Cross section measurements at colliders translate observed event counts into fundamental quantities that can be compared with theoretical predictions. The measured cross section sigma = (N_signal - N_background) / (efficiency * luminosity) must be corrected for detector acceptance, efficiency, and resolution effects. Fiducial and differential cross sections, corrected through unfolding, provide the most model-independent comparisons with theory.
Cross section measurements are the primary quantitative output of collider experiments. A cross section sigma has units of area (typically picobarns or femtobarns at the LHC) and represents the effective target area for a particular process. Multiplied by the integrated luminosity (the total amount of data collected, in units of inverse cross section), it gives the expected number of events: N = sigma * L. The reverse -- extracting sigma from the observed N after correcting for efficiency and backgrounds -- is the measurement.
The measurement chain proceeds as follows. Events are selected by the trigger and offline analysis cuts. The efficiency of each selection step (trigger, reconstruction, identification, isolation, kinematic cuts) is measured in data using tag-and-probe techniques on known processes. The background is estimated from data-driven methods or simulation and subtracted. The remaining signal event count is divided by the efficiency and luminosity to give the cross section. Systematic uncertainties from each step (efficiency correction, background estimation, luminosity, and theory modeling) are propagated and combined.
Fiducial cross sections restrict the measurement to the kinematic region directly accessible to the detector, avoiding model-dependent extrapolations. A fiducial region is defined at particle level (using stable particles with lifetime > 10 ps) with cuts that closely mirror the detector-level selection. Differential fiducial cross sections -- binned in kinematic variables like p_T, rapidity, jet multiplicity, or angular correlations -- provide the most detailed comparison with theory. They test not just the total rate but the shape of distributions, probing QCD dynamics, PDF effects, and electroweak corrections.
Unfolding is the mathematical procedure that corrects a measured distribution for detector effects. The detector response is encoded in a migration matrix that maps particle-level bins to detector-level bins. Inverting this matrix is ill-conditioned (small statistical fluctuations are amplified into large oscillations), so regularized methods are used. The result is a distribution at particle level that can be compared with any theoretical prediction without passing the prediction through detector simulation. This separation of measurement (corrected to particle level) and theory comparison is a key principle of modern collider physics, ensuring measurements remain useful long after the experiments that produced them.