betaFeedback

WorksPhysics and Astronomy

Yang-Lee edge singularity at O(∂4)\mathcal{O}(\partial^4) from the functional renormalization group: revised analysis

Abstract

We determine the critical exponents of the Yang-Lee edge singularity as continuous functions of dimension using the functional renormalization group, retaining all operators at second and fourth order in the derivative expansion. At fourth order, we obtain η=−0.153(7)η=-0.153(7), −0.352(9)-0.352(9) and −0.575(9)-0.575(9) in d=5d=5, 44 and 33, respectively, in agreement with the six-loop εε expansion. In three dimensions, these results give σ=0.076σ=0.076 and νc=0.359ν_c=0.359. We locate the zero of the edge exponent at dσ=2.65(1)d_σ=2.65(1), also consistent with the six-loop estimate. Continuing the fixed point through dσd_σ, we obtain η=−0.795(21)η=-0.795(21) and ω=1.90(6)ω=1.90(6) in d=2d=2, compared with the exact values −4/5-4/5 and 22. For the complex Yang-Lee fixed point, convergence of the field expansion depends crucially on its reference point. Expansions about ϕ=0ϕ=0 fail to converge with the smooth regulator used here, whereas expansions about the running imaginary field value at which the mass vanishes converge down to d=1.3d=1.3. This representation preserves the exact relation between the thermal eigenvalue and the field dimension, and hence the hyperscaling identity θh+θt=dθ_h+θ_t=d, at every field-truncation order. We trace the large violations reported in earlier works to the omission of the expansion-point motion when linearizing the flow, rather than to the truncation itself.

Introduction

The Yang–Lee edge (YLE) arises from the zeros of a ferromagnetic partition function in a complex magnetic field. In the thermodynamic limit, these zeros accumulate along lines [1, 2], which, above the critical temperature, terminate at h=±ihc(T)h=\pm ih_c(T). The density of zeros becomes singular at these endpoints, and each endpoint is itself a critical point [3, 4]. Its long-distance behavior is described by a scalar field theory with an imaginary cubic interaction,

L=12(∂ϕ)2+igϕ3.(1)\mathcal{L}=\frac{1}{2}(\partial\phi)^2+ig\phi^3. \tag*{(1)}

The upper critical dimension of this theory is six. Below six dimensions, the field has a negative anomalous dimension η\eta. The equation of motion relates ϕ2\phi^2 to ϕ\phi through a shift of the field, leaving η\eta as the only independent critical exponent [4, 5]. Interest in this universality class has recently been renewed by its role in studies of the QCD phase diagram: the location of the edge in the complex chemical-potential plane constrains the critical end point [6, 7, 8, 9, 10, 11]. For an introduction to this connection, see Ref. [12].

The edge exponents have been studied using the ϵ=6−d\epsilon=6-d expansion, now available to six loops [13, 14, 15, 16, 17, 18, 19], the exact solution in d=2d=2 [5], high-temperature series [20, 21], the conformal bootstrap [22, 23], lattice simulations [24, 25], and fuzzy-sphere regularizations [26, 27, 28]. The functional renormalization group (FRG) [29, 30, 31, 32, 33, 34] offers a complementary non-perturbative approach that can be applied directly in both integer and non-integer dimensions [35, 36, 37, 38, 39, 40, 41]. Examples include the Potts model, whose cubic interaction is closely related to the Yang–Lee case [42], and long-range models, whose exponents follow from short-range ones at an effective fractional dimension [43]. At fourth and sixth order in the derivative expansion, it yields the exponents of the three-dimensional Ising and O(N)O(N) universality classes in agreement with the conformal bootstrap, with error estimates that are under control [44, 45]. Recent developments include the use of conformal invariance to constrain truncations and the choice of regulator [46, 47]; field reparametrisations that reduce the flow to its essential couplings or expand it about the physical ground state [48, 49, 50]; and real-time and spectral formulations, which give access to dynamic critical exponents [51, 52, 53] and to the critical scaling of spectral functions [54, 55, 56].

Few previous FRG studies focused on the edge exponents.1 An, Mesterházy and Stepanov (AMS) [61] used a derivative expansion with the exponential regulator Rk(q)=αZkq2/(eq2/k2−1)R_k(q)=\alpha Z_kq^2/(e^{q^2/k^2}-1) at α=1\alpha=1. They expanded the coefficient functions about the imaginary field value at which the mass vanishes. This expansion point runs with the FRG scale kk. Their truncation included one of the three independent four-derivative operators. They reported a violation of an exact hyperscaling relation between two eigenvalues of the linearized flow: 15% in d=5d=5 and 60–70% in d=4d=4. Zambelli and Zanusso (ZZ) [62] worked at second order in the derivative expansion with the Litim regulator Rk(q)=αZk(k2−q2)θ(k2−q2)R_k(q)=\alpha Z_k(k^2-q^2)\theta(k^2-q^2). They represented the field dependence by polynomials about ϕ=0\phi=0. These expansions converged slowly, and no fixed point was found below d=4d=4. Benedetti, Eustachon and Zanusso (BEZ) [63] solved the LPA and LPA’ with the Litim regulator for a purely imaginary potential, by numerical shooting rather than polynomial expansion. Within the LPA′, they followed the fixed point through the dimension d0≃2.725d_0 \simeq2.725 at which Δϕ\Delta_\phi vanishes, down to d=2d=2. They obtained Δϕ=0.167\Delta_\phi=0.167, i.e. η=−0.667\eta=-0.667, in d=3d=3. Most recently, Beretta, Sánchez-Villalobos and Codello [64] applied functional dimensional regularization (FDR) [65], a new functional approach, at second order in the derivative expansion; their values of η\eta are close to our O(∂2)O(\partial^2) results (Sec. VII).

These studies leave two issues unresolved: the origin of the hyperscaling violation reported by AMS and the convergence of the polynomial expansion at lower dimensions. We revisit both issues here at O(∂4)O(\partial^4). We also determine η\eta and σ\sigma at fourth order in the derivative expansion for 1≤d≤61 \le d \le6, and locate the dimension dσd_\sigma at which the edge exponent σ\sigma changes sign,

σ(dσ)=0.(2)\sigma(d_\sigma)=0. \tag*{(2)}

For the YLE complex-valued fixed point, the convergence of the field expansion depends crucially on the choice of expansion point. With a suitable choice, we obtain convergent expansions down to d≈1.3d \approx1.3, overcoming the convergence problem encountered by ZZ. This allowed us to find dσ=2.65(1)d_\sigma=2.65(1). This result agrees with the six-loop ϵ\epsilon expansion [19, 66].

The rest of the paper is organized as follows. Section II summarizes the exact relations among the exponents. We introduce the FRG and the derivative expansion in Sec. III, derive the flow equations in § 1.3, and describe the representation of the field dependence in § 1.4. Tests of the method are presented in Sec. VI, followed by the results in Sec. VII and our conclusions in Sec. VIII. Technical details are given in the appendices.

Exponents and exact relations

We denote the scaling dimension of the field by Δϕ=(d−2+η)/2\Delta_\phi=(d-2+\eta)/2. The Yang–Lee fixed point has two relevant directions. The magnetic field hh, which couples linearly to ϕ\phi, has eigenvalue/scaling dimension

θh=d−Δϕ=d+2−η2.(3)\theta_h=d-\Delta_\phi=\frac{d+2-\eta}{2}. \tag*{(3)}

The second, “thermal” direction corresponds to the ϕ2\phi^2 deformation of the cubic theory. The equation of motion, □ϕ∝iϕ2\Box\phi\propto i\phi^2, relates this deformation to a constant shift of the field, ϕ→ϕ+c\phi\to\phi+c. Its eigenvalue is therefore the scaling dimension of cc,

θt=Δϕ=d−2+η2,(4)\theta_t=\Delta_\phi=\frac{d-2+\eta}{2}, \tag*{(4)}

so that θh+θt=d\theta_h+\theta_t=d [4, 5]. The edge exponent and the correlation-length exponent follow as

σ=θtθh=d−2+ηd+2−η,νc=1θh=2d+2−η.(5)\sigma=\frac{\theta_t}{\theta_h}=\frac{d-2+\eta}{d+2-\eta},\qquad\nu_c=\frac{1}{\theta_h}=\frac{2}{d+2-\eta}. \tag*{(5)}

Near six dimensions, the anomalous dimension is η=−ϵ/9−43ϵ2/729+…\eta=-\epsilon/9-43\epsilon^2/729+\ldots [13, 14, 18]. In two dimensions, the exact solution gives η=−4/5\eta=-4/5, Δϕ=−2/5\Delta_\phi=-2/5 and σ=−1/6\sigma=-1/6 [5], and in one dimension σ=−1/2\sigma=-1/2, i.e. η=−1\eta=-1 and Δϕ=−1\Delta_\phi=-1 [4]. Since σ\sigma is positive in d=3d=3 and negative in d=2d=2, it must vanish at an intermediate dimension dσ∈(2,3)d_\sigma\in(2,3). At this dimension, Δϕ=0\Delta_\phi=0 and η=2−dσ\eta=2-d_\sigma.

FRG and the derivative expansion

In the FRG calculation, the two eigenvalues are extracted from the stability matrix of the flow linearized about the fixed point. The relations θh=d−Δϕ\theta_h=d-\Delta_\phi and θt=Δϕ\theta_t=\Delta_\phi are exact properties of the full flow, but they have different sensitivities to approximations of the field dependence.

The first relation follows from the fact that the flow depends on the potential U(ϕ)U(\phi) only through U′′U'' and higher derivatives (see Sec. III). The coefficient of the linear term therefore decouples, giving θh=d−Δϕ\theta_h=d-\Delta_\phi exactly in any truncation.

The second relation follows by considering an infinitesimal field shift in every fixed-point coefficient function, F(ϕ)→F(ϕ+c)F(\phi)\to F(\phi+c). The resulting perturbation is proportional to F′F' and has eigenvalue θt=Δϕ\theta_t=\Delta_\phi. Unlike the magnetic eigenvalue relation, this identity can be violated when the field dependence is approximated, for example by a truncated Taylor expansion. We use the relative discrepancy

ϵscale=∣θtstab−Δϕ∣Δϕ,(6)\epsilon_{\rm scale}=\frac{\left|\theta_t^{\rm stab}-\Delta_\phi\right|}{\Delta_\phi}, \tag*{(6)}

where θtstab\theta_t^{\rm stab} is the stability-matrix eigenvalue, to quantify such violations. Comparing this eigenvalue with Δϕ\Delta_\phi obtained from η\eta provides a direct check on the treatment of the field dependence.

III. FRG AND THE DERIVATIVE EXPANSION

The FRG describes the evolution of the scale-dependent effective action Γk[ϕ]\Gamma_k[\phi]. Fluctuations with momenta q≳kq\gtrsim k are integrated out, while those with q≲kq\lesssim k are suppressed by a regulator term 12∫qRkϕqϕ−q\frac{1}{2}\int_q R_k\phi_q\phi_{-q}, acting as a mass of order k2k^2. The flow interpolates between the microscopic action at k=Λk=\Lambda and the full effective action as k→0k\to0. It obeys the exact Wetterich equation [29],

∂tΓk=12Tr⁡[∂tRk(Γk(2)+Rk)−1],t=ln⁡(k/Λ),(7)\partial_t\Gamma_k=\frac{1}{2}\operatorname{Tr}\left[\partial_t R_k\left(\Gamma_k^{(2)}+R_k\right)^{-1}\right],\qquad t=\ln(k/\Lambda), \tag*{(7)}

where Γk(2)\Gamma_k^{(2)} denotes the second functional derivative. Although this equation has a one-loop form, it contains the full, field-dependent propagator and vertices. In dimensionless variables, a critical point corresponds to a fixed point of the flow.

We solve Eq. (7) by performing a derivative expansion of Γk\Gamma_k. At leading order, the local potential approximation (LPA) retains a running potential U(ϕ)U(\phi) and a fixed kinetic term. At second order, O(∂2)O(\partial^2), the kinetic coefficient is promoted to a field-dependent wave-function renormalization Z(ϕ)Z(\phi). At fourth order, the ansatz is

Γk=∫ddx{U(ϕ)+12Z(ϕ)(∂ϕ)2+12W1(ϕ)(□ϕ)2+12W2(ϕ)(∂ϕ)2□ϕ+12W3(ϕ)[(∂ϕ)2]2}.(8)\Gamma_k=\int d^d x\left\{U(\phi)+\frac{1}{2}Z(\phi)(\partial\phi)^2+\frac{1}{2}W_1(\phi)(\Box\phi)^2+\frac{1}{2}W_2(\phi)(\partial\phi)^2\Box\phi+\frac{1}{2}W_3(\phi)\left[(\partial\phi)^2\right]^2\right\}. \tag*{(8)}

Up to total derivatives, this ansatz contains all independent operators with at most four derivatives. In the Ising model, Z2Z_2 symmetry requires W2W_2 to be odd in ϕ\phi. At the Yang–Lee edge, this symmetry is explicitly broken, and we retain all three functions W1,2,3W_{1,2,3}. To distinguish the complete fourth-order expansion from the truncation used by AMS, which retains only W1W_1, we denote them by O(∂4)O(\partial^4) and O(∂4)W1O(\partial^4)_{W_1}, respectively.

For the regulator, we use Rk(q)=Z0k2r(q2/k2)R_k(q)=Z_0 k^2 r(q^2/k^2) with the smooth exponential shape

r(y)=αyey−1,(9)r(y)=\alpha\frac{y}{e^y-1}, \tag*{(9)}

as in Refs. [44, 61, 67]. We write Z(ϕ)=Z0z(ϕ)Z(\phi)=Z_0z(\phi) and impose z(ϕ0)=1z(\phi_0)=1 at a reference field ϕ0\phi_0, whose choice is specified below. Varying the amplitude α\alpha allows us to assess the residual regulator dependence. This dependence would be absent in an exact solution of the Wetterich equation, but remains at finite truncation order and provides a useful diagnostic of the truncation.

The Litim regulator [68, 69] is not sufficiently smooth for the derivative expansion beyond O(∂2)O(\partial^2), since the O(∂4)O(\partial^4) flows require higher momentum derivatives of RkR_k. One may instead introduce terms that are non-analytic in momentum, but the resulting proliferation of coefficient functions makes this regulator impractical for the higher-order calculations considered here.

We introduce dimensionless fields and coefficient functions by rescaling with appropriate powers of kk and Z0Z_0, and define the anomalous dimension as η=−∂tln⁡Z0\eta=-\partial_t\ln Z_0. For a coefficient function ff multiplying an operator with 2s2s derivatives and nn explicit fields, the flow takes the form

∂tf=(−d+2s+n2(d−2+η))f+Δϕϕf′+∫0∞dy yd/2−1If(y),(10)\partial_t f=\left(-d+2s+\frac{n}{2}(d-2+\eta)\right)f+\Delta_\phi\phi f' +\int_0^\infty dy\,y^{d/2-1}I_f(y), \tag*{(10)}

where y=q2/k2y=q^2/k^2. The first two terms describe the scaling of the operator and the field argument. The integral contains the loop contribution, whose integrand depends on the coefficient functions, their field derivatives, the regulator, dd and η\eta. A fixed point is a set of functions for which all flows ∂tf\partial_t f vanish.

Although the Yang–Lee fixed point is complex, it is invariant under ϕ→−ϕ\phi\to-\phi combined with complex conjugation. The coefficient functions therefore satisfy F(−ϕ)∗=±F(ϕ)F(-\phi)^*=\pm F(\phi), with the minus sign applying to W2W_2 because its operator is odd under field reflection. For an expansion about a point on the imaginary axis, this symmetry constrains successive Taylor coefficients to alternate between real and purely imaginary values. Factoring out the corresponding powers of ii turns the fixed-point equations into a real system.

Deriving the flow equations

We derive the flows of UU, ZZ and W1,2,3W_{1,2,3} by matching the momentum dependence of the vertex functions at a constant background field ϕ\phi. Here Γ(n)\Gamma^{(n)} denotes the nn-th functional derivative of Γk\Gamma_k in momentum space. The ansatz (8) expresses each vertex as a polynomial in the external momenta p1,…,pnp_1,\ldots,p_n, with coefficients determined by the coefficient functions and their field derivatives. For the two-point function, for example,

Γ(2)(p)=U′′+Zp2+W1p4.(11)\Gamma^{(2)}(p)=U''+Zp^2+W_1p^4. \tag*{(11)}

Taking nn functional derivatives of the Wetterich equation (7) gives a second expression for ∂tΓ(n)\partial_t\Gamma^{(n)}: a sum of one-loop diagrams containing the propagator (Γ(2)+Rk)−1(\Gamma^{(2)}+R_k)^{-1} and the vertices Γ(3),…,Γ(n+2)\Gamma^{(3)},\ldots,\Gamma^{(n+2)} obtained from the ansatz.

We expand these loop expressions to fourth order in the external momenta and match the coefficients of the resulting momentum structures. The flows are extracted as follows:

  • ∂tU\partial_t U from the Wetterich equation evaluated at a constant field;

  • ∂tZ\partial_t Z and ∂tW1\partial_t W_1 from the p2p^2 and p4p^4 coefficients of the two-point flow;

  • ∂tW2\partial_t W_2 and ∂tW3\partial_t W_3 from the three- and four-point flows, respectively.

We perform the angular averages over the loop momentum analytically for arbitrary dd. Each flow then reduces to a one-dimensional integral over y=q2/k2y=q^2/k^2, of the form given in (10). These steps are implemented in Mathematica for a general operator basis, as described in Appendix A.

Overdetermined projections

Matching the momentum structures yields more equations than unknown flows. For the three-point function Γ(3)(p1,p2,p3)\Gamma^{(3)}(p_1,p_2,p_3), with p1+p2+p3=0p_1+p_2+p_3=0, the expansion through fourth order contains 10 momentum structures. Their coefficients involve only four unknown flows: ∂tW2\partial_t W_2, ∂tW1′\partial_t W_1', ∂tZ′\partial_t Z' and ∂tU′′′\partial_t U'''. The four-point function similarly yields 28 equations for five unknowns.

These additional equations provide consistency checks. Through fourth order in the external momenta, the loop expansion has the momentum dependence of a local functional that can be expressed in the basis (8). The coefficient equations must therefore be mutually consistent. In addition, the flows of Z′Z', W1′W'_1 and the other field derivatives extracted from higher-point vertices must agree with the field derivatives of the flows obtained from lower-point vertices.

We verify that the residuals of all redundant equations vanish identically and that the field-derivative relations hold for all yy. These checks test the vertex algebra, the diagrammatic combinatorics and the angular averages before numerical integration. The resulting integrands are exported as Julia code, which is used throughout the numerical calculations.

Representing the field dependence

The fixed-point equations form a set of coupled ordinary differential equations in the field for UU, ZZ and WiW_i. We solve them by expanding each function about a reference point ϕ=iχ\phi=i\chi on the imaginary axis,

F(ϕ)=∑n=0NFfnn!(ϕ−iχ)n,(12)F(\phi)=\sum_{n=0}^{N_F}\frac{f_n}{n!}(\phi-i\chi)^n, \tag*{(12)}

with NU=NZ+3N_U=N_Z+3, NW1=NW3=NZ−2N_{W_1}=N_{W_3}=N_Z-2 and NW2=NZ−1N_{W_2}=N_Z-1. These relative orders follow from the structure of the flow equations (Appendix § 1.6.2). We solve the resulting equations for the Taylor coefficients by Newton’s method, starting from the perturbative fixed point near d=6d=6 and continuing in small steps in dd towards d=1d=1. The stability matrix is obtained by automatic differentiation of the same code, and its eigenvalues are then computed. We check convergence in the field expansion by increasing NZN_Z.

Convergence depends strongly on the choice of χ\chi. The fixed-point functions have singularities in the complex field plane associated with zeros of the regulated inverse propagator, U′′+Zq2+⋯+RkU''+Zq^2+\cdots+R_k. Their distance from the expansion point limits the radius of convergence of Eq. (12). We compare three choices of reference point.

  1. Expansion about ϕ=0\phi=0. The origin is a natural choice for a PTPT-symmetric problem and was used in Ref. [62]. With the exponential regulator, however, this expansion does not converge. At d=5.5d=5.5, the anomalous dimension oscillates with period three as NZN_Z increases (Figure 7), indicating singularities at angles of approximately 120∘120^\circ around the origin. We therefore do not pursue this choice further.

Figure 7. Convergence of η\eta with the field order at d=5.5d = 5.5, O(∂2)O(\partial^2) and α=1\alpha= 1, for expansions about ϕ=0\phi= 0 and ϕ=−2i\phi= -2i. The upper panel shows η\eta; the lower panel shows the violation ϵscale\epsilon_{\mathrm{scale}} of the exact relation θt=Δϕ\theta_t = \Delta_\phi. The period-three oscillation for the expansion about ϕ=0\phi= 0 arises from complex singularities of the fixed-point functions.

  1. Fixed imaginary point. Choosing χ=χ0<0\chi=\chi_0<0 moves the expansion point away from the nearest singularity on the positive imaginary axis and substantially improves convergence. At d=5.5d=5.5, η\eta is stable to seven digits for NZ≥10N_Z\geq10 (Figure 7). As we lower dd, we choose χ0(d)=−2−max⁡(5.5−d,0)\chi_0(d)=-2-\max(5.5-d,0) and impose z(iχ0)=1z(i\chi_0)=1. The expansion point is thus fixed along each RG flow, but varies between calculations at different dimensions.

Where it converges, this representation gives the fastest convergence, with ϵscale<10−2\epsilon_{\mathrm{scale}}<10^{-2} (Figure 1). Below d≈3.2d\approx3.2, a pair of off-axis singularities approaches the imaginary axis and this choice of expansion point ceases to work. A different choice of χ0\chi_0 might extend its range, but the running-point representation described next avoids the need for such an adjustment.

Figure 1. Violation ϵscale\epsilon_{\mathrm{scale}} of the exact relation θt=Δϕ\theta_t = \Delta_\phi for expansions about fixed imaginary points. Each curve ends where the field expansion fails. For the running massless point, the violation vanishes identically, ϵscale≡0\epsilon_{\mathrm{scale}} \equiv0. Crosses show the violations reported in Ref. [61].

  1. Running massless point. Following Refs. [58, 59, 61], we expand about the point on the imaginary axis where the mass vanishes and normalize zz at the same point:

U′′(iχ)=0,z(iχ)=1.(13)U''(i\chi)=0,\qquad z(i\chi)=1. \tag*{(13)}

The position χ\chi now runs with kk. The flows of the Taylor coefficients at the massless point depend on it only through the combination ψ=i(∂tχ+Δϕχ)\psi=i(\partial_t\chi+\Delta_\phi\chi), see Appendix § 1.6.3. The two conditions (13) determine ψ\psi and η\eta. We therefore solve for the Taylor coefficients, η\eta and ψ\psi; the position itself can be found through ∂tχ=−Δϕχ−iψ\partial_t\chi=-\Delta_\phi\chi-i\psi, however, it does not feed back into the other flows. This choice of the expansion point has two useful properties.

  • The translation relation is preserved exactly. A translation of the fixed-point solution shifts the massless point but leaves the Taylor coefficients evaluated there unchanged. The corresponding perturbation is therefore the unit vector in the χ\chi direction. It is an eigenvector of the stability matrix with eigenvalue Δϕ\Delta_\phi at

every truncation order. Because χ\chi does not feed back into the other flows, this eigenvalue is exact, and the remaining eigenvalues are those of the stability matrix of the Taylor coefficients alone. The explicit χ\chi dependence of the dilatation term ΔϕϕF′\Delta_\phi\phi F' is cancelled by the motion of the expansion point; Appendix § 1.6.3 gives the explicit derivation. Consequently, ϵscale=0\epsilon_{\mathrm{scale}}=0 identically in this representation.

  • *The massless point recedes as Δϕ→0.∗At\Delta_\phi\to0.* AtU''(i\chi)=0$, only the dilatation term iΔϕχU′′′(iχ)i\Delta_\phi\chi U'''(i\chi) and the loop contribution remain in the fixed-point equation for U′′U'' (see Eq. (10)). Denoting the loop contribution by L2L_2, we obtain

iΔϕχU′′′(iχ)=−L2.i\Delta_\phi\chi U'''(i\chi)=-L_2.

Both U′′′U''' and L2L_2 remain finite as dd varies, so the product χΔϕ\chi\Delta_\phi is finite and χ∝1/Δϕ\chi\propto1/\Delta_\phi. As dd approaches dσd_\sigma and Δϕ\Delta_\phi vanishes, the massless point moves toward −i∞-i\infty, away from the singularities.

The drawback is slower convergence in NZN_Z at O(∂4)O(\partial^4) near six dimensions.

We choose the expansion point according to its convergence. At O(∂2)O(\partial^2), we use the running massless point throughout, with convergence in η\eta better than 5×10−45\times10^{-4} as NZN_Z is increased for d≥dσd\geq d_\sigma. At O(∂4)O(\partial^4), we use fixed imaginary points for d≥3.2d\geq3.2, where they give more accurate results, and the running massless point below this dimension, down to d=1d=1.

The two choices of the expansion point agree within 2×10−42\times10^{-4} for d≥4d\geq4 and within 3×10−33\times10^{-3} at d≃3.4d\simeq3.4. Because they impose the normalization of zz at different field values, we include their difference in the uncertainty estimate (Appendix E).

Tests: the three-dimensional Ising model, the six-dimensional limit and the AMS calculation

We test the calculation against three benchmarks: the three-dimensional Ising fixed point, the perturbative Yang–Lee result near six dimensions, and the results obtained with the AMS truncation.

Three-dimensional Ising universality. Our flow equations were derived without imposing a field symmetry. The Ising fixed point, for which Z2Z_2 symmetry requires UU, ZZ, W1W_1 and W3W_3 to be even and W2W_2 to be odd, therefore tests the derivation in a different symmetry setting. For this benchmark, we resolve the field dependence on a grid (Appendix D) and determine the stationary points of the exponents as functions of α\alpha, following the principle of minimal sensitivity (PMS) [70].

This workRef. [44]
ν\nuη\etaν\nuη\eta
LPA0.6506000.650590
O(∂2)O(\partial^2)0.627870.044260.627790.04500
O(∂4)O(\partial^4)0.629970.034570.630270.03454

Table I. Three-dimensional Ising exponents obtained with r(y)=αy/(ey−1)r(y)=\alpha y/(e^y-1) at their stationary points in α\alpha, compared with Ref. [44] for the same regulator family. For comparison, Ref. [67] gives ν=0.6281\nu=0.6281 and η=0.0443\eta=0.0443 at O(∂2)O(\partial^2), and ν=0.632\nu=0.632 and η=0.033\eta=0.033 at O(∂4)O(\partial^4).

Table I compares our results with those of Ref. [44], which used the same regulator family. The LPA values of ν\nu differ by 10−510^{-5}, and the O(∂4)O(\partial^4) values of η\eta by 3×10−53\times10^{-5}. The largest difference in the table is 7.4×10−47.4\times10^{-4}. These small differences are consistent with the use of different normalization points and PMS prescriptions.

Approach to six dimensions. As ϵ=6−d→0\epsilon=6-d\to0, the Yang–Lee fixed point approaches the perturbative fixed point of the cubic theory. A truncation that includes Z(ϕ)Z(\phi) must reproduce the leading result, η=−ϵ/9+O(ϵ2)\eta=-\epsilon/9+O(\epsilon^2). Fitting η(ϵ)\eta(\epsilon) over ϵ∈[0.005,0.04]\epsilon\in[0.005,0.04], we obtain the limiting ratios η/ϵ→−0.1110\eta/\epsilon\to-0.1110 at O(∂2)O(\partial^2) and −0.1112-0.1112 at O(∂4)O(\partial^4), both in agreement with −1/9-1/9.

Reproduction of the AMS calculation. AMS used the running massless point (13), the regulator amplitude α=1\alpha=1, and the polynomial truncation (nU,nZ,nW)=(7,5,0)(n_U,n_Z,n_W)=(7,5,0). Their notation specifies the highest power in UU by nUn_U, while nZn_Z and nWn_W count the terms retained in ZZ and W1W_1 [Eqs. (14)–(15) of Ref. [61]]. This corresponds to our O(∂2)O(\partial^2) truncation with (NU,NZ)=(7,4)(N_U,N_Z)=(7,4). With these settings, we obtain η=−0.1260\eta=-0.1260 in d=5d=5 and −0.3156-0.3156 in d=4d=4, reproducing the AMS values −0.126(6)-0.126(6) and −0.316(16)-0.316(16) [61].

For the same truncation, we find θtstab=Δϕ\theta_t^{\mathrm{stab}}=\Delta_\phi to machine precision, as required by the argument in § 1.4. The violations of 15% and 60–70% reported by AMS therefore do not arise from the truncation. They instead result from the treatment of the running expansion point when linearizing the flow, as discussed in Sec. IX.

This issue does not affect the exponents that AMS obtained from η\eta using Eq. (5). Their O(∂4)O(\partial^4) results are, however, affected by the omission of operators from the fourth-order truncation, which we examine in Sec. VIII B.

Results

Anomalous dimension and exponents

Figure 2(a) compares η(d)\eta(d) in the three truncations, while Figure 2(b) compares the complete O(∂4)O(\partial^4) result with earlier determinations. The corresponding edge and correlation-length exponents, σ\sigma and νc\nu_c, are shown in Figure 3. Table II lists the results at selected dimensions.

Figure 2. Anomalous dimension of the Yang–Lee edge as a function of dd at α=2\alpha= 2. (a) Comparison of the O(∂2)\mathcal{O}(\partial^2), complete O(∂4)\mathcal{O}(\partial^4) and O(∂4)W1\mathcal{O}(\partial^4)_{W_1} truncations. (b) Complete O(∂4)\mathcal{O}(\partial^4) result compared with the literature. Filled circles mark our values at d=2d = 2, 33, 44 and 55. Above dσd_\sigma, bands combine the field-truncation, regulator and normalization-point uncertainties, while below dσd_\sigma, only the first two. Solid lines use fixed imaginary expansion points; dashed lines use the running massless point (§ 1.4). The O(∂2)\mathcal{O}(\partial^2) curve uses the running massless point throughout and is dashed over the whole range. Where the field expansion is not converged (at O(∂4)\mathcal{O}(\partial^4) below d≃1.3d \simeq1.3, at O(∂2)\mathcal{O}(\partial^2) and O(∂4)W1\mathcal{O}(\partial^4)_{W_1} below d≃1.6d \simeq1.6), the lines are dotted and the hatched bands show the change from NZ=14N_Z = 14 to 1616 (1212 to 1414 at O(∂2)\mathcal{O}(\partial^2) and O(∂4)W1\mathcal{O}(\partial^4)_{W_1}). The dotted gray line in (b) is the [2/2] Padé approximant of the five-loop ϵ\epsilon expansion [18]. Other symbols show six-loop [19] and five-loop [18] ϵ\epsilon-expansion estimates, the exact results in d=2d = 2 and d=1d = 1 [4, 5], fuzzy-sphere results [27, 28], AMS [61], and ZZ [62]. All literature values are at integer dd; symbols are displaced horizontally for visibility.

ddη\eta [O(∂2)O(\partial^2)]η\eta [O(∂4)O(\partial^4)]σ\sigmaνc\nu_cη\eta [ϵ-exp./exact]
5.5−0.0588(16)−0.06855(29) ± 0.0030.45340.2643−
5−0.1324(27)−0.15307(52) ± 0.0070.39800.2796−0.152(1)
4.5−0.2214(35)−0.24868(58) ± 0.0090.33360.2964−
4−0.3244(38)−0.3523(19) ± 0.0090.25940.3148−0.355(6)
3.5−0.4388(52)−0.4603(34) ± 0.0070.17440.3356−
3−0.562(12)−0.5751(76) ± 0.0050.07620.3587−0.578(6)
2.5−0.688(17)−0.687(13) ± 0.001−0.03600.3856−
2−0.814(30)−0.795(20) ± 0.007−0.16570.4171−0.8 (exact)
1.5−0.93(6){sup}†`−0.893(23) ± 0.012−0.31700.4553−
1−1.07(8){sup}†`−0.97(4){sup}†` ± 0.03−0.4966{sup}†`0.5034{sup}†`−1 (exact)

Table II. Anomalous dimension η\eta at α=2\alpha= 2 at second and fourth order of the derivative expansion. The O(∂2)O(\partial^2) results use the running massless point. The O(∂4)O(\partial^4) results use fixed imaginary points for d≥3.2d \ge3.2 and the running massless point below this dimension (§ 1.4). Parenthetical errors combine the regulator uncertainty (half the spread over α∈[1,3]\alpha\in[1, 3], or the shift to the PMS point if larger), the field-order difference and the normalization-point dependence. For O(∂4)O(\partial^4), the second error, given after ±\pm, is the derivative-expansion estimate ∣ηO(∂4)−ηO(∂2)∣/3|\eta_{O(\partial^4)}-\eta_{O(\partial^2)}|/3 (Appendix E). The exponents σ\sigma and νc\nu_c follow from Eq. (5) using the O(∂4)O(\partial^4) value of η\eta. The final column gives the six-loop ϵ\epsilon-expansion estimate of Ref. [19]. Below dσd_\sigma (lower block), both truncations use the running massless point; the errors combine the regulator and field-order uncertainties, and the final column gives the exact values [4, 5]. Entries marked † are not converged in the field expansion (Sec. VII F); for these values of η\eta, the parenthetical error is the change from NZ=14N_Z = 14 to 16 (12 to 14 at O(∂2)O(\partial^2)).

Figure 3. Edge exponent σ\sigma and correlation-length exponent νc\nu_c, obtained from Eq. (5) using the complete O(∂4)\mathcal{O}(\partial^4) result in Figure 2(b). Filled circles mark our values of σ\sigma at d=2d=2, 33, 44 and 55; the error bar at d=2d=2 is the total uncertainty. As in Figure 2, the lines are solid where fixed imaginary expansion points are used (d≥3.2d \ge3.2) and dashed where the running massless point is used; they are dotted below d≃1.3d \simeq1.3, where the field expansion is not converged. Other symbols show results from the six-loop ϵ\epsilon expansion [19], conformal bootstrap [23], high-temperature series [21], and the exact results in d=2d=2 and d=1d=1 [4, 5], shown as stars for both σ\sigma and νc\nu_c. Literature values are at integer dd; symbols are displaced horizontally for visibility.

Throughout this section, we use the regulator amplitude α=2\alpha= 2; the reasons for this choice are discussed in Sec. VII C.

Going from O(∂2)O(\partial^2) to the complete O(∂4)O(\partial^4) truncation changes η\eta by about 16% in d=5d = 5, 9% in d=4d = 4 and 3% in d=3d = 3. Our fourth-order results are

η={−0.153(7),d=5,−0.352(9),d=4,−0.575(9),d=3.(14)\eta= \begin{cases} -0.153(7), & d = 5,\\ -0.352(9), & d = 4,\\ -0.575(9), & d = 3. \end{cases} \tag*{(14)}

The quoted uncertainties include the derivative-expansion estimate described in Appendix E. In d=3d = 3, these values give σ=0.076\sigma= 0.076 and νc=0.359\nu_c = 0.359.

The results agree with the six-loop ϵ\epsilon-expansion analysis of Ref. [19], which gives η=−0.152(1)\eta= -0.152(1), −0.355(6)-0.355(6) and −0.578(6)-0.578(6) in d=5d = 5, 4 and 3, respectively, and σ=0.078(9)\sigma= 0.078(9) in d=3d = 3. Our three-dimensional result also agrees with the fuzzy-sphere determination Δϕ=0.2151(8)\Delta_\phi= 0.2151(8), corresponding to η=−0.570(2)\eta= -0.570(2) [27], and with the high-temperature series estimate σ=0.077(2)\sigma= 0.077(2) [21]. The conformal-bootstrap estimate, σ=0.085(1)\sigma= 0.085(1) [23], has a larger central value.

At O(∂2)O(\partial^2), we find close agreement with ZZ [62], who used the Litim regulator. Our values, η=−0.1324(27)\eta= -0.1324(27) in d=5d = 5 and −0.3244(38)-0.3244(38) in d=4d = 4, compare with their −0.1344(1)-0.1344(1) and −0.325(3)-0.325(3), respectively. The FDR values at the same order, η=−0.1414\eta= -0.1414, −0.3276-0.3276 in d=5d = 5, 4 [64] agree with ours at O(∂2)O(\partial^2); their values at d=3d = 3 is −0.5451-0.5451 vs our −0.562(12)-0.562(12).

Effect of the omitted operators

Retaining only W1W_1, as in Ref. [61], gives η(5)=−0.163\eta(5) = -0.163 and η(4)=−0.361\eta(4) = -0.361. Measured relative to O(∂2)O(\partial^2), these shifts are 1.5 and 1.3 times those of the complete O(∂4)O(\partial^4) truncation in d=5d = 5 and 4, respectively. The combined contribution of W2(∂ϕ)2□ϕW_2(\partial\phi)^2\Box\phi and W3[(∂ϕ)2]2W_3[(\partial\phi)^2]^2 therefore opposes the shift produced by W1W_1 alone and is not negligible.

The pattern changes near d=3d = 3. There, O(∂4)W1O(\partial^4)_{W_1} gives η(3)=−0.571\eta(3) = -0.571, between the O(∂2)O(\partial^2) and complete O(∂4)O(\partial^4) values. The discrepancy caused by omitting W2W_2 and W3W_3 thus varies in both magnitude and sign with dimension.

Regulator dependence

At finite truncation order, the results depend on the regulator amplitude α\alpha in Eq. (9). The principle of minimal sensitivity (PMS) selects an amplitude αPMS\alpha_{\mathrm{PMS}} at which η\eta is stationary [70]. Figure 4 shows η(α)\eta(\alpha) in d=5d = 5, 4 and 3, calculated with the running massless point.

Figure 4. Regulator dependence of η\eta in d=5d=5, 44 and 33, calculated with the running massless point. At O(∂2)\mathcal{O}(\partial^2), NZ=10N_Z=10. At O(∂4)\mathcal{O}(\partial^4), solid lines show the highest field order computed: NZ=14N_Z=14 in d=5d=5 and 44, and NZ=12N_Z=12 in d=3d=3. Dashed lines show the preceding field order. Filled circles mark stationary points αPMS\alpha_{\mathrm{PMS}}; open circles mark α=2\alpha=2, used for the reported results. In d=3d=3, the two O(∂4)\mathcal{O}(\partial^4) curves separate for α≳3\alpha\gtrsim3, where the field expansion is not converged.

At O(∂2)O(\partial^2), each curve has a single extremum, at αPMS=2.7\alpha_{\mathrm{PMS}} = 2.7, 1.9 and 1.5 in d=5d = 5, 4 and 3, respectively. The corresponding values of η\eta, −0.1328-0.1328, −0.3245-0.3245 and −0.5655-0.5655, differ from the α=2\alpha= 2 results by at most 0.004.

At O(∂4)O(\partial^4), a reliable search for a stationary point requires convergence of the field expansion over a sufficiently wide range of α\alpha. With fixed imaginary expansion points, the variation of η\eta over α∈[1,3]\alpha\in[1,3] is only 1×10−31 \times10^{-3} in d=5d = 5 and 2.5×10−32.5 \times10^{-3} in d=4d = 4. Larger amplitudes are accessible only with the running point, for which convergence is slower. We therefore extended those calculations to NZ=14N_Z = 14 (Figure 4).

In d=5d = 5, the NZ=12N_Z = 12 and 14 results agree within 10−410^{-4} for α≳3\alpha\gtrsim3, and no stationary point is found over α∈[1.3,5]\alpha\in[1.3,5]. In d=4d = 4, they agree within 6×10−46 \times10^{-4} for α≳2.5\alpha\gtrsim2.5 and exhibit a shallow maximum at αPMS≃3.3\alpha_{\mathrm{PMS}} \simeq3.3–3.9. Its value, η=−0.3514\eta= -0.3514, lies 1.7×10−31.7 \times10^{-3} above the α=2\alpha= 2 result.

In d=3d = 3, the expansion is converged for α≲3\alpha\lesssim3, where the NZ=10N_Z = 10 and 12 results agree within 3×10−43 \times10^{-4}. Maxima appear at α≃3.4\alpha\simeq3.4 for NZ=8N_Z = 8 and at α≃4.0\alpha\simeq4.0 for NZ=10N_Z = 10, but move to larger amplitudes as the field order increases and are absent at NZ=12N_Z = 12. They are therefore artifacts of the field truncation. Within the converged range, η\eta increases monotonically with α\alpha and has no stationary point.

We consequently use α=2\alpha= 2 for all dimensions and truncations. The regulator uncertainty is estimated as half the spread of η\eta over α∈[1,3]\alpha\in[1,3], or as the shift to the PMS point when that is larger. The latter applies only at O(∂4)O(\partial^4) in d=4d = 4 (Appendix E).

Dimension at which σ\sigma vanishes

The change in sign of σ\sigma between d=3d = 3 and d=2d = 2 implies a zero at an intermediate dimension dσd_\sigma. Using the running massless point, we can follow the fixed point through this dimension: as Δϕ→0\Delta_\phi\to0, the expansion point recedes to infinity, but the flows of the Taylor coefficients do not depend on its position (§ 1.4).

Figure 5 shows Δϕ\Delta_\phi near dσd_\sigma. It passes through zero linearly. Increasing the field order to NZ=14N_Z = 14 and 16 at O(∂4)O(\partial^4) changes dσd_\sigma by less than 5×10−45 \times10^{-4}.

Figure 5. Field dimension Δϕ\Delta_\phi through dσd_\sigma at α=2\alpha=2, at O(∂2)\mathcal{O}(\partial^2) and in the complete O(∂4)\mathcal{O}(\partial^4) truncation (NZ=16N_Z=16). The dotted vertical line marks the zero of Δϕ\Delta_\phi at O(∂4)\mathcal{O}(\partial^4), where σ=0\sigma=0.

We obtain dσ=2.650(13)d_\sigma= 2.650(13) at O(∂2)O(\partial^2) and 2.653(10) at O(∂4)O(\partial^4). The uncertainties combine the field-order dependence and the regulator variation over α∈[1,3]\alpha\in[1,3]. Taking half the difference between the two orders as the derivative-expansion uncertainty gives

dσ=2.65(1).(15)d_\sigma= 2.65(1). \tag*{(15)}

At this dimension, η=2−dσ≃−0.65\eta= 2 - d_\sigma\simeq-0.65.

An independent estimate follows from the ϵ\epsilon expansion. Applying the six-loop Padé analysis of Ref. [19] to σ\sigma gives dσ=2.649(30)d_\sigma= 2.649(30) and 2.649(25) [66]. These estimates use, respectively, the mean and the loop-order-weighted mean of the valid Padé approximants, and both agree with our result. A conformal-bootstrap calculation using the determinant method found dσ=2.6199d_\sigma= 2.6199, without an uncertainty estimate [71].

Correction to scaling and ω\omega exponent

The leading correction to scaling2 is governed by the smallest irrelevant eigenvalue of the stability matrix, ω=−θ3>0\omega= -\theta_3 > 03. Unlike θh\theta_h and θt\theta_t, ω\omega is independent of η\eta.

We compute ω\omega with the same representations of the field dependence and the same choice α=2\alpha= 2 as for η\eta. The result is displayed in Figure 6. Numerically, at O(∂4)\mathcal{O}(\partial^4) we find

Figure 6. Correction-to-scaling exponent ω\omega in the complete O(∂4)\mathcal{O}(\partial^4) truncation at α=2\alpha=2. The band combines the field-truncation, regulator and normalization-point uncertainties; the solid line uses fixed imaginary expansion points and the dashed line the running massless point. Filled circles mark our values at d=2d=2, 33, 44 and 55, with error bars that include the derivative-expansion estimate. Below d=2d=2, where the field expansion of ω\omega is not converged, the line is dotted and the hatched band shows the change from NZ=14N_Z=14 to 1616. The dotted gray line is the [3/3] Padé approximant of the six-loop ϵ\epsilon expansion of ω/ϵ\omega/\epsilon [19], constrained by the exact value ω=2\omega=2 in d=2d=2. Other symbols show the six-loop [19] and five-loop [18] ϵ\epsilon-expansion estimates, the fuzzy-sphere results in d=3d=3 [27] and d=5d=5 [28], the estimates of Ref. [24] (including d=1d=1 and 22), and the exact d=1d=1 and d=2d=2 values. Literature symbols are displaced horizontally for visibility.

ω={0.701(8)±0.004,d=5,1.20(4)±0.02,d=4,1.59(3)±0.05,d=3,(16)\omega= \begin{cases} 0.701(8) \pm0.004, & d = 5, \\ 1.20(4) \pm0.02, & d = 4, \\ 1.59(3) \pm0.05, & d = 3, \end{cases} \tag*{(16)}

where the second error is the derivative-expansion estimate ∣ωO(∂4)−ωO(∂2)∣/3|\omega_{\mathcal{O}(\partial^4)}-\omega_{\mathcal{O}(\partial^2)}|/3. These values agree with the six-loop constrained-Padé estimates of Ref. [19], ω=0.6970(2)\omega= 0.6970(2), 1.1924(10)1.1924(10) and 1.615(1)1.615(1) in d=5d = 5, 44 and 33, respectively, and with the five-loop estimates 0.691(8)0.691(8), 1.15(7)1.15(7) and 1.5(2)1.5(2) [18]. In d=3d = 3, our value also agrees with the fuzzy-sphere result Δϕ3=4.613(6)\Delta_{\phi^3} = 4.613(6), i.e., ω=1.613(6)\omega= 1.613(6) [27]. In d=5d = 5, the fuzzy four-sphere gives Δϕ3=5.75(2)\Delta_{\phi^3} = 5.75(2), i.e., ω=0.75(2)\omega= 0.75(2) [28], somewhat above both our result and the ϵ\epsilon expansion. The older estimates of Ref. [24], 1.72(8)1.72(8), 1.30(3)1.30(3) and 0.73(2)0.73(2), lie above ours in all three dimensions.

At dσd_\sigma we find ω=1.72(8)\omega= 1.72(8), including the derivative-expansion uncertainty. Below dσd_\sigma, ω\omega continues to increase; we discuss this region in the next subsection.

Below dσd_\sigma: two and one dimensions

We continue the running-point calculation from dσd_\sigma to d=1d = 1 at O(∂2)\mathcal{O}(\partial^2), O(∂4)\mathcal{O}(\partial^4) and O(∂4)W1\mathcal{O}(\partial^4)_{W_1}, with α=2\alpha= 2. The resulting curves are included in Figure 2, Figure 3 and Figure 6, and Table II lists values at selected dimensions. We estimate the uncertainties as above dσd_\sigma, with three changes. The field-truncation error is the change from NZ=14N_Z = 14 to 1616 at O(∂4)\mathcal{O}(\partial^4), and from 1212 to 1414 at O(∂2)\mathcal{O}(\partial^2) and O(∂4)W1\mathcal{O}(\partial^4)_{W_1}. At O(∂4)\mathcal{O}(\partial^4), the regulator error, half the spread over α∈[1,3]\alpha\in[1,3], is evaluated at NZ=14N_Z = 14 for d≤1.5d \leq1.5 and at NZ=12N_Z = 12 above. The regulator error is not available below d≃1.04d \simeq1.04, as at NZ=14N_Z = 14 a spurious complex pair of eigenvalues becomes relevant for α=3\alpha= 3 and our calculations become unreliable. The field expansion converges more slowly as dd decreases. At O(∂4)O(\partial^4), η\eta changes by less than 5×10−35 \times10^{-3} from NZ=14N_Z = 14 to 16 down to d≃1.3d \simeq1.3; at O(∂2)O(\partial^2) and O(∂4)W1O(\partial^4)_{W_1}, by less than 5×10−35 \times10^{-3} from NZ=12N_Z = 12 to 14 down to d≃1.6d \simeq1.6. For ω\omega, the change from NZ=14N_Z = 14 to 16 stays below 0.05, and that from 12 to 14 below 0.15, only down to d=2d = 2. Below these dimensions, Figure 2–Figure 6 show the NZN_Z dependence as a hatched band, and we regard our results in this region as qualitative.

In d=2d = 2, we find η=−0.795(20)±0.007\eta= -0.795(20) \pm0.007 at O(∂4)O(\partial^4), in agreement with the exact value −4/5-4/5 [5]. For the correction-to-scaling exponent, we obtain ω=1.90(5)±0.03\omega= 1.90(5) \pm0.03, compared with the exact value ω=2\omega= 2. Since ω\omega is not determined by η\eta, this comparison is an independent test of the derivative expansion at d=2d = 2.

In d=1d = 1, the O(∂4)O(\partial^4) result η=−0.97\eta= -0.97 is consistent with the exact η=−1\eta= -1, with a field-truncation uncertainty of 0.04. Since the field expansion is not converged below d≃1.3d \simeq1.3, these values are qualitative. A quantitative result near d=1d = 1 would require higher orders of derivative expansion and higher field orders or a different representation of the field dependence.

Summary and outlook

We have determined the critical exponents of the Yang–Lee edge singularity using the functional renormalization group at second and fourth order in the derivative expansion. Two features distinguish our work from earlier FRG studies: First, complete derivative expansions: We retain all operators at O(∂2)O(\partial^2) and O(∂4)O(\partial^4). Omitting two of the three four-derivative operators changes the fourth-order correction by up to 50%, and the sign of the discrepancy depends on dimension. Second, a convergent representation of the field dependence: For the complex fixed point considered here, expansions about ϕ=0\phi= 0 fail to converge with our regulator. Expanding about imaginary field values resolves this problem. In particular, the running massless point used by AMS gives convergent expansions down to d≃1.3d \simeq1.3.

Our fourth-order values of η\eta agree, within the quoted uncertainties, with the six-loop ϵ\epsilon-expansion results in d=5d = 5, 4 and 3 [19]. The correction-to-scaling exponent, ω=0.701(8)\omega= 0.701(8), 1.20(4) and 1.59(3) in d=5d = 5, 4 and 3, is not fixed by η\eta and therefore tests the truncation independently. It agrees with the six-loop constrained-Padé estimates [19] and, in d=3d = 3, with the fuzzy-sphere result [27]; as for η\eta, the step from O(∂2)O(\partial^2) to O(∂4)O(\partial^4) moves it toward these values. In d=2d = 2, they agree with the exact values of η\eta, σ\sigma, νc\nu_c, and the correction-to-scaling exponent ω=1.90(6)\omega= 1.90(6) is close to the exact ω=2\omega= 2. Towards d=1d = 1 the field expansion converges more slowly; there our results are only qualitatively consistent with the exact values.

The main remaining limitation is the uncertainty associated with the derivative expansion itself. At present, we estimate it from the difference between second and fourth order. We have derived the strict O(∂6)O(\partial^6) flows for twelve of the thirteen coefficient functions. Deriving the remaining flow and solving the resulting equations numerically will require a more efficient evaluation of the large expressions. A sixth-order calculation would allow us to apply the error-estimation methods developed for the Ising model [44].

Our determination of dσd_\sigma agrees with the six-loop $\epsilonexpansion estimate 2.649(25) [19, 66].

Addendum

After completing this work, we found the Mathematica code accompanying the arXiv submission of Ref. [61]. Comparing its implementation with ours allowed us to identify the source of the reported eigenvalue-relation violations.

The code evaluates the flows of the Taylor coefficients U(n)(iχ)U^{(n)}(i\chi) and z(n)(iχ)z^{(n)}(i\chi) at fixed field, omitting the contribution F(n+1)(iχ)∂t(iχ)F^{(n+1)}(i\chi)\partial_t(i\chi) from the motion of the expansion point. Since this point runs with kk, the chain rule requires this contribution in the flow of every Taylor coefficient. In Ref. [61], however, it is retained only in the equation for η\eta.

The omission leaves the fixed-point solution unchanged, because ∂tχ=0\partial_t\chi=0 at the fixed point. It does change the linearized flow: the stability matrix no longer has Δϕ\Delta_\phi as an exact eigenvalue. Removing the same contribution from our equations reproduces the large violations reported by AMS. We obtain 15% in d=5d=5 and 74% in d=4d=4, with the violation increasing to 200% in d=3.5d=3.5.

Acknowledgments

We thank J. A. Gracey for correspondence and for the six-loop estimate of dσd_\sigma. We also thank Piero Beretta and Carlos Sánchez-Villalobos for discussions on FRG and YLE. V.S. is indebted to ITP Heidelberg for hospitality and for creating a stimulating environment that led to this work. K.K. acknowledges funding by the GSI Helmholtzzentrum für Schwerionenforschung and by HGS-HIRe for FAIR. This research is supported in part by the ExtreMe Matter Institute EMMI at the GSI Helmholtzzentrum für Schwerionenforschung, Darmstadt, Germany. This work is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy EXC 2181/1 - 390900948 (the Heidelberg STRUCTURES Excellence Cluster) and the Collaborative Research Centre SFB 1225 - 273811115 (ISOQUANT). The derivation, numerical code and data that support this work are available from the authors upon reasonable request.

Outline for the derivation of the derivative-expansion flows

This appendix describes the algorithm implemented in our Mathematica code. Given a local operator basis for a single scalar field, it generates the functional flow equations for all coefficient functions without imposing an internal symmetry. The dimension dd remains a symbolic parameter throughout the calculation.

Vertices of the ansatz

We write each operator in Eq. (8) as F(ϕ)O[∂ϕ,∂2ϕ,…]F(\phi)\mathcal{O}[\partial\phi,\partial^2\phi,\ldots]. The derivative structure O\mathcal{O} consists of nOn_{\mathcal{O}} slots, each representing a field with a specified set of derivatives, contracted in a prescribed pattern. We expand around a uniform background,

ϕ(x)=ϕ+∑iϵieipi⋅x,\phi(x)=\phi+\sum_i \epsilon_i e^{i p_i\cdot x},

and extract the coefficient of ϵ1⋯ϵn\epsilon_1\cdots\epsilon_n to obtain the nn-point vertex:

Γ(n)(p1,…,pn;ϕ)=∑OF(n−nO)(ϕ)∑σfO(pσ(1),…,pσ(nO)).(17)\Gamma^{(n)}(p_1,\ldots,p_n;\phi)=\sum_{\mathcal{O}}F^{(n-n_{\mathcal{O}})}(\phi)\sum_{\sigma}f_{\mathcal{O}}(p_{\sigma(1)},\ldots,p_{\sigma(n_{\mathcal{O}})}). \tag*{(17)}

Here σ\sigma runs over injective maps from the operator slots to the external legs. The function fOf_{\mathcal{O}} contains the momentum factors ipip generated by the derivatives, together with the operator’s explicit normalization. For example,

fZ(k1,k2)=−12k1⋅k2,f_Z(k_1,k_2)=-\frac{1}{2}k_1\cdot k_2,
fW2(k1,k2,k3)=12(k1⋅k2)k32,f_{W_2}(k_1,k_2,k_3)=\frac{1}{2}(k_1\cdot k_2)k_3^2,

and

fW3(k1,…,k4)=12(k1⋅k2)(k3⋅k4).f_{W_3}(k_1,\ldots,k_4)=\frac{1}{2}(k_1\cdot k_2)(k_3\cdot k_4).

All momenta are incoming, with ∑ipi=0\sum_i p_i=0.

Vertex flows

We expand Tr⁡[∂tRk(Γ0(2)+V)−1]\operatorname{Tr}[\partial_tR_k(\Gamma_0^{(2)}+V)^{-1}] in the fluctuation-dependent part VV of the two-point function. Extracting the multilinear coefficient gives

∂tΓ(n)(p1,…,pn)=12∑(B1,…,Bk)(−1)k∫q∂tRk(q)G(q)2∏j=1kΓ(2+∣Bj∣)(q+Qj−1,−q−Qj,{pi}i∈Bj)∏j=1k−1G(q+Qj).(18)\partial_t\Gamma^{(n)}(p_1,\ldots,p_n)=\frac{1}{2}\sum_{(B_1,\ldots,B_k)}(-1)^k\int_q\partial_tR_k(q)G(q)^2\prod_{j=1}^{k}\Gamma^{(2+|B_j|)}(q+Q_{j-1},-q-Q_j,\{p_i\}_{i\in B_j})\prod_{j=1}^{k-1}G(q+Q_j). \tag*{(18)}

where (B1,…,Bk)(B_1,\ldots,B_k) runs over all ordered set partitions of {1,…,n}\{1,\ldots,n\} and

Qj=∑l≤j∑i∈Blpi,Q0=0.Q_j=\sum_{l\leq j}\sum_{i\in B_l}p_i,\qquad Q_0=0.

The insertion of ∂tRk\partial_tR_k singles out a position on the loop, so the block order must be retained. Equation (A2) reproduces the usual diagrammatic expressions. For n=2n=2, for example, it reduces to

∂tΓ(2)(p)=∫q∂tR G2[Γ(3)G(q+p)Γ(3)−12Γ(4)].\partial_t\Gamma^{(2)}(p)=\int_q\partial_tR\,G^2\left[\Gamma^{(3)}G(q+p)\Gamma^{(3)}-\frac{1}{2}\Gamma^{(4)}\right].

Momentum expansion and angular average

We introduce a counting parameter λ\lambda by rescaling the external momenta as pi→λpip_i\to\lambda p_i. The shifted propagators are expanded according to

G((q+Q)2)=∑mG(m)(q2)m!(2q⋅Q+Q2)m.G((q+Q)^2)=\sum_m\frac{G^{(m)}(q^2)}{m!}(2q\cdot Q+Q^2)^m.

Every product is retained through order λs\lambda^s, where ss is the derivative order of the truncation. We then perform the angular averages using

⟨∏a=12m(q⋅ka)⟩=(q2)m∑pairings∏(ka⋅kb)d(d+2)⋯(d+2m−2).(19)\left\langle\prod_{a=1}^{2m}(q\cdot k_a)\right\rangle= \frac{(q^2)^m\sum_{\mathrm{pairings}}\prod(k_a\cdot k_b)}{d(d+2)\cdots(d+2m-2)}. \tag*{(19)}

This reduces the loop expressions to radial integrands while keeping their dependence on dd explicit.

Projection

After eliminating pnp_n by momentum conservation, we expand both sides of Eq. (18) in monomials of the independent invariants pi⋅pjp_i\cdot p_j. Matching their coefficients gives a linear system,

Ax=b,A x=b,

where xx contains the unknown flows ∂tF(j)\partial_t F^{(j)} entering Γ(n)\Gamma^{(n)}, and bb contains the one-dimensional loop integrands.

For n≥3n\geq3, the system is overdetermined. We solve it using the Moore–Penrose left inverse and verify that every component of the residual Ax−bAx-b vanishes identically at the integrand level. This checks that the momentum-expanded loop expression can be represented as the vertex of a local functional through O(∂s)\mathcal{O}(\partial^s).

We also check consistency between different vertex orders. For example, the flows of Z′Z' and W1′W_1' extracted from Γ(3)\Gamma^{(3)} must equal the field derivatives of the corresponding flows from Γ(2)\Gamma^{(2)}. Likewise, the flows of Z′′Z'', W1′′W_1'' and W2′W_2' extracted from Γ(4)\Gamma^{(4)} must agree with the appropriate derivatives of the lower-vertex flows. All these identities hold at any yy.

Convergence in the field expansion

We truncate the Taylor expansions (12) at NU=NZ+3N_U=N_Z+3, NW1=NW3=NZ−2N_{W_1}=N_{W_3}=N_Z-2 and NW2=NZ−1N_{W_2}=N_Z-1. These relative orders follow from the coupling between coefficient functions in the flow equations. For example, the flow of UU at order nn involves ZZ at order n−2n-2 through the term Zq2Zq^2 in the propagator. With these offsets, a single parameter, NZN_Z, controls the field truncation.

Figure 7 illustrates the effect of the expansion point. At d=5.5d=5.5, the expansion about ϕ=0\phi=0 produces a non-decaying period-three oscillation as NZN_Z increases. Moving the expansion point to ϕ=−2i\phi=-2i instead gives rapid convergence, reaching seven-digit stability. A similar convergence pattern was found in Ref. [72] in a different context.

For the running massless point, Eq. (13), convergence at α=2\alpha=2 is shown in Figure 8. At O(∂2)\mathcal{O}(\partial^2), η\eta changes by less than 3×10−43\times10^{-4} beyond NZ=8N_Z=8 in every dimension shown above dσd_\sigma. We use NZ=10N_Z=10 for these calculations above dσd_\sigma.

Figure 8. Convergence of η\eta with the field order NZN_Z for the running massless point at α=2\alpha= 2. The left panel shows O(∂2)\mathcal{O}(\partial^2) and the right panel O(∂4)\mathcal{O}(\partial^4). Each curve gives the difference between η(NZ)\eta(N_Z) and its value at the highest field order computed in that dimension. The reference order is NZ=14(16)N_Z = 14(16) at O(∂2)\mathcal{O}(\partial^2) (O(∂4)\mathcal{O}(\partial^4)).

At O(∂4)\mathcal{O}(\partial^4), convergence is slower. Near d=3d=3, increasing NZN_Z from 8 to 10 changes η\eta by 2×10−32\times10^{-3}, while the change from 10 to 12 is 3×10−43\times10^{-4}. Convergence deteriorates closer to six dimensions, where the running massless point approaches the origin. At d=5d=5, the changes from NZ=10N_Z=10 to 12 and from 12 to 14 are about 10−310^{-3} and 7×10−47\times10^{-4}, respectively.

We therefore use fixed imaginary expansion points at O(∂4)\mathcal{O}(\partial^4) wherever they give convergent results, namely for d≥3.2d\geq3.2. For these expansions, the change from NZ=8N_Z=8 to 10 is below 2×10−42\times10^{-4}. The running-point results remain within 10−310^{-3} of the fixed-expansion-point results; at NZ=14N_Z=14, the differences are 3×10−43\times10^{-4} in d=5d=5 and 8×10−48\times10^{-4} in d=4d=4.

Below d=3.2d=3.2, we use the running massless point with NZ=12N_Z=12 at O(∂4)\mathcal{O}(\partial^4). We estimate the field-truncation uncertainty from the change between NZ=10N_Z=10 and 12. This change is at most 1.1×10−31.1\times10^{-3} over the dimensions considered in this range, and is 3×10−43\times10^{-4} at d=3d=3.

We have also computed the running-point solution at O(∂4)\mathcal{O}(\partial^4) with NZ=14N_Z=14 and 16 over the whole range 1≤d≤61\leq d\leq6. Above dσd_\sigma, the two orders differ by ≤3×10−4\leq3\times10^{-4} in η\eta and ≤7×10−3\leq7 \times10^{-3} in ω\omega.

Below dσd_\sigma, the field expansion converges more slowly as dd decreases and requires larger NZN_Z. At O(∂2)O(\partial^2), the change from NZ=12N_Z = 12 to 14 is 10−310^{-3} in d=2d = 2. In d=2d = 2, the O(∂4)O(\partial^4) values of η\eta at NZ=10N_Z = 10, 12, 14 and 16 are −0.7997-0.7997, −0.7938-0.7938, −0.7942-0.7942 and −0.7945-0.7945 (Figure 8); in d=1.5d = 1.5, NZ=14N_Z = 14 and 16 give −0.8915-0.8915 and −0.8926-0.8926, while in d=1d = 1 they differ by 0.04.

Exact translation eigenvalue with the running massless point

Here we demonstrate that with the running massless point, Eq. (13), the stability matrix has the eigenvalues θt=Δϕ\theta_t = \Delta_\phi and θh=d−Δϕ\theta_h = d - \Delta_\phi exactly, for any order of the field expansion. We give the explicit derivation in the LPA, where every step can be written out explicitly, and then explain why it carries over to O(∂2)O(\partial^2) and O(∂4)O(\partial^4). Eigenvalues θ\theta are defined as the eigenvalues of −M-M, where Mij=∂(∂tgi)/∂gjM_{ij} = \partial(\partial_t g_i)/\partial g_j is the stability matrix of the couplings gig_i, so that relevant directions have θ>0\theta> 0.

In the LPA, Eq. (10) for the potential reads

∂tU=−dU+ΔϕϕU′+ℓ(U′′).(20)\partial_t U = -d U + \Delta_\phi\phi U' + \ell(U''). \tag*{(20)}

where ℓ(m)=∫0∞dy yd/2−1IU(y;m)\ell(m) = \int_0^\infty\mathrm{d}y\, y^{d/2-1} I_U(y;m) is the loop integral evaluated with the mass mm. Differentiating nn times with respect to ϕ\phi gives

∂tU(n)=(−d+nΔϕ)U(n)+ΔϕϕU(n+1)+∂ϕnℓ(U′′).(21)\partial_t U^{(n)} = (-d + n\Delta_\phi) U^{(n)} + \Delta_\phi\phi U^{(n+1)} + \partial_\phi^n \ell(U''). \tag*{(21)}

The field enters explicitly only through the dilatation term ΔϕϕU(n+1)\Delta_\phi\phi U^{(n+1)}, while the loop term is a simple polynomial in U(3),…,U(n+2)U^{(3)},\ldots,U^{(n+2)}.

The FRG couplings are the Taylor coefficients at the massless point, we denote them by un≡U(n)(iχ)u_n \equiv U^{(n)}(i\chi). Because the expansion point moves with kk, the flow of unu_n is the total derivative

∂tun=[∂tU(n)](iχ)+un+1i ∂tχ.(22)\partial_t u_n = [\partial_t U^{(n)}](i\chi) + u_{n+1} i\,\partial_t\chi. \tag*{(22)}

The second term describes the drift of the expansion point. Inserting Eq. (21), the explicit χ\chi of the dilatation term and the drift term combine into

∂tun=(−d+nΔϕ)un+Ln+un+1ψ,(23)\partial_t u_n = (-d + n\Delta_\phi)u_n + L_n + u_{n+1}\psi, \tag*{(23)}

with ψ≡i(∂tχ+Δϕχ)\psi\equiv i(\partial_t\chi+\Delta_\phi\chi) and Ln≡∂ϕnℓ(U′′)∣ϕ=iχL_n \equiv\left.\partial_\phi^n\ell(U'')\right|_{\phi=i\chi}. Because U′′(iχ)=0U''(i\chi)=0, the loop derivatives are taken at zero mass, and ℓ(j)≡ℓ(j)(0)\ell^{(j)} \equiv\ell^{(j)}(0) are merely numbers that depend only on dd, η\eta, and the regulator, e.g.

L0=ℓ,L1=ℓ(1)u3,L2=ℓ(1)u4+ℓ(2)u32,L3=ℓ(1)u5+3ℓ(2)u3u4+ℓ(3)u33.(24)\begin{aligned} L_0 &= \ell, & L_1 &= \ell^{(1)}u_3, & L_2 &= \ell^{(1)}u_4 + \ell^{(2)}u_3^2,\\ L_3 &= \ell^{(1)}u_5 + 3\ell^{(2)}u_3u_4 + \ell^{(3)}u_3^3. \tag*{(24)} \end{aligned}

Note that LnL_n do not depend on χ\chi.

The condition u2=0u_2 = 0 holds for all kk. Equation (C4) for n=2n = 2 then fixes ψ\psi:

0=L2+u3ψ⟹ψ=−L2u3.(25)0 = L_2 + u_3\psi\quad\Longrightarrow\quad\psi= -\frac{L_2}{u_3}. \tag*{(25)}

With this, the flow equations become

∂tun=(−d+nΔϕ)un+Ln−un+1L2u3,n≠2.(26)\partial_t u_n = (-d + n\Delta_\phi)u_n + L_n - u_{n+1}\frac{L_2}{u_3}, \qquad n \ne2. \tag*{(26)}
∂tχ=−Δϕχ+iL2u3.(27)\partial_t\chi= -\Delta_\phi\chi+ i\frac{L_2}{u_3}. \tag*{(27)}

Here one has to account for the field truncation un=0u_n = 0 for n>NUn > N_U. The right-hand side of Eq. (26) then depends on u3,…,uNUu_3,\ldots,u_{N_U} (and on u0,u1u_0,u_1 for n=0,1n = 0,1), but, most importantly, not on χ\chi. The position χ\chi appears only in Eq. (C8); it enters there linearly and with exact coefficient −Δϕ-\Delta_\phi!

Explicit example: NU=4N_U = 4. With u5=u6=0u_5 = u_6 = 0, the complete set of flow equations is

∂tu4=(−d+4Δϕ)u4+3ℓ(2)u32+6ℓ(3)u3u4+ℓ(4)u34,\partial_t u_4 = (-d + 4\Delta_\phi)u_4 + 3\ell^{(2)}u_3^2 + 6\ell^{(3)}u_3u_4 + \ell^{(4)}u_3^4,
∂tu3=(−d+3Δϕ)u3+2ℓ(2)u3u4+ℓ(3)u33−ℓ(1)u42u3,\partial_t u_3 = (-d + 3\Delta_\phi)u_3 + 2\ell^{(2)}u_3u_4 + \ell^{(3)}u_3^3 - \ell^{(1)}\frac{u_4^2}{u_3},
∂tu1=(−d+Δϕ)u1+ℓ(1)u3,\partial_t u_1 = (-d + \Delta_\phi)u_1 + \ell^{(1)}u_3,
∂tu0=−du0+ℓ+u1ψ,\partial_t u_0 = -d u_0 + \ell+ u_1\psi,
∂tχ=−Δϕχ+iℓ(1)u4+ℓ(2)u32u3.(28)\partial_t\chi= -\Delta_\phi\chi+ i\frac{\ell^{(1)}u_4 + \ell^{(2)}u_3^2}{u_3}. \tag*{(28)}

In the u1u_1 equation the drift term is absent because u2=0u_2 = 0. With the couplings ordered as (u4,u3,u1,u0,χ)(u_4,u_3,u_1,u_0,\chi), the stability matrix is lower triangular in blocks,

M=(K44K43000K34K330000ℓ(1)−(d−Δϕ)00∗∗ψ−d0∗∗00−Δϕ),(29)M = \begin{pmatrix} K_{44} & K_{43} & 0 & 0 & 0 \\ K_{34} & K_{33} & 0 & 0 & 0 \\ 0 & \ell^{(1)} & -(d-\Delta_\phi) & 0 & 0 \\ * & * & \psi& -d & 0 \\ * & * & 0 & 0 & -\Delta_\phi \end{pmatrix}, \tag*{(29)}

where KK is the 2×22 \times2 matrix of derivatives of the first two equations in Eq. (28) and the entries marked ∗* do not affect the eigenvalues. The eigenvalues are therefore those of −K-K, which describe the irrelevant directions, together with the obvious

θh=d−Δϕ,θ=d,θt=Δϕ.(30)\theta_h = d-\Delta_\phi,\qquad\theta= d,\qquad\theta_t = \Delta_\phi. \tag*{(30)}

The eigenvalue dd belongs to the field-independent part of the potential. The column of MM belonging to χ\chi vanishes except for its diagonal entry, so the unit vector eχe_\chi is an exact eigenvector. It describes the infinitesimal translation of the fixed point: U(ϕ)→U(ϕ−i δχ)U(\phi) \to U(\phi-i\,\delta\chi) moves the massless point to i(χ+δχ)i(\chi+\delta\chi) and leaves every unu_n unchanged.

The same block structure holds for every NUN_U: the equations for u3,…,uNUu_3,\ldots,u_{N_U} form a closed subsystem, u1u_1 feeds only into u0u_0, and χ\chi feeds into nothing. Hence θt=Δϕ\theta_t=\Delta_\phi and θh=d−Δϕ\theta_h=d-\Delta_\phi for every truncation order.

Comparison with a fixed expansion point. For a fixed point iχ0i\chi_0, there is no drift term and Eq. (23) is replaced by

∂tun=(−d+nΔϕ)un+iΔϕχ0un+1+∂ϕ2ℓ(U′′)∣iχ0,(31)\partial_t u_n = (-d+n\Delta_\phi)u_n+i\Delta_\phi\chi_0u_{n+1}+\left.\partial_\phi^2\ell(U^{\prime\prime})\right|_{i\chi_0}, \tag*{(31)}

with u2≠0u_2\ne0. The translation corresponds to δun=un+1\delta u_n=u_{n+1}. Differentiating the fixed-point equation with respect to ϕ\phi shows that this is an eigenvector with eigenvalue −Δϕ-\Delta_\phi of the untruncated system. In a truncated system the components δuNU=uNU+1=0\delta u_{N_U}=u_{N_U+1}=0. The relation θt=Δϕ\theta_t=\Delta_\phi is then violated by an amount that decreases only as the Taylor series converges; this leads to a non-zero ϵscale\epsilon_{\mathrm{scale}} of Figure 1.

Higher orders of the derivative expansion. At O(∂2)\mathcal{O}(\partial^2) and O(∂4)\mathcal{O}(\partial^4) each coefficient function FF obeys Eq. (10). Its loop term depends on the field only through the functions and their derivatives, and the dilatation term is again ΔϕϕF′\Delta_\phi\phi F'. For the Taylor coefficients fn=F(n)(iχ)f_n=F^{(n)}(i\chi) the dilatation and drift terms therefore combine into fn+1ψf_{n+1}\psi, as in Eq. (23). The two conditions defining the expansion point, U′′(iχ)=0U^{\prime\prime}(i\chi)=0 and Z(iχ)=1Z(i\chi)=1, give two equations, 0=L2+u3ψ0=L_2+u_3\psi and 0=η+z1ψ+Lz0=\eta+z_1\psi+L_z, where z1=z′(iχ)z_1=z'(i\chi) and LzL_z is the loop term of ZZ. All coefficients in these equations are independent of χ\chi, and they determine ψ\psi and η\eta. All flows of the Taylor coefficients are therefore independent of χ\chi, while ∂tχ=−Δϕχ−iψ\partial_t\chi=-\Delta_\phi\chi-i\psi. The block structure of Eq. (29) is thus the exact eigenvalues θt=Δϕ\theta_t=\Delta_\phi and θh=d−Δϕ\theta_h=d-\Delta_\phi follow in the same way as in the LPA.

The anomalous dimension, and with it Δϕ=(d−2+η)/2\Delta_\phi=(d-2+\eta)/2, is a function of the Taylor coefficients, and we include this dependence when computing MM. It contributes to the rows of MM, including the row of χ\chi through the term −Δϕχ-\Delta_\phi\chi, but not to the column belonging to χ\chi, because η\eta does not depend on χ\chi. The exact eigenvalues are therefore obtained whether or not the dependence of η\eta on the couplings is included. The other eigenvalues do depend on this choice. Including the dependence is required to reproduce the one-loop result ω=ϵ+O(ϵ2)\omega=\epsilon+O(\epsilon^2); with η\eta held fixed at its fixed-point value, one finds instead ω≃4ϵ/3\omega\simeq4\epsilon/3 near six dimensions.

Full field dependence for the Ising test

For the Ising benchmark in Sec. VI, we resolve the full field dependence using Chebyshev collocation. We compactify the real field axis through

ξ=ϕ2ϕ2+ℓ2∈[0,1)\xi=\frac{\phi^2}{\phi^2+\ell^2}\in[0,1)

and factor out the known large-field behavior of each coefficient function,

F(ϕ)=(1+ϕ2ℓ2)pF/2F~(ξ),F(\phi)=\left(1+\frac{\phi^2}{\ell^2}\right)^{p_F/2}\widetilde{F}(\xi),

where pFp_F is its large-field power. We expand F~\widetilde{F} in Chebyshev polynomials, with Z2Z_2 symmetry incorporated into the basis, and impose the fixed-point equations at the Chebyshev nodes. The critical exponents are then obtained from the eigenvalues of the discretized stability matrix.

This method also works for the Yang–Lee fixed point close to six dimensions. At d=5.95d=5.95, it reproduces the Taylor-expansion result for η\eta to seven digits. Below d≈5d\approx5, however, the collocation problem becomes ill-conditioned.

For the Ising model, the large-field behavior along the real axis selects the physical solution. For the complex Yang–Lee solution, this condition is no longer sufficient, and the discretized equations admit nearly degenerate spurious solutions. Selecting the physical branch requires control of analyticity in the complex field plane. A Taylor expansion incorporates local analyticity directly, making it better suited to the present calculation. We therefore use Taylor expansions for all Yang–Lee results.

Uncertainty estimates

We have four major sources of uncertainty.

Field truncation. We estimate the uncertainty from the finite Taylor expansion by the change in η\eta when NZN_Z is increased by two. This change is below 5×10−45 \times10^{-4} at O(∂2)O(\partial^2) and below 3×10−33 \times10^{-3} at O(∂4)O(\partial^4) (Appendix § 1.6.2) for d≥dσd\ge d_\sigma. Below dσd_\sigma, we use the change from NZ=14N_Z=14 to 1616 at O(∂4)O(\partial^4) and from 1212 to 1414 at O(∂2)O(\partial^2) (Sec. VII F).

Regulator dependence. We take half the spread of η\eta over α∈[1,3]\alpha\in[1,3] as the regulator uncertainty. When a PMS point is available (Sec. VII C), we instead use the shift from the α=2\alpha=2 result to the PMS value if it is larger. This applies only at O(∂4)O(\partial^4) in d=4d=4, where the shift is 1.7×10−31.7 \times10^{-3}.

Normalization point. At finite order in the derivative expansion, η\eta depends weakly on the field value at which zz is normalized, since the regulator Rk∝Z0R_k\propto Z_0 is not invariant under field rescalings. We estimate this uncertainty from the difference between the two field representations described in § 1.4. It is below 3×10−43 \times10^{-4} for d≥4d\ge4 and reaches 3×10−33 \times10^{-3} near d=3.4d=3.4. This uncertainty is most likely correlated with the regulator dependence. Below dσd_\sigma, only the running massless point converges, and this estimate is not available.

Derivative expansion. The finite order of the derivative expansion is the dominant source of uncertainty. For the Ising model, the change in η\eta from O(∂4)O(\partial^4) to O(∂6)O(\partial^6) is several times smaller than that from O(∂2)O(\partial^2) to O(∂4)O(\partial^4) [44]. Assuming a similar reduction here, we assign the fourth-order result an uncertainty of

δDEη=13∣ηO(∂4)−ηO(∂2)∣.\delta_{\mathrm{DE}}\eta=\frac{1}{3}\left|\eta_{O(\partial^4)}-\eta_{O(\partial^2)}\right|.

References

References

  1. [1]C.-N. Yang and T. D. Lee, Statistical theory of equations of state and phase transitions. 1. Theory of condensation, Phys. Rev. 87, 404 (1952).
  2. [2]T. D. Lee and C.-N. Yang, Statistical theory of equations of state and phase transitions. 2. Lattice gas and Ising model, Phys. Rev. 87, 410 (1952).
  3. [3]P. J. Kortman and R. B. Griffiths, Density of Zeros on the Lee-Yang Circle for Two Ising Ferromagnets, Phys. Rev. Lett. 27, 1439 (1971).DOI
  4. [4]M. E. Fisher, Yang-Lee Edge Singularity and phi***3 Field Theory, Phys. Rev. Lett. 40, 1610 (1978).DOI
  5. [5]J. L. Cardy, Conformal Invariance and the Yang-lee Edge Singularity in Two-dimensions, Phys. Rev. Lett. 54, 1354 (1985).pubmed.ncbi.nlm.nih.gov/10031010
  6. [6]S. Mukherjee and V. Skokov, Universality driven analytic structure of the QCD crossover: radius of convergence in the baryon chemical potential, Phys. Rev. D 103, L071501 (2021), arXiv:1909.04639 [hep-ph].DOI
  7. [7]M. Bryant, C. Schmidt, and V. V. Skokov, Asymptotic behavior of the Fourier coefficients and the analytic structure of the QCD equation of state, Phys. Rev. D 109, 076021 (2024), arXiv:2401.06489 [hep-ph].arxiv.org/abs/2401.06489
  8. [8]D. A. Clarke, P. Dimopoulos, F. Di Renzo, J. Goswami, C. Schmidt, S. Singh, and K. Zambello, Searching for the QCD critical end point using multipoint Padé approximations, Phys. Rev. D 112, L091504 (2025), arXiv:2405.10196 [hep-lat].
  9. [9]Z.-Y. Wan, Y. Lu, F. Gao, and Y.-X. Liu, Lee–Yang edge singularities in QCD via the Dyson –Schwinger equations, Eur. Phys. J. C 84, 899 (2024), arXiv:2401.04957 [hep-ph].arxiv.org/abs/2401.04957
  10. [10]G. Basar, QCD critical point, Lee-Yang edge singularities, and Padé resummations, Phys. Rev. C 110, 015203 (2024), arXiv:2312.06952 [hep-th].arxiv.org/abs/2312.06952
  11. [11]H.-T. Ding, S. Mukherjee, P. Petreczky, and K.-F. Ye, QCD Chiral Crossover Line from Lee-Yang Edge Singularities (2026), arXiv:2608.05752 [hep-lat].arxiv.org/abs/2608.05752
  12. [12]V. V. Skokov, Two lectures on Yang-Lee edge singularity and analytic structure of QCD equation of state, SciPost Phys. Lect. Notes 91, 1 (2025), arXiv:2411.02663 [hep-ph].DOI
  13. [13]A. J. Macfarlane and G. Woo, φ3 Theory in Six Dimensions and the Renormalization Group, Nucl. Phys. B 77, 91 (1974), [Erratum: Nucl.Phys.B 86, 548-548 (1975)].DOI
  14. [14]O. F. de Alcantara Bonfim, J. E. Kirkham, and A. J. McKane, Critical Exponents to Order ε3 for φ3 Models of Critical Phenomena in Six ε-dimensions, J. Phys. A 13, L247 (1980), [Erratum: J.Phys.A 13, 3785 (1980)].
  15. [15]O. F. de Alcantara Bonfim, J. E. Kirkham, and A. J. McKane, Critical Exponents for the Percolation Problem and the Yang-lee Edge Singularity, J. Phys. A 14, 2391 (1981).
  16. [16]J. A. Gracey, Four loop renormalization of ϕ3 theory in six dimensions, Phys. Rev. D 92, 025012 (2015), arXiv:1506.03357 [hep-th].arxiv.org/abs/1506.03357
  17. [17]M. Kompaniets and A. Pikelner, Critical exponents from five-loop scalar theory renormalization near six-dimensions, Phys. Lett. B 817, 136331 (2021), arXiv:2101.10018 [hep-th].DOI
  18. [18]M. Borinsky, J. A. Gracey, M. V. Kompaniets, and O. Schnetz, Five-loop renormalization of ϕ3 theory with applications to the Lee-Yang edge singularity and percolation theory, Phys. Rev. D 103, 116024 (2021), arXiv:2103.16224 [hep-th].DOI
  19. [19]J. A. Gracey, Six loop critical exponent analysis for Lee-Yang and percolation theory, Phys. Rev. D 112, 105019 (2025), arXiv:2510.05723 [hep-th].arxiv.org/abs/2510.05723
  20. [20]D. A. Kurtze and M. E. Fisher, Yang-Lee edge singularities at high temperatures, Phys. Rev. B 20, 2785 (1979).DOI
  21. [21]P. Butera and M. Pernici, Yang-Lee edge singularities from extended activity expansions of the dimer density for bipartite lattices of dimensionality 2 <= d <= 7, Phys. Rev. E 86, 011104 (2012), arXiv:1206.0872 [cond-mat.stat-mech].DOI
  22. [22]F. Gliozzi, More constraining conformal bootstrap, Phys. Rev. Lett. 111, 161602 (2013), arXiv:1307.3111 [hep-th].DOI
  23. [23]F. Gliozzi and A. Rago, Critical exponents of the 3d Ising and related models from Conformal Bootstrap, JHEP 10, 042, arXiv:1403.6003 [hep-th].DOI
  24. [24]S.-N. Lai and M. E. Fisher, The universal repulsive-core singularity and Yang–Lee edge criticality, J. Chem. Phys. 103, 8144 (1995).
  25. [25]H.-P. Hsu, W. Nadler, and P. Grassberger, Simulations of lattice animals and trees, J. Phys. A 38, 775 (2005).DOI
  26. [26]J. Elias Miró and O. Delouche, Flowing from the Ising model on the fuzzy sphere to the 3D Lee-Yang CFT, JHEP 10, 037, arXiv:2505.07655 [hep-th].
  27. [27]E. Arguello Cruz, I. R. Klebanov, G. Tarnopolsky, and Y. Xin, Yang-Lee Quantum Criticality in Various Dimensions, Phys. Rev. X 16, 011022 (2026), arXiv:2505.06369 [hep-th].arxiv.org/abs/2505.06369
  28. [28]J. Qian, Regulating Free-scalar and Yang–Lee 5d CFTs on the Fuzzy Four Sphere (2026), arXiv:2609.00148 [hep-th].arxiv.org/abs/2609.00148
  29. [29]C. Wetterich, Exact evolution equation for the effective potential, Phys. Lett. B 301, 90 (1993), arXiv:1710.05815 [hep-th].
  30. [30]T. R. Morris, The Exact renormalization group and approximate solutions, Int. J. Mod. Phys. A 9, 2411 (1994), arXiv:hep-ph/9308265.DOI
  31. [31]J. Berges, N. Tetradis, and C. Wetterich, Nonperturbative renormalization flow in quantum field theory and statistical physics, Phys. Rept. 363, 223 (2002), arXiv:hep-ph/0005122.DOI
  32. [32]J. M. Pawlowski, Aspects of the functional renormalisation group, Annals Phys. 322, 2831 (2007), arXiv:hep-th/0512261.DOI
  33. [33]B. Delamotte, An Introduction to the nonperturbative renormalization group, Lect. Notes Phys. 852, 49 (2012), arXiv:cond-mat/0702365.
  34. [34]N. Dupuis, L. Canet, A. Eichhorn, W. Metzner, J. M. Pawlowski, M. Tissier, and N. Wschebor, The nonperturbative functional renormalization group and its applications, Phys. Rept. 910, 1 (2021), arXiv:2006.04853 [cond-mat.stat-mech].DOI
  35. [35]H. Ballhausen, J. Berges, and C. Wetterich, Critical phenomena in continuous dimension, Phys. Lett. B 582, 144 (2004), arXiv:hep-th/0310213.DOI
  36. [36]A. Codello and G. D’Odorico, O(N)-Universality Classes and the Mermin-Wagner Theorem, Phys. Rev. Lett. 110, 141601 (2013), arXiv:1210.4037 [hep-th].DOI
  37. [37]A. Codello, N. Defenu, and G. D’Odorico, Critical exponents of O(N) models in fractional dimensions, Phys. Rev. D 91, 105003 (2015), arXiv:1410.3308 [hep-th].DOI
  38. [38]T. Hellwig, A. Wipf, and O. Zanusso, Scaling and superscaling solutions from the functional renormalization group, Phys. Rev. D 92, 085027 (2015), arXiv:1508.02547 [hep-th].DOI
  39. [39]N. Defenu and A. Codello, Scaling solutions in the derivative expansion, Phys. Rev. D 98, 016013 (2018), arXiv:1711.01809 [hep-th].DOI
  40. [40]S. Yabunaka and B. Delamotte, Surprises in O(N) Models: Nonperturbative Fixed Points, Large N Limits, and Multicriticality, Phys. Rev. Lett. 119, 191602 (2017), arXiv:1707.04383 [cond-mat.stat-mech].DOI
  41. [41]A. Chlebicki and P. Jakubczyk, Analyticity of critical exponents of the O(N) models from nonperturbative renormalization, SciPost Phys. 10, 134 (2021), arXiv:2012.00782 [cond-mat.stat-mech].DOI
  42. [42]R. Ben Ali Zinati and A. Codello, Functional RG approach to the Potts model, J. Stat. Mech. 1801, 013206 (2018), arXiv:1707.03410 [cond-mat.stat-mech].DOI
  43. [43]N. Defenu, A. Trombettoni, and A. Codello, Fixed-point structure and effective fractional dimensionality for O(N) models with long-range interactions, Phys. Rev. E 92, 052113 (2015), arXiv:1409.8322 [cond-mat.stat-mech].DOI
  44. [44]I. Balog, H. Chaté, B. Delamotte, M. Marohnić, and N. Wschebor, Convergence of Nonperturbative Approximations to the Renormalization Group, Phys. Rev. Lett. 123, 240604 (2019), arXiv:1907.01829 [cond-mat.stat-mech].DOI
  45. [45]G. De Polsi, I. Balog, M. Tissier, and N. Wschebor, Precision calculation of critical exponents in the O(N) universality classes with the nonperturbative renormalization group, Phys. Rev. E 101, 042113 (2020), arXiv:2001.07525 [cond-mat.stat-mech].DOI
  46. [46]B. Delamotte, M. Tissier, and N. Wschebor, Scale invariance implies conformal invariance for the three-dimensional Ising model, Phys. Rev. E 93, 012144 (2016), arXiv:1501.01776 [cond-mat.stat-mech].DOI
  47. [47]I. Balog, G. De Polsi, M. Tissier, and N. Wschebor, Conformal invariance of the nonperturbative renormalization group: a rationale for choosing the regulator, Phys. Rev. E 101, 062146 (2020), arXiv:2004.02521 [cond-mat.stat-mech].DOI
  48. [48]A. Baldazzi, R. B. A. Zinati, and K. Falls, Essential renormalisation group, SciPost Phys. 13, 085 (2022), arXiv:2105.11482 [hep-th].
  49. [49]F. Ihssen and J. M. Pawlowski, Flowing fields and optimal RG-flows (2023), arXiv:2305.00816 [hep-th].arxiv.org/abs/2305.00816
  50. [50]F. Ihssen and J. M. Pawlowski, Physics-informed renormalisation group flows, Annals Phys. 481, 170177 (2025), arXiv:2409.13679 [hep-th].arxiv.org/abs/2409.13679
  51. [51]C. Duclut and B. Delamotte, Frequency regulators for the nonperturbative renormalization group: A general study and the model A as a benchmark, Phys. Rev. E 95, 012107 (2017), arXiv:1611.07301 [cond-mat.stat-mech].DOI
  52. [52]J. V. Roth and L. von Smekal, Critical dynamics in a real-time formulation of the functional renormalization group, JHEP 10, 065, arXiv:2303.11817 [hep-ph].
  53. [53]J. V. Roth, Y. Ye, S. Schlichting, and L. von Smekal, Dynamical critical behavior of the chiral phase transition from the real-time functional renormalization group, JHEP 01, 118, arXiv:2403.04573 [hep-ph].arxiv.org/abs/2403.04573
  54. [54]J. Braun et al., Renormalised spectral flows, SciPost Phys. Core 6, 061 (2023), arXiv:2206.10232 [hep-th].arxiv.org/abs/2206.10232
  55. [55]J. Horak, F. Ihssen, J. M. Pawlowski, J. Wessely, and N. Wink, Scalar spectral functions from the spectral functional renormalization group, Phys. Rev. D 110, 056009 (2024), arXiv:2303.16719 [hep-th].
  56. [56]K. Kockler, J. M. Pawlowski, and J. Wessely, Critical scaling for spectral functions, Eur. Phys. J. C 85, 950 (2025), arXiv:2506.09142 [hep-th].arxiv.org/abs/2506.09142
  57. [57]A. Connelly, G. Johnson, F. Rennecke, and V. Skokov, Universal Location of the Yang-Lee Edge Singularity in O(N) Theories, Phys. Rev. Lett. 125, 191602 (2020), arXiv:2006.12541 [cond-mat.stat-mech].DOI
  58. [58]F. Rennecke and V. V. Skokov, Universal location of Yang–Lee edge singularity for a one-component field theory in 1≤d≤ 4, Annals Phys. 444, 169010 (2022), arXiv:2203.16651 [hep-ph].DOI
  59. [59]G. Johnson, F. Rennecke, and V. V. Skokov, Universal location of Yang-Lee edge singularity in classic O(N) universality classes, Phys. Rev. D 107, 116013 (2023), arXiv:2211.00710 [hep-ph].DOI
  60. [60]F. Ihssen and J. M. Pawlowski, Functional flows for complex effective actions, SciPost Phys. 15, 074 (2023), arXiv:2207.10057 [hep-th].DOI
  61. [61]X. An, D. Mesterházy, and M. A. Stephanov, Functional renormalization group approach to the Yang-Lee edge singularity, JHEP 07, 041, arXiv:1605.06039 [hep-th].
  62. [62]L. Zambelli and O. Zanusso, Lee-Yang model from the functional renormalization group, Phys. Rev. D 95, 085001 (2017), arXiv:1612.08739 [hep-th].DOI
  63. [63]D. Benedetti, F. Eustachon, and O. Zanusso, Critical and multicritical Lee-Yang fixed points in the local potential approximation, JHEP 05, 281, arXiv:2601.15087 [hep-th].DOI
  64. [64]P. Beretta, C. A. Sánchez-Villalobos, and A. Codello, Functional Dimensional Regularization for the Lee-Yang Model (2026), arXiv:2601.03318 [hep-th].
  65. [65]P. Beretta and A. Codello, Rethinking dimensional regularization in critical phenomena (2026), arXiv:2604.25103 [hep-th].arxiv.org/abs/2604.25103
  66. [66]J. A. Gracey, (2026), private communication.
  67. [67]L. Canet, B. Delamotte, D. Mouhanna, and J. Vidal, Nonperturbative renormalization group approach to the Ising model: A Derivative expansion at order partial**4, Phys. Rev. B 68, 064421 (2003), arXiv:hep-th/0302227.DOI
  68. [68]D. F. Litim, Optimization of the exact renormalization group, Phys. Lett. B 486, 92 (2000), arXiv:hep-th/0005245.
  69. [69]D. F. Litim, Optimized renormalization group flows, Phys. Rev. D 64, 105007 (2001), arXiv:hep-th/0103195.
  70. [70]L. Canet, B. Delamotte, D. Mouhanna, and J. Vidal, Optimization of the derivative expansion in the nonperturbative renormalization group, Phys. Rev. D 67, 065004 (2003), arXiv:hep-th/0211055.DOI
  71. [71]S. Hikami, Conformal bootstrap analysis for the Yang–Lee edge singularity, PTEP 2018, 053I01 (2018), arXiv:1707.04813 [hep-th].DOI
  72. [72]J. M. Pawlowski and F. Rennecke, Higher order quark-mesonic scattering processes and the phase structure of QCD, Phys. Rev. D 90, 076002 (2014), arXiv:1403.1179 [hep-ph].DOI

Paper details

vlnc:W_ana7zwwn
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents