research38 min studyLesson 696 in Physics

Wilson-coefficient running and operator mixing

Evolve matched coefficients between scales and track the compensating operator evolution with a declared transpose convention.

Before you begin

On this page

The question for today

What you’ll understand

Distinguish a matching boundary condition from an RG evolution equation.
Keep coefficient and operator anomalous-dimension signs consistent.
Solve a triangular mixing example and test scale independence.

Matching specifies the Wilson coefficients at a threshold, while renormalization-group evolution transports those coefficients within an EFT containing a fixed set of active fields. These are different operations. Running reorganizes logarithms of scale ratios, and mixing means that a coefficient absent at the matching scale can appear at another scale. Neither a single coefficient nor an operator matrix element is generally observable by itself. Their scale dependence cancels in the complete physical combination through the accuracy of the calculation.

Fix the matrix convention before solving the flow

Write a set of operators of the same mass dimension as a column O and their coefficients as a column C, with the interaction proportional to the transpose of C times O. For this lesson define the coefficient evolution matrix Gamma directly by $dC/d\ln\mu=\Gamma C$ . The corresponding operator evolution must then be minus the transpose of Gamma acting on O. Other references define the anomalous-dimension matrix on operators first, so their transpose and sign may appear in different places.

Differentiating the coefficient-operator product verifies the cancellation without relying on a remembered convention. This check is particularly useful after a basis change. Mixing can arise because a divergent loop with one operator insertion has the tensor structure of another operator. A symmetry can forbid some entries, but a zero matching coefficient alone is not such a symmetry. The basis must be closed under renormalization to the order being used, including any required evanescent structures in dimensional regularization.

\mathcal L\supset\Lambda^{4-d}\mathbf C^T\mathbf O,\qquad \frac{d\mathbf C}{d\ln\mu}=\Gamma\mathbf C,\qquad \frac{d\mathbf O}{d\ln\mu}=-\Gamma^T\mathbf O,\qquad \frac{d}{d\ln\mu}(\mathbf C^T\mathbf O)=0

The displayed operators share canonical dimension d, their C coefficients are dimensionless, and Λ is fixed. Gamma is defined here as the coefficient evolution matrix. The last equality holds to the order at which coefficients and matrix elements are treated consistently.↔ Scroll wide equations horizontally.

Integrate a declared triangular toy matrix

To isolate the mechanism, take a constant two-by-two matrix with only the upper-right entry gamma nonzero. This is a toy approximation that freezes any coupling dependence and is not a claimed anomalous dimension of a particular ultraviolet theory. Its square vanishes, so its exponential terminates after the term linear in the logarithm. The second coefficient stays constant, while the first receives a contribution from it.

For downward evolution the logarithm of mu divided by the matching scale is negative. A positive mixing entry can therefore generate a negative shift; confusing this with an infrared scale variable of opposite sign reverses the answer. General matrices need not be nilpotent or constant. Running couplings enter Gamma, and matrices at different scales need not commute, requiring an ordered evolution rather than one unqualified matrix exponential.

\Gamma=\begin{pmatrix}0&\gamma\\0&0\end{pmatrix},\quad \ell=\ln\frac\mu M,\quad U(\mu,M)=I+\Gamma\ell,\qquad C_2(\mu)=C_2(M),\quad C_1(\mu)=C_1(M)+\gamma\ell C_2(M)

γ and ell are dimensionless and Gamma²=0 in this toy model. M is the matching scale. The exact finite formula applies to this constant triangular system; an actual loop anomalous dimension generally depends on running couplings.↔ Scroll wide equations horizontally.

Running sums logarithms but does not add a missing threshold

When a small loop factor multiplies a large scale logarithm, fixed-order expressions can develop large repeated powers. Solving the RG equations resums the appropriate logarithmic series, provided the underlying couplings remain in a controlled regime. Matching near each heavy threshold and evaluating matrix elements near their characteristic light scale avoids large logarithms in either boundary calculation. Running alone cannot supply an unknown finite threshold constant or the contribution of an omitted operator.

Crossing a heavy mass generally changes the active theory, its beta functions, and its operator basis, so a threshold matching step connects the two evolution problems. Within one EFT, residual mu dependence estimates some omitted perturbative terms, but it does not sample every higher-dimensional operator or guarantee a complete truncation uncertainty. A stable prediction checks both the RG cancellation and the separate expansion in inverse heavy scales. These are complementary requirements rather than interchangeable notions of convergence.

Work it through

A downward mixing shift with no new matching calculation

Use the triangular toy matrix with γ=0.020, initial coefficients C1(M)=1 and C2(M)=2, and lower scale μ=Me−2.5. Find both evolved coefficients and the numerical evolution matrix. State which part of this result is an assumption.

The scale logarithm is $\ell=\ln(\mu/M)=-2.5$ . Thus the upper-right evolution entry is $\gamma\ell=0.020(-2.5)=-0.050$ .
The second coefficient remains two. The first becomes $C_1(\mu)=1+(-0.050)(2)=0.90$ . The negative shift follows from downward evolution with the stated sign convention.
The matrix is $U=\left(\begin{smallmatrix}1&-0.050\\0&1\end{smallmatrix}\right)$ . Its terminating exponential is exact for the constant nilpotent toy Gamma, while treating an actual anomalous dimension as that constant matrix would require a separate approximation.

The result

The evolved coefficients are (0.90,2), with an upper-triangular evolution matrix whose off-diagonal entry is −0.050.

The living notebook

Explore the model

Interactive lab

Interactive lab · matrix

An off-diagonal map mixes components

For each fresh comparison, Reset and set BOTH Diagonal scale a=1 and Shear b. Compare b=0 with b=−0.50. Predict selected-final determinants one in both runs and singular values (1,1) versus approximately (1.281,0.781). The geometry starts at identity after Reset and interpolates under Play or Step; readouts always describe the final selected A. This triangular family illustrates a component-mixing map, not a general RG solver. Unit determinant is a feature of this chosen shear, not a law for Wilson evolution.

Interpolated geometryOriginal unit squareSelected final transform

Area scale · det A: 1.4
Singular values · σmax, σmin: 1.597, 0.876
Condition number · κ₂(A): 1.823

Diagonal scale a1.4Shear b0.6

t = 0 s

The selected matrix is A = [[a,b],[0,1]]. Playback interpolates geometry between the identity and A; readouts always describe A. Its determinant is a and its singular values come from AᵀA. Positive a makes every supported matrix invertible. This two-dimensional family is not a general eigensolver or a learned network. 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 forMixing changes coefficient components even when a special map preserves area; physical invariance requires compensating operator evolution.

Pause & reason

Check your understanding

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

Read the checkpoint explanation

Differentiation gives CᵀΓᵀO from the coefficient and −CᵀΓᵀO from the operator. This explicit check fixes both the transpose and the sign.

Your turn

Try it yourself

For the same γ=0.020 and ell=−2.5, suppose a matrix element at M is O1=3 and O2=4, with coefficients (1,2). Evolve the matrix elements with the inverse-transpose map and verify the amplitude CᵀO at both scales.

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

Give me a hint

Since Gamma²=0, operator evolution is I−Gammaᵀell. Only O2 changes in this convention, while O1 stays fixed.

Compare with the solution

The operator map has lower-left entry $-\gamma\ell=+0.050$ . Thus $O_1(\mu)=3$ and $O_2(\mu)=4+0.050(3)=4.15$ . Initially the amplitude is $1(3)+2(4)=11$ . After running, the coefficients are (0.90,2), giving $0.90(3)+2(4.15)=2.70+8.30=11$ . Holding the operators fixed while evolving coefficients would give a false scale dependence. The exact cancellation here follows the toy matrix; perturbative real-world calculations cancel only up to the order consistently included.

Keep the idea

Take it with you

Matching supplies threshold data; RG equations evolve it within an active theory.
Matrix conventions must fix the coefficient/operator transpose and sign.
Running coefficients alone do not define a scale-independent observable or a complete error estimate.

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.