Pith. Sign in

REVIEW

Critical exponents from optimised renormalisation group flows

PaperPDF from arXiv

data files

ReviewThe objections and verdict, as JSON Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
Daniel F. Litim

High Energy Physics - Theory · Submitted 1 Mar 2002 · 37 references

Pith's verdict: The Gaussian fixed point derivation and the O(N) tables are solid and worth citing; the bounded-ν claim is genuinely conjectural and the abstract states it more firmly than the body supports.

T0 · AI review

How this review was made: An AI referee (selfhosted/deepseek-v4.1-flash) read the paper on November 11, 2026 and answered a fixed list of questions about it. T0 is the starting level every paper gets. The levels describe how deeply Pith has checked the paper, never the authors.

  1. T0 AI review An AI read the paper and answered a fixed list of questions.
  2. T1 Audit Automatic checks that the references and files the paper relies on exist and match.
  3. T2 Compute The paper's code or calculations re-run with the same results.
  4. T3 Formal statement The claims written out in a formal math language a computer can read.
  5. T4 Proof Those claims proved, and the proof checked by computer.

How Pith checks papers →

· 3 major objections · 9 minor · Ask about this paper · The compiled record · Read the full review →
Challenge this review Comment (0) Read on arXiv ↗
Cite ▾
arXiv IDhep-th/0203006 (version 2)
Pith IDEVVJ6VTD, Pith's permanent record of this paper and its review
CitationDaniel F. Litim. "Critical exponents from optimised renormalisation group flows." arXiv:hep-th/0203006 (2002). https://pith.science/paper/EVVJ6VTD
Cite this reviewBibTeX for Pith's review of this paper
Claim paper

Why Pith says this

Highlight any text to comment on it.

Referee report

How Pith checks papers → · Full audit of this paper →

The paper

The paper claims

The critical exponent ν of O(N) scalar theories is trapped between the optimised regulator and the large-N limit, and the same regulator makes crude polynomial truncations converge fast.

The reading

The paper asks a practical question about the exact renormalisation group: when you truncate the flow — here to leading order in the derivative expansion plus a polynomial in the field — how much of your answer is physics and how much is an artefact of the infrared regulator you chose to implement the cutoff? Litim's answer is that the regulator dependence is not arbitrary noise but an ordered interval. For O(N) symmetric scalar theories in three dimensions, the leading critical exponent $ν$ obeys $ν_{opt} ≤ ν_{ERG} ≤ ν_{large-N} = 1$: the lower end is attained by a specific optimised regulator $R_{opt}(q^2) = (k^2 - q^2)θ(k^2 - q^2)$, the upper end by a mass-term (Callan-Symanzik) regulator that reproduces the exact large-$N$ eigenvalues for every $N$. If this holds, a reader who wants a number close to the physical one does not need a fancier truncation, only a better-chosen cutoff — and the paper backs that up by showing that six couplings already reproduce $ν$ at the percent level for the Ising case, where $ν_{opt} = 0.649562$.

What carries the argument

The load-bearing object is the threshold function $ℓ(ω) = ½∫_0^∞ dy y^{d/2} ∂_t r(y) / [y(1+r) + ω]$, which is the only place the choice of regulator enters the leading-order flow $∂_t u + du - (d-2)ρu' = 2v_d(N-1)ℓ(u') + 2v_d ℓ(2ρu''+...)$. Optimisation is formulated as maximising the gap $C = min_{y≥0} y(1+r(y))$ of the regularised inverse propagator, which is what singles out $R_{opt}$ and gives $ℓ_{opt}(ω) = (4/d)(1+ω)^{-1}$ in $d=3$; the exponent $γ$ in $ℓ_γ(ω) ∼ (1+ω)^{1-γ}$ then organises the whole regulator space, with $γ=2$ the best possible decay and $γ=1/2$ the mass-term (Callan-Symanzik) end. Critical exponents come from the eigenvalues of the stability matrix $M_{ij} = ∂β_i/∂λ_j$ evaluated at the fixed-point couplings, which reproduces the spectrum of the linearised eigenperturbation equation without solving it directly.

Extended reading notes

Core claim

Working to leading order in the derivative expansion of the exact renormalisation group for O(N) scalar theories in $d=3$, the paper computes the eigenvalues of the stability matrix at the Wilson-Fisher fixed point for all $N$ up to truncation order 20, and then scans the space of infrared regulators to see how the leading exponent $ν = -1/ω_0$ moves. The central result is that $ν$ is bounded: $ν_{opt} ≤ ν_{ERG} ≤ ν_{large-N} = 1$. The lower boundary is the optimised regulator $R_{opt}(q^2) = (k^2-q^2)θ(k^2-q^2)$, and the upper boundary is a mass-term regulator whose flow is of Callan-Symanzik type and which, for every $N$, yields the exact large-$N$ spectrum $ω_n = 2n-1$. The scan is carried out over four families chosen to probe different directions: $R_γ$, indexed by the decay exponent $γ$ of the threshold function $ℓ(ω) ∼ (1+ω)^{1-γ}$ with $γ ∈ [1/2, 2]$; $R_a$, which scales the optimised regulator by $a$; $R_b$, which flattens it below $q^2 = k^2/2$; and $R_c$, which shifts the cutoff by the running mass. A second, separate claim is that the optimised regulator is what makes polynomial truncations converge: for the Ising case the accuracy of $ν_{trunc}$ improves by one decimal place every $Δn ≈ 2$–$2.5$ orders, and the poor convergence reported earlier for sharp cutoffs is attributed to the cutoff rather than to the polynomial method. The appendix adds a regulator-independent result: at the Gaussian fixed point in $d>2$ the eigenvalues are $ω = (d-2)(n+m) - d$ for any admissible regulator, with Laguerre-polynomial eigenfunctions, so the scheme dependence lives entirely in the argument of those polynomials.

Load-bearing premise

The claim that the optimised regulator gives the smallest possible $ν$ assumes the four scanned regulator families are representative enough that no unexplored admissible regulator produces a smaller value; the paper says plainly that it has no general proof, and it labels the matching upper bound a conjecture.

Editorial extensions

If this is right

  • A regulator-dependence band, rather than a single number, becomes the honest error bar for any ERG computation of $ν$ at this order; for $N=1$ the band spans roughly $0.6496$ to $1$, a spread of about 54 percent.
  • Low-order polynomial truncations become usable for complex theories where higher operators are algebraically out of reach, provided the regulator is chosen by the gap criterion — the paper names quantum gravity and QCD as targets.
  • The sharp cutoff's reputation for producing spurious fixed points and non-convergent expansions is reassigned from the polynomial method to the regulator, which reopens polynomial truncation as a legitimate technique.
  • Because the Callan-Symanzik flow saturates the upper bound at the exact large-$N$ spectrum for every $N$, the large-$N$ limit serves as a calibration point against which any proposed regulator can be checked.
  • At the Gaussian fixed point the eigenvalues are regulator-independent and given in closed form, so any scheme dependence seen at the Wilson-Fisher point is a genuine artefact of the approximation, not of the regularisation per se.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the lower bound is real, it suggests a selection principle that could be exported: pick the regulator that extremises the observable within the allowed band, and the truncation error and the scheme error shrink together rather than trading off.
  • The paper's own structure hints that the bound is really a statement about the analyticity radius of the fixed-point solution in the complex field plane — the gap $C$ controls where the nearest poles sit, and the exponent ordering tracks that radius. A direct study of those poles across the $R_γ$ family would test the bound without scanning regulators one at a time.
  • The observation that $ν$ decreases monotonically as convergence improves suggests a cheap diagnostic for any new truncation scheme: compute $ν$ at low order under two regulators with different gaps and check that the better-converging one gives the smaller value.
  • Since the $R_c$ family shifts the cutoff by the running mass and leaves universal quantities untouched, the same trick might suppress the pole at $1 + u' = -1$ in theories with spontaneous symmetry breaking, which is where the flow is hardest to integrate numerically.
Original abstract

Within the exact renormalisation group, the scaling solutions for O(N) symmetric scalar field theories are studied to leading order in the derivative expansion. The Gaussian fixed point is examined for d>2 dimensions and arbitrary infrared regularisation. The Wilson-Fisher fixed point in d=3 is studied using an optimised flow. We compute critical exponents and subleading corrections-to-scaling to high accuracy from the eigenvalues of the stability matrix at criticality for all N. We establish that the optimisation is responsible for the rapid convergence of the flow and polynomial truncations thereof. The scheme dependence of the leading critical exponent is analysed. For all N > 0, it is found that the leading exponent is bounded. The upper boundary is achieved for a Callan-Symanzik flow and corresponds, for all N, to the large-N limit. The lower boundary is achieved by the optimised flow and is closest to the physical value. We show the reliability of polynomial approximations, even to low orders, if they are accompanied by an appropriate choice for the regulator. Possible applications to other theories are outlined.

The compiled record

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Compiled by classifier-1.13-20260917 via OpenRouter, question bank abstract-p1-20260923, 2026-09-23: 1 request, 7,963 input tokens, $0.0003 reported.

What the abstract itself states, as probabilities: claims first 0.03 · states limitation 0.39 · human participants 0.01 · controlled comparison 0.07 · randomized assignment 0.01 · states quantitative result 0.66 · compares with prior results 0.75 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v2. Results and the main claim are underlined; .

Within the exact renormalisation group, the scaling solutions for O(N) symmetric scalar field theories are studied to leading order in the derivative expansion. The Gaussian fixed point is examined for d>2 dimensions and arbitrary infrared regularisation. The Wilson-Fisher fixed point in d=3 is studied using an optimised flow. We compute critical exponents and subleading corrections-to-scaling to high accuracy from the eigenvalues of the stability matrix at criticality for all N. We establish that the optimisation is responsible for the rapid convergence of the flow and polynomial truncations thereof. The scheme dependence of the leading critical exponent is analysed. For all N > 0, it is found that the leading exponent is bounded. The upper boundary is achieved for a Callan-Symanzik flow and corresponds, for all N, to the large-N limit. The lower boundary is achieved by the optimised flow and is closest to the physical value. We show the reliability of polynomial approximations, even to low orders, if they are accompanied by an appropriate choice for the regulator. Possible applications to other theories are outlined.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#sentencerolethe authors' own workstates the main conclusionstrength
s0 Within the exact renormalisation group, the scaling solutions for O(N) symmetric scalar field theories are studied to leading order in the derivative expansion. contribution 0.75 0.77 0.04 asserted 0.94
s1 The Gaussian fixed point is examined for d>2 dimensions and arbitrary infrared regularisation. contribution 0.87 0.77 0.03 asserted 0.92
s2 The Wilson-Fisher fixed point in d=3 is studied using an optimised flow. contribution 0.81 0.89 0.06 asserted 0.97
s3 We compute critical exponents and subleading corrections-to-scaling to high accuracy from the eigenvalues of the stability matrix at criticality for all N. result 0.65 0.97 0.35 asserted 1.00
s4 We establish that the optimisation is responsible for the rapid convergence of the flow and polynomial truncations thereof. result 0.69 0.86 0.31 asserted 1.00
s5 The scheme dependence of the leading critical exponent is analysed. contribution 0.97 0.86 0.06 asserted 0.83
s6 For all N > 0, it is found that the leading exponent is bounded. result 1.00 0.93 0.16 asserted 1.00
s7 The upper boundary is achieved for a Callan-Symanzik flow and corresponds, for all N, to the large-N limit. result 0.96 0.82 0.14 asserted 1.00
s8 The lower boundary is achieved by the optimised flow and is closest to the physical value. result 0.92 0.88 0.23 asserted 1.00
s9 We show the reliability of polynomial approximations, even to low orders, if they are accompanied by an appropriate choice for the regulator. result 0.82 0.88 0.18 asserted 1.00
s10 Possible applications to other theories are outlined. future work 1.00 0.46 0.04 hedged possibility 0.68

What kind of paper the classifier reports this is: theoretical result 0.88 · evidence from proof or derivation 0.54.

What was examined, and what was not

sources read
  • arXiv listing abstract v2
not examined
  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers
levels run
  • P abstract
not run
  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

Figures
Figure 1. The scaling potential, with cN = 40 − 2N. From left to right: N = 0, 1, · · · , 10
Figure 3. The amplitude u ′ ⋆ + 2ρu′′ ⋆ of the scaling solution. The dots indicate the points where u ′ ⋆ = 0. From left to right: N = 0, 1, · · · , 10. 0 0.1 0.2 0.3 0.4 0 1 2 3 4 5
Figure 4. The exponent ν(N) as a func￾tion of N and of the order of the truncation. From top to bottom: N = 10, 4, 3, 2, 1, 0
Figure 6. The critical index ω2(N) as a func￾tion of N and of the order of the truncation. From top to bottom: N = 0, 1, 2, 3, 4, 10
Figure 8. The critical index ω4(N) as a func￾tion of N and of the order of the truncation. From top to bottom: N = 0, 1, 2, 3, 4, 10. The isolated point at n = 7 corresponds to N = 10, and the two at n = 9 to N = 2 (upper) and N = 3 (lower). The intermediate points (at n = 8 and n = 10, resp.) are missing because they have a small imaginary part. 6 8 10 12 14 16 18 20 7 8 9 10 11 12 13 14 For low orders of the truncation, the eigenvalues ω2, ω3 and ω4 (Figs. 6, 7 and 8, respec￾tively) depend more strongly on ntrunc, and do not even lead to real eigenvalues in some cases (c.f. Figs. 7 an
Figure 9. Ising universality class. Conver￾gence of νtrunc (expansion II) towards νopt with increasing truncation. Points where νtrunc is larger (smaller) than νopt are denoted by o (•). Roughly speaking, for 2 < ntrunc < 20, the ac￾curacy of the critical exponents improves by one decimal point every ∆n ≈ 2 − 2.5. -10 -8 -6 -4 -2 0 5 10 15 20 Let us have a closer look at the rate of convergence towards the asymptotic values of expansion II (cf
Figure 10. The critical index ν for the Ising universality class as a function of the order of the truncation ntrunc, for expansion I around vanishing field and expansion II around the non-trivial minimum. Left panel: I and II converge towards the asymptotic value. Right panel: Magnification by a factor of 625. The variation of all data points is below 10−3 . II converges faster than I. Expansion I (II) fluctuates about the asymptotic value with decreasing amplitude and a six-fold (four-fold) periodicity; see also
Figure 11. The critical index ν for the Ising universality class. Results are given for Expansion I (left panel) and II (right panel), and for the sharp cutoff (upper curves) the quartic regulator (middle curves) and the optimised regulator (lower curves). In [7], it has been shown that the gap C (the radius of convergence for amplitude ex￾pansions) is linked to the radius of convergence C ′ for the expansions (3.1) and (3.2). An alternative way for identifying C ′ consists in studying the complex structure of the scaling solution. For real ρ, the scaling solution is finite and real for all ρ…
Figure 12. (for more details, see Appendix B and Tab. 3). We find that νa is a monotonously increasing function for decreasing a ≤ 1. In particular, for small a we find indeed that νa > νsharp. Hence, this result confirms the above picture: Flows with a poor radius of convergence lead to large numerical values for ν. γ 1 2 3 4 4 5 5 6 1 3 2 2 νγ 1 .7354 .7216 .7142 .6895 .6604 .6496
Figure 14. The function y(1+ra) for the class of regulators (B.1). All lines for different a co￾incide for large momenta y ∈ [1,∞] (full line), but differ for small momenta y ∈ [0, 1[ (dashed lines). The dashed lines clock-wise from the bottom: a = 1 2 , 1, 2 and ∞. The short dashed 0 0.5 1 1.5 2 line corresponds to a = 0 (no regulator). 0 0.5 1 1.5 2 1. Definition In this appendix, we discuss two variants of the optimised regulator (2.11). First, we consider the class of regulators given by Ra(q 2 ) = a (k 2 − q 2 ) θ(k 2 − q 2 ). (B.1) These regulators have a compact support. They vanish…
Figure 15. The function y(1+rb) for the class of regulators (B.3). All lines for different b co￾incide for large momenta y ∈ [ 1 2 ,∞] (full line), but differ for small momenta y ∈ [0, 1 2 ] (dashed lines). The dashed lines clock-wise from the bottom: b = 0, 1 2 , 1, 2 and ∞. The standard sharp cutoff regulator (dotted line) is given for comparison. 0 0.5 1 1.5 2 0 0.5 1 1.5 2 The regulators Ra and Rb have a similar low momentum limit for q 2 → 0, e.g. Ra(q 2 → 0) = Rb=a(q 2 → 0) = a k2 . The crucial difference between them concerns the intermediate momentum regime q 2 ≈ k 2

Showing 11 of the paper's 15 figures. See all figures in the PDF ↗ Figures reproduced from arXiv:hep-th/0203006 by the authors.

References
How later papers used it

69 later papers cite it

Evidence: substantial. 59 of the citing papers share no author with it, and 27 of those carry a confident label from Pith's citation classifier.

Pith read 17 of the 126 citing sentences, choosing only those whose citation markers point at this paper alone, and wrote the paraphrases from them. The summary, the counts, the evidence level, the links and the quotes come from code; the summary counts the papers listed below. A later paper is listed under a label when Pith's citation classifier gives one of its sentences that label at probability 0.8 or more (0.95 for support), Pith's reading finds that the citing paper itself did it, and a second check confirms the paraphrase against the sentence. This section lists uses only. It does not list criticism or disagreement, so their absence here says nothing about how the paper has been received. Papers that share an author with this one are listed apart, under "By the same authors". A quote is the citing authors' own sentence, shown when their paper's licence is CC BY, CC BY-SA or CC0. JSON

Uses its method or tool 5 of 27 papers with this label
How the literature uses it

57 arXiv papers cite this in 107 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Papers citing it · 60 sentences without a confident label

Uses its method or tool 22 papers
  • Using the common optimized regulator function [ref] [equation] where $θ$ is the Heaviside function, we can perform the trace in the momentum space of $U(1)$ representations and find for the effective potential $U_k(ρ)=μ_kρ+V_k(ρ)$ [ref] [equation] where the particular threshold functions $s^s$ in the dimensionless variable $= k$ result from an integral approximation of the $s$-dimensional traces with quadratic cutoff [We have also explicitly calculated the trace as a sum with simplex and box cutoffs taking into account subleading orders in $$ leading to the same qualitative results [ref] .], thus including the usual unit volume $s=π^s/2/Γ(1+s/2)$.

    (No) phase transition in tensorial group field theory · Functional renormalization group in the melonic potential approximation · 0.98

  • Explicitly, [equation] In the present calculation we are going to use the optimised regulator [equation] following [ref] .

    Critical $O(N)$ models in the complex field plane · Derivative expansion · 0.99

  • For convenience we choose the optimized cutoff $R_k(z)=(k^2-z)θ(k^2-z)$ [ref] , for which one can easily calculate Q_n (Ṙ_k+ηR_k/(P_k +q)^l) = k^2(n-l+1)/Γ(n+1)2+η/n+1/(1 +q̃)^l

    Functional renormalization of N scalars with O(N) invariance · Calculation of beta functions · 0.99

Compares with it 4 papers
Cites it as support 1 paper
Cites it as background 8 papers

Discussion

Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.