Measurement Functions and Inclusive Observables

A measurement function is the mathematical definition of what an idealized experiment counts. It maps each final-state configuration to a weight, bin, or test function, so amplitudes for different multiplicities can be combined into one observable. Making that map explicit fixes the normalization, the inclusiveness of the final-state sum, and the unresolved limits that later determine whether perturbation theory is finite.

Required background. Cross Sections and Decay Rates supplies flux, phase-space, and symmetry-factor normalizations. The Optical Theorem and Cut Interpretation explains why sums over physical intermediate states implement unitarity.

Weighted sums over final states

For a scattering process with fixed incoming state, define a family $F_n$ on the physical $n$-body phase space. The associated cross section is

$$\begin{aligned} \sigma[F] ={}&\frac{1}{\mathcal F}\sum_{n}\frac{1}{S_n} \int \mathrm d\Phi_n(P;\{p\}_n)\\ &\times\overline{|\mathcal M_n(\{p\}_n)|^2}\,F_n(\{p\}_n). \end{aligned}$$

Here $P$ is the total incoming momentum, $S_n$ removes overcounting of identical final particles, and $\mathcal F=4\sqrt{(p_A\!\cdot p_B)^2-m_A^2m_B^2}$ is the invariant flux for two incoming particles. The bar denotes the declared final-state sums and initial-state averages. These normalizations follow the standard invariant scattering convention Weinberg 1995, §§ 3.1–3.4, pp. 107–141. The measurement function may be dimensionless, as for a cut or event count, or dimensionful, as for an energy moment.

The same notation covers several familiar objects:

• a total inclusive rate has $F_n=1$ for every admitted state;
• a fiducial rate has $F_n$ equal to a product of indicator functions for the accepted region;
• a moment has $F_n=w(O_n)$ for an event observable $O_n$;
• a distribution is defined weakly by $F_n(O)=\delta(O-O_n)$, or more safely by integration against a smooth test function;
• a histogram bin $B=[O_a,O_b)$ uses $F_n^{(B)}=\mathbf1_B(O_n)$.

This formulation of jet cross sections in terms of functions on each multiplicity is used explicitly in Catani and Seymour 1997, § 2.1, pp. 297–299. It is not merely notation: the relationship between $F_n$ and $F_{n+1}$ in unresolved limits decides whether cancellation can occur.

Bins, densities, and normalization

A differential cross section is a distribution. Its operational meaning is

$$\begin{aligned} \int \mathrm dO\,f(O)\frac{\mathrm d\sigma}{\mathrm dO} ={}&\frac{1}{\mathcal F}\sum_n\frac{1}{S_n}\int \mathrm d\Phi_n\\ &\times\overline{|\mathcal M_n|^2}\,f(O_n) \end{aligned}$$

for suitable test functions $f$. A bin is therefore primary and a pointwise density is an idealization. This matters at a Born endpoint, where a contribution such as $C\,\delta(O-O_0)$ is perfectly meaningful after binning but cannot be interpreted as an ordinary finite function.

For a normalized shape,

$$\frac{1}{\sigma_{\mathrm{norm}}}\frac{\mathrm d\sigma}{\mathrm dO},$$

the numerator and denominator must be stated at compatible perturbative orders. Expanding the ratio and dividing two truncated numbers are different prescriptions beyond the claimed order. If

$$\begin{aligned} N&=N_0+\alpha N_1+O(\alpha^2),\\ D&=D_0+\alpha D_1+O(\alpha^2), \end{aligned}$$

then the consistently expanded ratio is

$$\frac ND =\frac{N_0}{D_0} +\alpha\left( \frac{N_1}{D_0}-\frac{N_0D_1}{D_0^2} \right)+O(\alpha^2).$$

Reporting which prescription was used prevents an apparent disagreement that is only a higher-order convention.

Inclusiveness is relative to a resolution

“Inclusive” does not mean that every final state is ignored. It means that states not distinguished by the observable receive the same weight. A rate can be inclusive over soft photons but exclusive in the number of resolved jets; a lepton-energy spectrum can sum over hadronic states while resolving one lepton momentum.

Unitarity supplies cancellations only across the degenerate states actually summed with compatible weights. If two configurations become experimentally indistinguishable in a soft or collinear limit, a perturbatively safe measurement must approach the same value on both. The next page turns this statement into the infrared-and-collinear-safety conditions.

This distinction also separates an idealized theory observable from detector response. $F_n$ acts on exact final-state momenta and quantum numbers. Smearing, inefficiencies, reconstruction, migration matrices, and likelihoods are additional maps; they should not be silently folded into a parton-level definition.

Example: an energy-flow moment

In the center-of-mass frame with energy $Q$, consider

$$V_n(\{p\}_n)=\frac1Q\sum_{i=1}^n E_i\,w(\widehat{\boldsymbol p}_i),$$

where $w$ is a bounded angular weight. The moment $\langle V\rangle$ uses $F_n=V_n$. If a particle of momentum $p$ splits collinearly into $zp$ and $(1-z)p$, the two daughters point in the same direction and their energies add, so $V_{n+1}\to V_n$. Adding a particle with energy $E_s\to0$ changes $V$ by $O(E_s/Q)$. The observable therefore passes the elementary unresolved-limit test provided $w$ remains finite and its angular boundaries have a declared prescription. A bounded step boundary can still be IRC safe, although it may introduce non-global logarithms.

By contrast, the bare particle multiplicity $N_n=n$ changes by one under either an arbitrarily soft emission or an arbitrarily collinear splitting. In a massless theory it is not an infrared-and-collinear-safe partonic observable, even though it is easy to describe experimentally after hadronization.

Common pitfalls

Defining a plot instead of an observable. Axis labels and selection prose are not enough. Write $F_n$, including cuts, recombination, normalization, and the treatment of boundaries.

Dropping symmetry or averaging factors. The measurement function does not repair an inconsistent amplitude normalization. Keep identical-particle factors, initial averages, and discrete sums explicit until conventions have been matched.

Calling a quantity inclusive without naming the unresolved states. Inclusiveness is always relative to what is not distinguished. State which radiation, flavors, spins, or multiplicities are summed.

Treating a delta-function density pointwise. Singular distributions are tested by bins or smooth weights. A finite bin integral can coexist with a distributional endpoint.

Check your understanding

Define a two-bin measurement for the energy-flow moment above and show that the two bin indicators sum to one away from the shared boundary. Then verify that the sum of the two bin cross sections equals the corresponding inclusive rate and that changing the convention for the shared endpoint affects only a set of measure zero unless a distributional endpoint sits there.

Where to continue

• Infrared and Collinear Safety tests the unresolved limits of the entire family $F_n$.
• Real–Virtual Cancellation and Subtraction constructs finite integrals once those limits agree.
• Jets and Event-Shape Observables supplies important energy-flow examples.

References

• Catani, Stefano, and Michael H. Seymour. “A General Algorithm for Calculating Jet Cross Sections in NLO QCD.” Nuclear Physics B 485 (1997): 291–419; erratum 510 (1998): 503–504. DOI. Open preprint.
• Weinberg, Steven. The Quantum Theory of Fields, Volume I: Foundations. Cambridge: Cambridge University Press, 1995, §§ 3.1–3.4, pp. 107–141. DOI.