Gaussian Fixed Lines in RG Theory

• Gaussian Fixed Lines are continuous sets of renormalization-group fixed points connected to a Gaussian theory, characterized by exactly marginal directions and vanishing beta functions.
• The analysis involves local deformation theory, perturbative searches in two-dimensional four-fermion models, and functional RG flows that distinguish genuine fixed lines from mere RG trajectories.
• These constructions clarify the differences between exactly marginal deformations and superficially similar flows, offering insights into conformal manifolds and the structure of free-boson boundary sectors.

Gaussian fixed lines (GFLs) are continuous families of renormalization-group fixed points that pass through a Gaussian, or free, fixed point and are usually associated with exactly marginal directions. In the material summarized here, that definition is both central and a source of ambiguity, because closely related literatures also analyze local deformations of a single Gaussian point, RG trajectories connecting distinct Gaussian endpoints, and a distinguished $c=1$ Gaussian/free-boson theory with rich boundary sectors. These are not equivalent notions. The sharp distinction is that a genuine fixed line requires vanishing beta functions along a continuous locus of scale-invariant theories, whereas local operator algebra, perturbative candidate loci, or trajectories between isolated Gaussian fixed points do not by themselves establish such a structure (A et al., 2024).

1. Definitional framework and scope

In the strict RG sense, a Gaussian fixed line is a continuous set of fixed points connected to a Gaussian theory. Along such a line, the theory remains scale invariant at every point on the family, and the tangent directions are exactly marginal. This is the standard against which the other constructions discussed here must be judged.

The distinction matters because several nearby notions can look superficially similar. A perturbation may be classically marginal at a Gaussian point yet fail to remain marginal once nonlinear effects are included. A set of couplings may satisfy $\beta=0$ through one or two loops without defining an exact conformal manifold. A theory may also flow between two Gaussian fixed points without containing any continuous family of Gaussian fixed points. The summarized literature makes all of these distinctions explicit.

A useful taxonomy emerges. One class of results concerns the local deformation theory of a Gaussian fixed point, including operator products and nonlinear source flow. Another concerns perturbatively accessible candidate fixed lines in two-dimensional four-fermion theories. A third studies trajectories connecting isolated Gaussian fixed points associated with different kinetic operators. A fourth develops one $c=1$ Gaussian/free-boson CFT with explicit boundary conformal blocks and $\mathrm{SLE}_4$ geometry. Only some of these are properly about GFLs.

2. Local deformation theory around the four-dimensional scalar Gaussian point

Sonoda’s study of the Gaussian fixed point in $D=4$ is formulated in the exact renormalization group (ERG) language for a free massless real scalar with Wilson action

$$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$

which becomes, after rescaling to dimensionless variables,

$$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$

The analysis is restricted to scalar, $\phi\to-\phi$ even, relevant or marginal composite operators. In this setting the only three independent even scalar relevant/marginal operators are $O_2$, corresponding to $\phi^2$, $\beta=0$0, corresponding to $\beta=0$1, and $\beta=0$2, the number operator corresponding to $\beta=0$3, with momentum-space scaling dimensions $\beta=0$4, $\beta=0$5, and $\beta=0$6, respectively; $\beta=0$7 is not counted separately (Sonoda, 2020).

The central object is not a fixed line but the renormalized multiple-product algebra

$\beta=0$8

where the braces denote a properly renormalized local composite operator product. The need for counterterms in such products is the ERG manifestation of short-distance singularities. Sonoda packages the resulting local operator algebra into a source-dependent generating functional $\beta=0$9, where the couplings are allowed to be momentum-dependent sources $c=1$0. This generalizes constant couplings such as $c=1$1, $c=1$2, and $c=1$3 to local source fields, so the construction probes the full local source space near the Gaussian point rather than only constant deformations.

The main structural claim is that the short-distance singularities of multiple products determine the nonlinear RG scaling of the sources. In coordinate-space language, momentum dependence in $c=1$4 is equivalent to local couplings such as

$c=1$5

This makes the paper a study of local deformation theory and operator algebra around one Gaussian fixed point.

For the fixed-line question, the decisive result is negative. Although $c=1$6 is marginal at the linearized Gaussian point, it is not shown to be exactly marginal. For constant sources,

$c=1$7

the quartic beta function is

$c=1$8

With the paper’s IR-flow sign convention, the quartic coupling runs. The free Gaussian point is therefore isolated rather than exhibited as part of a fixed line. The paper studies the operator algebra near a single Gaussian fixed point and the perturbative construction of $c=1$9 theory from it; it does not construct a Gaussian fixed line.

3. Perturbative candidate fixed lines in two-dimensional four-fermion models

A direct search for Gaussian fixed lines appears in the study of renormalizable $\mathrm{SLE}_4$0-dimensional massless four-fermion models built from $\mathrm{SLE}_4$1 free fermions perturbed by current-current interactions. For Dirac fermions the Lagrangian is

$\mathrm{SLE}_4$2

with dimensionless couplings $\mathrm{SLE}_4$3, so the interaction is classically marginal in two dimensions. The Gaussian point is $\mathrm{SLE}_4$4. The current-current form is natural because two-dimensional Fierz identities allow generic four-fermion interactions to be rewritten in that basis. For Dirac fermions $\mathrm{SLE}_4$5, while for Majorana fermions one additionally has

$\mathrm{SLE}_4$6

The notion of fixed line used here is explicitly perturbative: the search is for continuous families of couplings solving the vanishing beta-function conditions near the free theory, through two loops (A et al., 2024).

The renormalization is performed in dimensional regularization around $\mathrm{SLE}_4$7, with

$\mathrm{SLE}_4$8

A key simplification is that the field-strength renormalization vanishes through two loops,

$\mathrm{SLE}_4$9

At one loop,

$D=4$0

and at two loops

$D=4$1

In the scheme used, vanishing of the one-loop beta functions implies vanishing of the two-loop beta functions as well, because the two-loop terms arise by replacing one vertex by the one-loop counterterm. The fixed-line search therefore reduces to the quadratic algebraic constraints

$D=4$2

The standard exact benchmark is the Abelian Thirring family,

$D=4$3

For one Dirac fermion,

$D=4$4

and bosonization gives equivalence to a compact free boson with

$D=4$5

In the summarized material this is the canonical Gaussian fixed line: a bona fide family of CFTs continuously connected to free fermions.

Beyond that known example, the paper finds a broader class of perturbative solutions. A simple sufficient ansatz is

$D=4$6

which is described as “an obvious solution” of the one-loop constraints, although numerical and low-$D=4$7 analytic work indicate more general families. For Dirac fermions the total number of independent couplings is

$D=4$8

and the authors report solution spaces larger than the Abelian Thirring subfamily or the simple product-like ansatz. For $D=4$9 they give an explicit analytic solution with minimal zero couplings that leaves $$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$0 independent couplings out of the original $$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$1. For $$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$2, numerical evidence suggests a family with effective number of free parameters

$$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$3

The status of these loci is deliberately cautious. They are candidate GFLs in the weak perturbative sense: classically marginal couplings, $$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$4 at one loop, and in the chosen scheme also $$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$5 at two loops. They are not proven exactly marginal deformations, and the authors explicitly note that three-loop terms are scheme independent in this situation and may be nonzero. They also do not provide a stability-matrix analysis, a discussion of redundant operators, a conformal perturbation theory check of exact marginality, or the operator-theoretic data needed to prove a conformal manifold. In that precise sense, the new families are perturbative candidate fixed lines rather than established ones.

4. RG trajectories between isolated Gaussian fixed points

A different but related use of Gaussian language appears in functional-RG studies of scalar theories with many free fixed points associated with quadratic kinetic operators of the form

$$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$6

Here the Gaussian fixed points are denoted GFP$$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$7, and in $$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$8 dimensions the canonical field dimension is

$$S_\Lambda[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p/\Lambda)}\,\phi(p),$$9

In $$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$0, GFP$$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$1 has canonical field dimension $$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$2, GFP$$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$3 has canonical field dimension $$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$4, and GFP$$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$5 is the “trivial” fixed point with field dimension $$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$6. The central claim is that RG trajectories can interpolate between such isolated Gaussian endpoints, with the anomalous dimension changing continuously so that the field has the correct scaling dimension at each endpoint (Buccio et al., 2022).

The main example uses the truncated effective action

$$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$7

Near GFP$$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$8, the natural chart $$S[\phi] = -\frac12 \int_p \phi(-p)\,\frac{p^2}{K(p)}\,\phi(p).$$9 uses dimensionless couplings

$\phi\to-\phi$0

with anomalous dimension

$\phi\to-\phi$1

The flow equations imply

$\phi\to-\phi$2

Approaching the UV Gaussian endpoint GFP$\phi\to-\phi$3 from this chart requires $\phi\to-\phi$4, hence $\phi\to-\phi$5. This produces the continuous change in effective scaling dimension from $\phi\to-\phi$6 at GFP$\phi\to-\phi$7 to $\phi\to-\phi$8 at GFP$\phi\to-\phi$9.

The geometric point is that different Gaussian fixed points lie naturally in different coordinate charts on theory space. In the chart $O_2$0, where the four-derivative term defines the free theory, the dimensionless variables are

$O_2$1

and the coordinate transformation between the two charts is

$O_2$2

This shows directly that a diverging coupling in one chart may correspond to a vanishing coupling in another. The asymptotically free UV endpoint seen as “strong coupling” in $O_2$3 is free in the coordinates adapted to GFP$O_2$4.

This framework is not a GFL in the standard sense. The Gaussian theories are isolated fixed points, not a continuous family of fixed points with vanishing beta functions everywhere along a locus. The continuity lies in the RG trajectory and in the anomalous dimension, not in a manifold of Gaussian fixed points. The paper also emphasizes significant limitations: higher-derivative free theories with $O_2$5 contain ghosts at perturbative level, the asymptotically free trajectories involve negative couplings, omitted operators are generated in the middle of the flow, and some quantitative features are regulator dependent. The result is therefore an analysis of flows between isolated Gaussian fixed points, not a realization of a Gaussian fixed line.

5. The $O_2$6 Gaussian/free-boson point and boundary sectors

Another adjacent construction arises from the two-dimensional Gaussian free field (GFF), explicitly interpreted as the massless free boson with central charge

$O_2$7

The setting is a simply connected domain with piecewise constant Dirichlet boundary data. The paper studies level lines of this GFF and shows that their crossing probabilities are determined by conformal blocks of primary fields that are degenerate at each insertion. The basic boundary data are encoded by generalized Dyck paths $O_2$8, with jumps of size $O_2$9 where

$\phi^2$0

and the fused conformal block is

$\phi^2$1

The pure partition functions $\phi^2$2 are finite linear combinations of these blocks, and the crossing probabilities are

$\phi^2$3

The same functions also describe fused multiple $\phi^2$4 sectors and solve BPZ PDEs of arbitrary order (Karrila et al., 20 Jun 2026).

The inserted fields are

$\phi^2$5

so the relevant observables are explicit $\phi^2$6 degenerate conformal blocks. The functions $\phi^2$7 are positive, linearly independent, and Möbius covariant. In the GFF/$\phi^2$8 coupling, a single curve conditioned on connectivity $\phi^2$9 has Loewner drift

$\beta=0$00

The paper also proves scaling-limit convergence from metric graph GFF first-passage-set geometry to the same continuum formulas.

Its relevance to GFLs is structural rather than exhaustive. The theory is a distinguished $\beta=0$01 Gaussian/free-boson point with a rich family of admissible boundary conditions and fused degenerate insertions. It does not study a varying compactification radius, a line of inequivalent $\beta=0$02 CFTs, or marginal deformations generating a Gaussian fixed line in the standard condensed-matter or conformal-field-theory sense. The “family” here is a family of boundary sectors inside one fixed normalization of the noncompact GFF/free-boson theory.

6. Established results, candidate structures, and common misunderstandings

The literature summarized here separates naturally into four statuses. First, there is the exact benchmark: the Abelian Thirring model, which is known nonperturbatively to define a CFT continuously connected to free fermions and thus serves as the canonical Gaussian fixed line through a free point (A et al., 2024). Second, there are perturbative candidate GFLs in generalized two-dimensional four-fermion models, for which $\beta=0$03 is shown through two loops but exact marginality is not established. Third, there is local deformation theory at a single Gaussian fixed point, exemplified by Sonoda’s $\beta=0$04 scalar analysis, where the main result is a source-dependent operator algebra and a nonzero nonlinear beta function for the quartic direction rather than a fixed line (Sonoda, 2020). Fourth, there are conceptually adjacent constructions—flows between isolated Gaussian endpoints and explicit $\beta=0$05 Gaussian/free-boson boundary sectors—that illuminate the geometry of theory space or the structure of one Gaussian theory without producing a GFL in the strict sense (Buccio et al., 2022, Karrila et al., 20 Jun 2026).

A recurrent misconception is that classical marginality is enough. It is not. In the four-dimensional scalar example, the $\beta=0$06 operator is marginal at the linearized Gaussian point but ceases to be marginal once nonlinear effects are included. Another misconception is that perturbative vanishing of beta functions automatically proves a conformal manifold. The two-dimensional four-fermion analysis explicitly avoids that claim: vanishing through two loops is only a necessary condition, and higher-loop effects may obstruct exact marginality. A third misconception is that any continuous interpolation between Gaussian theories defines a fixed line. The functional-RG flows between GFP$\beta=0$07 show the opposite: trajectories may connect isolated Gaussian endpoints without any continuous family of fixed points.

This suggests a practical criterion. To identify a genuine GFL, one needs more than a Gaussian origin and a marginal interaction. One needs a continuous locus of couplings with vanishing beta functions, together with enough operator-theoretic control to distinguish tangent exactly marginal directions from transverse perturbations. The summarized papers show that this criterion is stringent. They also show why GFLs are important: they sit at the intersection of local operator algebra, conformal manifolds, theory-space geometry, and exactly solvable free-boson structures, but each of those topics captures only one aspect of the full concept.