research30 min studyLesson 680 in Physics

Real–virtual cancellation in simple processes

Combine regulated endpoint terms into a plus distribution and distinguish universal singular structure from supplied hard matching constants.

Before you begin

Infrared divergences and inclusive observables

On this page

The question for today

What you’ll understand

Cancel a shared endpoint regulator at the weighted-integral level.
Evaluate a plus distribution on simple smooth measurement functions.
Explain why singular cancellation does not determine finite hard-process terms.

Real and virtual corrections become meaningful together only when they describe the same measurement and use compatible regulators. A one-dimensional endpoint model makes this cancellation explicit without pretending to compute a complete gauge-theory process. It captures the algebra behind plus distributions: the singular real-emission kernel acts on the difference between a measurement and its unresolved limit. A finite hard matching term remains independent information. This separation prevents a successful logarithm cancellation from being mistaken for a full next-order prediction.

Match the measurement at the singular endpoint

Let x be a dimensionless resolved energy fraction, with the unresolved configuration at x equal to one. Use a dimensionless cutoff delta to stop the real integral at one minus delta. A smooth measurement weight f approaches a finite value at the endpoint. The real contribution integrates f divided by one minus x; the corresponding virtual singular term multiplies the endpoint value by logarithm of delta. Their coefficients must match because both describe the same unresolved limit of the physical event.

Add and subtract the endpoint value inside the real integral. The piece containing that constant produces minus logarithm of delta and cancels the virtual logarithm. The remaining numerator vanishes linearly or faster near the endpoint for a smooth f, so its integral is finite. A discontinuous or unresolved-unsafe measurement can fail this argument. The cancellation concerns an infrared regulator after ultraviolet renormalization; it is not implemented by assigning a detector-dependent finite ultraviolet counterterm.

R_\delta[f]=a\int_0^{1-\delta}\frac{f(x)}{1-x}\,dx,\qquad V_\delta[f]=a f(1)\ln\delta\qquad(0<\delta<1)

a is a small dimensionless perturbative coefficient, delta is an unphysical endpoint regulator, and f is a dimensionless smooth measurement weight. These terms define a toy singular kernel, not a complete real or virtual matrix element.↔ Scroll wide equations horizontally.

The plus distribution is a rule for integration

Taking the cutoff to zero defines the plus distribution through its action on a test function. It is not an ordinary positive function whose value can be read at x equal to one. On a constant test function, its integral vanishes because the real and virtual singular pieces cancel completely in this normalized toy kernel. On a nonconstant weight, a finite remainder survives and can have either sign. A negative next-order contribution is not by itself a negative probability for a separately observed event.

For f equal to x, the subtraction numerator is x minus one, so dividing by one minus x leaves minus one and the integral is minus one. For f equal to x squared, the quotient is minus x minus one, giving minus three halves. These elementary examples show how the measurement affects the finite remainder even when the endpoint value is the same. A generic ordinary integral of f, such as the area under a parabola, is a different functional and cannot replace the plus-weighted calculation.

\int_0^1dx\left[\frac1{1-x}\right]_+f(x)=\int_0^1dx\,\frac{f(x)-f(1)}{1-x}=\lim_{\delta\to0}\frac{R_\delta[f]+V_\delta[f]}a

This definition assumes sufficient regularity near x = 1, for example a smooth test function. Every term is dimensionless and uses the same endpoint convention and cutoff; other distributions and kinematic variables require their own Jacobians.↔ Scroll wide equations horizontally.

Finite matching is additional physical input

After ultraviolet renormalization and singular cancellation, a complete process also contains finite virtual and real terms that the leading endpoint approximation does not determine. Model them here by a supplied constant C multiplying the Born-like endpoint contribution. A weighted cross section combines that constant with the plus-distribution remainder. Its numerical value is stipulated, not derived from the cancellation theorem. Changing the hard process, renormalization prescription, or measurement can change finite terms while leaving the universal singular mechanism intact.

A regulator-independence check varies delta and verifies convergence of the combined quantity, while a resolution-independence claim would be different and usually false. Physical detector cuts can remain in a valid prediction. Fixed-order reliability also needs the perturbative correction to remain appropriately small; large endpoint logarithms can demand resummation. The quadratic instrument computes ordinary height, slope, and area only. It helps contrast an ordinary test-function integral with the distributional operation, but cannot generate real radiation or combine a gauge-theory loop with phase space.

\sigma[f]=\sigma_0\left\{f(1)+a\left[Cf(1)+\int_0^1dx\left[\frac1{1-x}\right]_+f(x)\right]\right\}+O(a^2)

sigma_0 has cross-section units, while a, C, f, and the plus integral are dimensionless. This is a declared endpoint model with a supplied finite matching constant; it does not claim to be the full correction for a named physical process.↔ Scroll wide equations horizontally.

Work it through

A finite weighted toy cross section

Take $f(x)=x$ , $a=0.02$ , supplied $C=3$ , and $\sigma_0=100\,\mathrm{fb}$ . Evaluate the finite endpoint contribution and the modeled weighted cross section through first order.

At finite cutoff, the real integral divided by a is $-(1-\delta)-\ln\delta$ . The virtual term divided by a is $\ln\delta$ because $f(1)=1$ . Their sum is $-1+\delta$ .
Taking delta to zero leaves the plus-weighted integral minus one. Add the independently supplied hard constant three, giving a total finite coefficient $3-1=2$ .
The first-order model gives $\sigma=100[1+0.02(2)]\,\mathrm{fb}=104\,\mathrm{fb}$ . The singular logs disappeared, while the finite matching input and measurement dependence remained.

The result

The plus contribution is −1 and the stipulated endpoint model gives 104 fb; this is not a full calculated QED or QCD process rate.

The living notebook

Explore the model

Interactive lab

Interactive lab · function

Ordinary area and distributional weighting differ

Reset and set coefficient $a=1$ , position $x=1$ ; predict ordinary signed area $1/3$ under $y=x^2$ . Repeat from Reset with both controls $a=1$ , $x=0.5$ ; predict area $1/24$ . These ordinary areas differ from the plus-weighted value −3/2 for the test function $x^2$ on [0,1]. The lab has no endpoint-distribution or real–virtual cancellation control.

y = ax²TangentSigned integral

y at chosen x: 0.608
Tangent slope · 2ax: 1.35
Signed area · ax³/3: 0.182

Coefficient a0.75Position x0.9

t = 0 s

This lab uses only y = ax². Playback traces the curve within the fixed integration interval from 0 to x; it does not estimate numerical error or change the chosen point. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forA plus distribution specifies a subtraction rule inside an integral, not merely a plotted function with a difficult endpoint.

Pause & reason

Check your understanding

Interactive checks need JavaScript. You can still read the explanation.

Read the checkpoint explanation

The definition subtracts the same constant endpoint value, so the numerator is identically zero. Finite hard matching terms are separate contributions.

Your turn

Try it yourself

In the same declared model, use $f(x)=x^2$ , $a=0.04$ , supplied $C=2$ , and $\sigma_0=50\,\mathrm{fb}$ . Compute the plus-weighted integral and the first-order weighted cross section.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Factor x squared minus one before dividing by one minus x; retain f(1) in the endpoint term.

Compare with the solution

The subtraction quotient is $(x^2-1)/(1-x)=-(x+1)$ , so its integral from zero to one is $-3/2$ . Since $f(1)=1$ , the finite coefficient is $2-3/2=1/2$ . Thus $\sigma=50[1+0.04(1/2)]\,\mathrm{fb}=51\,\mathrm{fb}$ . The cutoff has canceled, but this finite value still depends on the stipulated matching constant and the chosen measurement weight.

Keep the idea

Take it with you

Real and virtual endpoint terms cancel only with matched coefficients and the same unresolved measurement.
A plus distribution acts on test functions through endpoint subtraction and need not give a positive correction.
Finite matching terms and physical resolution dependence survive beyond cancellation of an unphysical regulator.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.