Monotonicity Theorems and Flow Constraints

Monotonicity theorems compare the number of effective degrees of freedom along renormalization-group flow, but the rigorous quantity and the strength of the statement depend on dimension. Two-dimensional $c$, three-dimensional sphere free energy, and four-dimensional $a$ are not interchangeable functions. Each theorem requires unitarity or reflection positivity, locality, relativistic invariance, and a controlled pair of endpoints in a form specific to its proof.

Required background. Local RG and Weyl Consistency Conditions separates identities from positivity. Sphere Partition Functions and Universal CFT Data fixes $F$ and type-A response. Helpful background. Anomaly Matching constrains endpoints independently; UV and IR Fixed Points fixes flow orientation.

Three dimension-specific endpoint theorems

Use the convention that the RG flows from a UV CFT to an IR CFT as distances increase.

Two dimensions

For a unitary, local, Poincaré-invariant two-dimensional QFT, Zamolodchikov constructs a function $c(r)$ from stress-tensor two-point functions such that

$$\frac{dc(r)}{d\log r}\le0,$$

and at conformal fixed points it equals the Virasoro central charge in the standard normalization. Therefore

$$c_{\rm UV}>c_{\rm IR}$$

for a nontrivial flow between fixed points. This is a strong statement: an explicit monotone function exists along the flow Zamolodchikov 1986.

Three dimensions

For a unitary relativistic flow between three-dimensional CFTs, strong subadditivity and Lorentz symmetry of vacuum entanglement imply the endpoint inequality

$$F_{\rm UV}>F_{\rm IR}, \qquad F=-\log|Z_{S^3}|.$$

The proof is naturally phrased through the renormalized entanglement entropy of a disk Casini and Huerta 2012. Identifying its endpoint constants with sphere free energies requires the CFT vacuum map and the standard treatment of topological or gapped sectors.

Four dimensions

For unitary, local, Lorentz-invariant four-dimensional flows with suitable UV and IR behavior, the dilaton scattering argument gives

$$a_{\rm UV}>a_{\rm IR}$$

for a nontrivial flow. The positive dispersive integral is tied to the forward dilaton amplitude, while its low-energy coefficient is $a_{\rm UV}-a_{\rm IR}$ Komargodski and Schwimmer 2011, §§2–3. This proves the endpoint inequality; it does not by itself provide a unique scheme-independent $a(\mu)$ at every intermediate scale.

Weak, strong, and gradient statements

Three claims should be kept distinct:

• weak: a fixed-point quantity satisfies $Q_{\rm UV}>Q_{\rm IR}$;
• strong: a physical function $Q(r)$ is monotone all along the flow;
• gradient: beta functions are gradients of a potential with a positive metric.

The two-dimensional theorem supplies all three in an appropriate scheme. The established three- and four-dimensional endpoint theorems do not automatically imply a global gradient formula on the full coupling space. Perturbative local-RG constructions can yield candidate interpolating functions in restricted regimes, with their own scheme and positivity conditions.

The theorem matrix

The dimension-specific portion of the chapter’s scheme and flow table is:

Dimension and quantity	Proven conclusion	Core hypotheses	Endpoint class	Outside the theorem
2D $c$	monotone function and $c_{\rm UV}>c_{\rm IR}$	unitarity, locality, Poincaré invariance	conformal endpoints	nonunitary/logarithmic flows
3D $F$	$F_{\rm UV}>F_{\rm IR}$ via entanglement monotonicity	unitarity, Lorentz invariance, vacuum locality	CFT endpoints, with gapped/topological terms handled	arbitrary thermal states or nonrelativistic flow
4D $a$	$a_{\rm UV}>a_{\rm IR}$	unitarity, locality, Lorentz invariance, dispersive assumptions	well-defined UV and IR behavior	a universal pointwise $a(\mu)$ without extra input
Boundary/defect quantities	particular $g$, $b$, or defect inequalities in specified dimensions	ambient plus defect unitarity and the proof’s geometry	specified boundary/defect fixed points	a universal codimension-independent theorem
Higher-dimensional candidates	model- or regime-dependent constraints	varies	varies	a general theorem inferred by analogy

The strict sign can become equality for a physically trivial flow or for decoupled sectors treated inconsistently. Gauge zero modes, accidental symmetries, moduli, spontaneous breaking, and topological IR theories require the endpoint observable to include all surviving degrees of freedom.

Applying a flow constraint

1. Identify the spacetime dimension and the exact endpoint invariant.
2. Confirm unitarity/reflection positivity, locality, and Lorentz symmetry in the theorem’s sense.
3. Establish that both endpoints exist and include decoupled, topological, or Goldstone sectors.
4. Match anomaly and sphere conventions between UV and IR.
5. Check relevant deformations and symmetry breaking rather than comparing unrelated CFTs.
6. State only the theorem’s strength: endpoint order, monotone function, or gradient formula.

A failed inequality can reveal a missing sector or normalization mismatch. It is not automatically a counterexample to the theorem.

Boundary and defect analogues belong to Defect Entropy and Monotonicity, where support dimension, codimension, and subtraction are specified. General RG mechanics and EFT threshold matching remain in the renormalization volume.

Evidence scope

The theorem status and qualifications on this page were checked against the cited sources through 9 August 2026. The durable claims are the dimension-specific theorems above, not a ranking of current numerical tests. Proposed higher-dimensional, boundary, or defect monotones should be presented with their precise proof or counterexample domain rather than promoted by analogy.

Common pitfalls

Using $F$ in even dimensions without defining the universal part. Even-dimensional sphere finite terms are scheme dependent. Use the type-A logarithmic coefficient instead.

Turning an endpoint theorem into a pointwise flow function. The four-dimensional $a$ theorem orders endpoints; a global monotone $a(\mu)$ requires additional construction.

Dropping IR sectors. Goldstone bosons, topological field theories, and decoupled free fields contribute to endpoint data. Omitting them can reverse an otherwise valid comparison.

Exercises

Classify the statement “$a_{\rm UV}>a_{\rm IR}$” as weak, strong, or gradient.

Solution

It is the weak, endpoint form. The dilaton proof establishes the ordered fixed-point values. It does not by itself construct a unique scheme-independent function at every intermediate scale or express all beta functions globally as its gradient.

References

• Casini, H., and Huerta, M. “On the RG Running of the Entanglement Entropy of a Circle.” Physical Review D 85, 125016 (2012). arXiv. DOI.
• Komargodski, Z., and Schwimmer, A. “On Renormalization Group Flows in Four Dimensions.” Journal of High Energy Physics 2011, 099 (2011). arXiv. DOI.
• Zamolodchikov, A. B. “Irreversibility of the Flux of the Renormalization Group in a 2D Field Theory.” JETP Letters 43 (1986): 730–732. INSPIRE record.