Finite Counterterms, Schemes, and Multi-Trace Data

After divergent counterterms have been fixed, finite boundary terms still matter—but they do not all mean the same thing. A finite local functional of fixed sources changes the renormalization scheme and contact terms. A Legendre transform or multi-trace boundary functional changes which canonical variable is held fixed and can define a different quantization, deformation, or ensemble. This page gives operational tests that separate those cases.

Required background. Boundary Conditions, Alternate Quantization, and Deformations supplies the two scalar falloffs and their admissible quantizations. Scalar Counterterms and the Renormalized On-Shell Action constructs the finite functional. Renormalized One-Point Functions and the Variational Problem fixes the source–response pairing.

Helpful background. Renormalization Conditions, Schemes, and Finite Parts and Scheme Transformations and RG Invariants give the corresponding field-theory distinctions.

First application. Add a finite scalar counterterm and a double-trace boundary term separately and show which change only contact data and which changes the boundary condition.

Renormalized canonical data

For a scalar in the mass window where both falloffs may be normalizable, write

$$\phi(z,x)=z^{d-\Delta}\alpha(x)+z^\Delta\beta(x)+\cdots, \qquad m^2L^2=\Delta(\Delta-d).$$

In standard quantization, $\alpha$ is the source and the renormalized variation takes the form

$$\delta S_{\mathrm{ren}} =\int d^dx\sqrt{\lvert g_{(0)}\rvert}\, \mathcal B(x)\,\delta\alpha(x),$$

where $\mathcal B$ is a convention-dependent normalization of $\beta$ plus local terms. Thus $(\alpha,\mathcal B)$ are renormalized canonical variables. A finite boundary functional acts either as a canonical transformation of these variables or as a change in the condition imposed on them. The distinction is visible in the variational principle.

Finite local counterterms

Let

$$S_{\mathrm{fin}}[\alpha] =\frac12\int d^dx\sqrt{\lvert g_{(0)}\rvert}\, \alpha\,P(-\nabla^2)\alpha,$$

where $P$ is a polynomial compatible with the symmetries and dimensions. Then

$$\mathcal B\longrightarrow \mathcal B+P(-\nabla^2)\alpha, \qquad G(k)\longrightarrow G(k)+P(k^2).$$

The shift in the two-point function is polynomial in momentum, hence supported at coincident points in position space. It can change one-point conventions, contact terms, and anomaly representatives, but not the nonanalytic structure or separated-point correlator. This is ordinary scheme freedom. The same logic applies to finite local functionals of the boundary metric and background gauge fields, subject to their Ward identities.

Locality is the decisive restriction. A term such as

$$\frac12\int\!\frac{d^dk}{(2\pi)^d}\, \alpha(k)\,\lvert k\rvert^{2\nu}\alpha(-k),$$

with noninteger $\nu$, is generally nonlocal. Calling it a counterterm would change the nonanalytic part of the correlator and therefore the physical separated-point response. It is not a harmless scheme choice.

Multi-trace deformations and mixed conditions

Now add a boundary functional $W(\alpha)$ and allow $\alpha$ to vary. At zero external source the stationary condition is

$$\mathcal B(x)+\frac{1}{\sqrt{\lvert g_{(0)}\rvert}} \frac{\delta W}{\delta\alpha(x)}=0.$$

For $W=\tfrac12 f\int\sqrt{\lvert g_{(0)}\rvert}\,\alpha^2$, this is a mixed boundary condition. With a conventional choice of signs and normalization, the deformed connected two-point function has the large-N form

$$G_f(k)=\frac{G_0(k)}{1+fG_0(k)}.$$

Unlike a polynomial contact shift, this transformation moves poles and changes separated-point physics. It represents a double-trace deformation or a different boundary condition, not merely a subtraction convention. The sign in the denominator follows the sign chosen for $W$; physical statements should be phrased in terms of the stated variational condition.

More general $W$ produces nonlinear mixed conditions and higher multi-trace couplings. At finite radial cutoff, such couplings are naturally momentum dependent. Their running can be scheme dependent, but a change in boundary condition that alters the spectrum remains physical.

Legendre transforms and alternate quantization

Within the Breitenlohner–Freedman window, a Legendre transform exchanges which member of the canonical pair is held fixed. Schematically,

$$\widetilde S[\mathcal B] =S_{\mathrm{ren}}[\alpha] -\int d^dx\sqrt{\lvert g_{(0)}\rvert}\,\alpha\mathcal B,$$

so that $\delta\widetilde S=-\int\alpha\,\delta\mathcal B$. This implements alternate quantization when both falloffs satisfy the required normalizability and stability conditions. Klebanov and Witten 1999, §§2–3 explain the dimension exchange, and Witten 2001, §§2–4 develops the multi-trace prescription.

A Legendre transform is an invertible change of thermodynamic or generating variables only after its domain and convexity properties are fixed. In AdS/CFT it may also select a different CFT quantization. One must therefore state whether two functionals describe the same theory in different variables or different boundary theories.

Classification by observable effect

Operation	Variational data	Effect at separated points	Interpretation
Add a finite local functional of fixed sources	Response shifts by a local source functional	None, apart from distributions at coincidence	Renormalization-scheme change
Redefine an operator by local source terms	One-point and contact conventions shift	Nonlocal response unchanged	Operator convention
Add a nonlocal finite source functional	Response shifts nonlocally	Correlators generally change	New dynamics or an inadmissible subtraction
Impose a mixed boundary condition	Source–response relation changes	Spectrum and poles can move	Multi-trace deformation or ensemble choice
Legendre transform admissible scalar data	Fixed canonical variable is exchanged	Operator dimension and correlator can change	Alternate quantization when allowed

The table is an observable test, not a test based on terminology: calculate the variation, inspect locality, and compare separated-point quantities.

Running and invariant comparisons

Logarithmic counterterms introduce a renormalization scale $\mu$. Under $\mu$ evolution, finite local coefficients and multi-trace couplings can run. A meaningful comparison between schemes transforms both the coupling and the operator definition. Useful invariants include pole locations, spectral densities away from contact support, properly normalized conserved charges, and differences of generating functionals in which the same local ambiguity cancels.

At large $N$, double-trace beta functions often close at leading order, but subleading connected correlators generate additional multi-trace structures. A truncation must state its order in $1/N$ and in derivatives. It is unsafe to promote a leading large-N flow to an exact finite-N statement.

Adversarial check: a false scheme equivalence

Start from $G_0(k)$ and compare:

$$G_{\mathrm{local}}(k)=G_0(k)+c_0+c_2k^2, \qquad G_f(k)=\frac{G_0(k)}{1+fG_0(k)}.$$

The first differs only by contact terms. The second changes nonanalytic structure and can create, remove, or move poles. If either is labelled simply “a finite term,” that label conceals the physics. A proposed classification fails whenever it groups these two transformations together.

Practical decision rule

For every finite boundary term, ask in order:

1. Is it local in the fixed sources and compatible with the symmetries?
2. Does it preserve the same variational problem?
3. Does it leave separated-point observables and the spectrum unchanged?
4. Is a Legendre transform merely changing variables, or is it selecting another admissible quantization?
5. Which ensemble, scale, and large-N order are held fixed?

Only an affirmative answer to the first three licenses “scheme change.” A changed boundary condition, pole structure, or operator spectrum must be reported as a physical change.

The chapter overview contains the structure diagram and validity and failure diagram. They are embedded there once so that their shared chapter-level context is not repeated on every article.

For the chapter-wide comparison of assumptions, counterevidence, falsifiers, and claim ceilings, see the claim-domain table.

References

• Berkooz, Micha, Amit Sever, and Avishay Shomer. “Double-Trace Deformations, Boundary Conditions and Space-Time Singularities.” Journal of High Energy Physics 2002, 034 (2002). DOI; arXiv:hep-th/0112264.
• Klebanov, Igor R., and Edward Witten. “AdS/CFT Correspondence and Symmetry Breaking.” Nuclear Physics B 556, 89–114 (1999). DOI; arXiv:hep-th/9905104.
• Papadimitriou, Ioannis. “Multi-Trace Deformations in AdS/CFT: Exploring the Vacuum Structure of the Deformed CFT.” Journal of High Energy Physics 2007, 075 (2007). DOI; arXiv:hep-th/0703152.
• Witten, Edward. “Multi-Trace Operators, Boundary Conditions, and AdS/CFT Correspondence.” arXiv preprint (2001). arXiv:hep-th/0112258.