Skip to main content

Scale dependence of a model

Defining a flow​

A Wilsonian model specifies its fields, kinetic terms, interactions and regulator. Eliminating a momentum shell generates an effective action. Extracting a few running couplings from it requires a declared projection onto operators. The 21 unordered edges and seven lines of a Fano plane count possible indices. A loop coefficient also depends on the propagator, vertex normalization, dimension and contractions of those indices.

An exact sign check​

Consider a scalar truncation whose equation on a scale interval is dλ/dlog⁡μ=−Δλd\lambda/d\log\mu=-\Delta\lambda with fixed Δ>0\Delta>0. The solution for a nonzero initial coupling is

λ(μ)λ(μ0)=(μμ0)−Δ.\frac{\lambda(\mu)}{\lambda(\mu_0)} =\left(\frac{\mu}{\mu_0}\right)^{-\Delta}.

Taking Δ=5/42\Delta=5/42 and μ/μ0=10−61\mu/\mu_0=10^{-61} gives a factor 10305/4210^{305/42}, about 107.2610^{7.26}. The magnitude grows towards the infrared under the stated equation. Whether a scalar truncation describes a measured coupling requires matching its running and boundary value to independent observables.

Comparing two scales​

For a positive constant rescaling α=cλ\alpha=c\lambda, the beta functions obey βα=cβλ\beta_\alpha=c\beta_\lambda. A positive quadratic beta function in four dimensions cannot reproduce the negative one-loop beta function of asymptotically free QCD by this rescaling. A test must specify gauge and matter dynamics and compare the coupling at a scale withheld from calibration. A different scale comparison is already defined for transport times on networks. Its observables are the measured times of multiple probes. Network scale tests.