowed universal inputs from suitable data and use them in genuinely independent measurements. Fitting and testing on overlapping event samples can obscure that independence. Event generators add radiation, hadronization, and underlying-event models to construct complete events, but a generated event is still model-dependent. Compare data with a defined perturbative calculation and uncertainty budget rather than assuming that an attractive simulated event display is a direct photograph of the incoming partons.