Pith. Sign in

REVIEW

Critical slowing down of topological modes

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
Luigi Del Debbio, Gian Mario Manca, Ettore Vicari

High Energy Physics - Lattice · Submitted 1 Mar 2004 · 42 references

Pith's verdict: A clean, well-measured demonstration that topological autocorrelation times in CP^(N-1) blow up far faster than the magnetic ones; the functional form is honestly left open, and the QCD extrapolation is heuristic.

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 →

· 2 major objections · 6 minor · Ask about this paper · The compiled record · Read the full review →
Challenge this review Comment (0) Read on arXiv ↗
Cite ▾
arXiv IDhep-lat/0403001 (version 1)
Pith IDBQAHKK7D, Pith's permanent record of this paper and its review
CitationEttore Vicari, Gian Mario Manca, Luigi Del Debbio. "Critical slowing down of topological modes." arXiv:hep-lat/0403001 (2004). https://pith.science/paper/BQAHKK7D
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

In two-dimensional CP^(N-1) models, the topological charge decorrelates far more slowly than the magnetic modes, and the gap widens as the correlation length grows.

The reading

Monte Carlo simulations of lattice field theories become expensive near the continuum limit because successive configurations are correlated; the number of sweeps needed for a fresh, independent configuration is the autocorrelation time $t$, and it grows with the correlation length $xi$. This paper asks whether everything grows at the same rate. Using local updating algorithms (Metropolis, and a mixture of overrelaxed microcanonical, over-heat-bath and Metropolis steps) on two-dimensional $CP^{N-1}$ models with $N = 10, 15, 21$, the authors measure two autocorrelation times side by side: one for the magnetic susceptibility, built from the quasi-Gaussian spin modes, and one for the topological susceptibility, built from the integer-valued winding of the field. The magnetic one behaves as expected, $t_{mag} = c xi^z$ with $z$ slightly above 2. The topological one does not: over the same range of $xi$ it rises by orders of magnitude more, and no single power law with a small exponent fits it. If this is right, the cost of simulating anything that depends on topology is set by a much steeper function of the correlation length than the cost of simulating ordinary observables, which changes how one should budget lattice QCD runs aimed at topology-dependent physics.

What carries the argument

The load-bearing device is a side-by-side measurement of two integrated autocorrelation times as functions of a single control parameter. Both are estimated by the binning method, $t_O = E^2/(2 E_0^2)$, where $E_0$ is the naive error and $E$ the error after blocking, so that the same statistical machinery is applied to both observables and the comparison is internally consistent. The topological side uses the geometrical lattice charge $Q = sum_n (1/2 pi) Im [ln Tr(P_{n+mu+nu} P_{n+mu} P_n) + ln Tr(P_{n+nu} P_{n+mu+nu} P_n)]$, built from the gauge-invariant bilinear $P = zbar otimes z$, which is integer-valued on every configuration and needs no renormalization. The comparison is then made twice: once as a log-log plot of $t_{top}$ against $xi$, where curvature rules out a single small-exponent power law, and once as $N^{-1} log_{10} t_{top}$ against $xi^{-1/2}$, where the data collapse onto straight lines of common slope if the exponential form with $theta = 1/2$ is assumed. The second plot is what makes the large-$N$ scaling of the barrier visible.

Extended reading notes

Core claim

The paper's claim is a comparison of two autocorrelation times measured in the same runs. For the magnetic susceptibility, $t_{mag}$ follows $t_{mag} = c xi^z$ with $z = 2.30(5)$ in the $CP^9$ Metropolis data, consistent with the random-walk picture in which information must propagate across a distance of order $xi$. For the topological susceptibility, $t_{top}$ grows far faster: in the $N = 21$ runs it goes from $10.6(2)$ sweeps at $xi = 1.42$ to $19000(3000)$ sweeps at $xi = 4.21$, and the log-log plot shows curvature rather than a straight line. Two descriptions are offered and neither is excluded by the data. A power law $t_{top} = b_N xi^{c_N N}$ fitted only to the largest-$xi$ points gives $c_{10} = 0.51(2)$, $c_{15} = 0.49(4)$, $c_{21} = 0.41(2)$, i.e. an effective exponent of order $N/2$ or larger, with the fitted exponent creeping upward as more small-$xi$ points are discarded. An exponential $N^{-1} log_{10} t_{top} = a_N + b_N xi^{theta_N}$ fits the whole range with a common $theta = 0.49(2)$, and with $theta$ fixed to $1/2$ the fits are good for all three $N$. The authors read this as evidence for free-energy barriers between topological sectors, tunnelled through by instanton-like configurations whose action scales as $N$, so that $ln t_{top} = O(N)$; the same reasoning is offered as an explanation of the exponential-looking topological autocorrelation times reported in four-dimensional $SU(N)$ gauge theory. Along the way they extract the continuum values $chi_t xi^2 = 0.0175(3)$, $0.0113(2)$ and $0.0080(2)$ for $N = 10, 15, 21$, slightly above the large-$N$ expansion $1/(2 pi N) - 0.060/N^2$.

Load-bearing premise

The measured slow mode is genuine physical tunnelling between topological sectors, which requires the geometrical lattice charge to be free of dislocation contamination at the $N$ and couplings actually simulated, a point the paper supports by citation rather than by a direct check at these parameters.

Editorial extensions

If this is right

  • The autocorrelation time relevant to topological observables should be budgeted as a much steeper function of the correlation length than the factor of $xi$ usually assumed in cost estimates for dynamical-fermion lattice QCD.
  • Physics that depends on topology, such as the mass and matrix elements of the $eta'$ meson and the breaking of the $U(1)_A$ symmetry, inherits this cost penalty directly.
  • Because $N^{-1} log_{10} t_{top}$ approaches a common large-$N$ limit, the barrier picture predicts that the problem does not disappear as $N$ grows; the exponent per colour scales like $N$.
  • The measured ratio $t_Q / t_{top} = 2.3$ (Metropolis) and $2.1$ (overrelaxed) is close to the value 2 that simple Gaussian propagation would give, so the excess slowing is in the susceptibility's long-wavelength topological content rather than in the charge operator's own fluctuations.
  • Multigrid Monte Carlo, which substantially reduces the slowing of the quasi-Gaussian modes, is left as an open question for the topological modes; cluster algorithms are reported not to help in these models.

Reading between the lines

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

  • If the exponential form with $theta = 1/2$ survives at larger $xi$, then the cost of a fixed number of independent topological configurations grows like $exp(c sqrt{xi})$, which is a far milder wall than a naive $exp(c xi)$ reading of the four-dimensional gauge data would suggest; distinguishing the two is a concrete target for the next round of simulations.
  • The instanton-barrier argument predicts that the effect should weaken as $N$ grows at fixed $xi$ in the combination $N^{-1} ln t_{top}$, but not in $ln t_{top}$ itself; a plot of the raw logarithm against $N$ at matched $xi$ would separate the barrier interpretation from a generic slowing of the algorithm.
  • The same side-by-side protocol could be applied to four-dimensional $SU(N)$ runs with a fixed lattice definition of $Q$, which would test whether the apparent $exp(c xi)$ there is the same phenomenon or an artifact of the cooling procedure used to define the charge.
  • If a non-local or topology-aware update were found that restores $t_{top} \sim \xi^2$, that would confirm the barrier picture by showing the slow mode is a genuine tunnelling rate rather than a property of the local algorithm's information propagation.
Original abstract

We investigate the critical slowing down of the topological modes using local updating algorithms in lattice 2-d CP^(N-1) models. We show that the topological modes experience a critical slowing down that is much more severe than the one of the quasi-Gaussian modes relevant to the magnetic susceptibility, which is characterized by $\tau_{\rm mag} \sim \xi^z$ with $z\approx 2$. We argue that this may be a general feature of Monte Carlo simulations of lattice theories with non-trivial topological properties, such as QCD, as also suggested by recent Monte Carlo simulations of 4-d SU(N) lattice gauge theories.

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, 3,065 input tokens, $0.0001 reported.

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.16 · human participants 0.01 · controlled comparison 0.35 · randomized assignment 0.01 · states quantitative result 0.91 · compares with prior results 0.83 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v1. Results and the main claim are underlined; .

We investigate the critical slowing down of the topological modes using local updating algorithms in lattice 2-d CP^(N-1) models. We show that the topological modes experience a critical slowing down that is much more severe than the one of the quasi-Gaussian modes relevant to the magnetic susceptibility, which is characterized by $\tau_{\rm mag} \sim \xi^z$ with $z\approx 2$. We argue that this may be a general feature of Monte Carlo simulations of lattice theories with non-trivial topological properties, such as QCD, as also suggested by recent Monte Carlo simulations of 4-d SU(N) lattice gauge theories.

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 We investigate the critical slowing down of the topological modes using local updating algorithms in lattice 2-d CP^(N-1) models. contribution 0.96 0.90 0.07 asserted 0.66
s1 We show that the topological modes experience a critical slowing down that is much more severe than the one of the quasi-Gaussian modes relevant to the magnetic susceptibility, which is characterized by $\tau_{\rm mag} \sim \xi^z$ with $z\approx 2$. result 1.00 0.92 0.81 asserted 1.00
s2 We argue that this may be a general feature of Monte Carlo simulations of lattice theories with non-trivial topological properties, such as QCD, as also suggested by recent Monte Carlo simulations of 4-d SU(N) lattice gauge theories. interpretation 1.00 0.11 0.19 hedged possibility 1.00

What kind of paper the classifier reports this is: simulation study 1.00 · evidence from simulation or computation 0.99.

What was examined, and what was not

sources read
  • arXiv listing abstract v1
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. Integrated autocorrelation time of the magnetic suscep
Figure 2. Log-log plot of the integrated autocorrelation time
Figure 3. The quantity N −1 log10 τtop versus ξ −1/2 . The lines show the results of the exponential fits τtop ∼ exp(bN ξ θ ) with θ = 1/2; see text. modes under local updating algorithms should be a general feature of Monte Carlo simulations of lattice models with non-trivial topological properties, since the mechanism behind this phenomenon should be similar. This is also supported by recent Monte Carlo simulations of 4-d lattice SU(N) gauge the￾ories reported in Refs. [5, 4]. Indeed, the estimates of autocorrelation time τQ of the topological charge 3 recently reported in Ref. [5] (measu
Figure 4. χtξ 2 versus ξ for N = 10, 15, 21. ical sectors correctly. In particular, it may worsen the current cost estimates of the dynamical fermion simulations for lattice QCD, see e.g. Ref. [38], where it is usually assumed that the autocorrelation time only contributes a factor of ξ. An interesting question would be whether other, possibly non-local, up￾dating algorithms may eliminate or at least improve the severe form of CSD of topological modes. Cluster algorithms turn out not to be effective in CPN−1 models [17, 40]. Instead, as shown in Ref. [39], multigrid Monte Carlo al￾gorithms…

Figures reproduced from arXiv:hep-lat/0403001 by the authors.

References
How later papers used it

59 later papers cite it

Evidence: substantial. 55 of the citing papers share no author with it, and 7 of those carry a confident label from Pith's citation classifier.

Pith read 10 of the 89 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

Compares with it 1 of 3 papers with this label
How the literature uses it

60 arXiv papers cite this in 92 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 · 74 sentences without a confident label

Uses its method or tool 2 papers
  • First, the issue of topology freezing [ref] when approaching the continuum limit is solved by the use of winding transformations [ref] .

    Strong CP problem in the quantum rotor · Introduction · 0.87

  • In this study, we will use the following estimator, which allows to evaluate the integated auto-correlation time through a standard binned jack-knife analysis [ref] :

    Full QCD with milder topological freezing · Zero-temperature results · 0.99

Compares with it 3 papers
Cites it as support 2 papers
  • In particular, finite-sample pathologies have been demonstrated for traditional methods in cases where the sampling algorithm remains trapped in one or several of many modes of a multimodal distribution, such as when “topological freezing” occurs [ref] .

    Flow-based sampling for multimodal and extended-mode distributions in lattice field theory · Introduction · 0.97

  • This has been observed empirically to result in topological freezing in practical simulations, where depending on the gauge and Fermion action simulations with inverse lattice spacings between 3 and 4 GeV have been seen to become practically non-ergodic with frozen topology, and thus demonstrable non-ergodicity [ref] .

    Advances in algorithms for solvers and gauge generation · Topological tunneling · 0.96

Cites it as background 11 papers
  • The fact that gauge configurations relevant to the path-integral divide, in the continuum, in homotopy classes characterized by different integer values of $Q$, is at the origin of the infamous critical slowing down of topological modes in Monte Carlo simulations, also known as topological freezing, when the continuum limit is approached [ref] : the tunneling between different topological sectors becomes more and more suppressed, as it involves the sampling of large action configurations.

    Compact Gauge Fields on Causal Dynamical Triangulations: a 2D case study · Gauge observables · 0.96

  • Topological freezing [ref] constitutes a computational challenge that pervades Monte Carlo simulations of all lattice field theories possessing non-trivial topological properties when employing conventional local algorithms in the vicinity of the continuum limit.

    Scale setting of SU($N$) Yang--Mills theory, topology and large-$N$ volume independence · Introduction · 0.90

  • This algorithm was tested for the CP$^N-1$ model, which suffers from topology freezing [ref] , by Engel and Schaefer [ref] .

    Learning Trivializing Flows · Trivializing flows · 0.87

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.