Pith. sign in

REVIEW 2 major objections 4 minor 59 references

A hybrid algorithm unfreezes lattice QCD simulations that a standard method leaves completely stuck for roughly 40,000 molecular dynamics units.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 07:02 UTC pith:52EUJWVT

load-bearing objection First credible demonstration of PT-MetaD unfreezing full QCD with dynamical fermions; the quantitative numbers need an independent benchmark, but the core time-series evidence is solid. the 2 major comments →

arxiv 2607.21575 v1 pith:52EUJWVT submitted 2026-07-23 hep-lat

Parallel Tempered Metadynamics for full QCD

Timo Eichhorn , Gianluca Fuwa , Christian Hoelbling , Lukas Varnhorst This is my paper
classification hep-lat
keywords topological freezinglattice QCDparallel temperingmetadynamicstopological chargecollective variableenhanced samplingRHMC
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper aims to show that combining parallel tempering with metadynamics—using a bias on a single collective variable, the stout-smeared topological charge—can overcome topological freezing in full QCD with dynamical fermions. At a coupling where conventional RHMC remains trapped in the zero topological sector for 39,600 MDUs, the PT-MetaD algorithm repeatedly crosses between sectors and produces a nonzero topological susceptibility estimate. Even at a coarser coupling where RHMC already tunnels, PT-MetaD reduces the autocorrelation time of the topological susceptibility by about a factor of 2.6, with a CPU efficiency gain of roughly 3.0 (or 1.8 including the one-time bias build-up). If correct, this offers a practical path to reliable sampling of topological observables in lattice QCD at fine lattice spacings, where freezing is a major systematic obstacle.

Core claim

The central discovery is that biasing a single collective variable—the clover-based topological charge after six stout smearing steps—while running a parallel tempered replica with that bias, keeps the physical replica's distribution unchanged while substantially increasing the rate of topological sector transitions. In the fully frozen beta=1.15 system, PT-MetaD yields a nonzero topological susceptibility estimate with an integrated autocorrelation time of 180(76) MDUs, compared with no transitions at all in 39,600 MDUs for RHMC. The authors also show that the swap acceptance reduces to a simple function of the bias potential, so the overhead of the method is small relative to the sampling

What carries the argument

The central object is the collective variable Q_CV, the clover-based topological charge after six stout smearing steps at the fermion smearing parameter rho=0.125, which serves as the reaction coordinate for topological sector changes. A bias potential V(Q_CV) is constructed from a well-tempered metadynamics buildup, then filtered by singular spectrum analysis to retain only the periodic barriers between sectors. The second replica samples with this bias, and swap acceptances between replicas depend only on the bias values at the two configurations, not on the full action, making swaps cheap. This combination lets the biased replica cross barriers and the swaps propagate mixing to the physic

Load-bearing premise

The load-bearing premise is that the single collective variable Q_CV—the clover-based topological charge after six stout smearing steps—captures the slow direction that blocks topological sector changes, so that biasing it alone can drive mixing; if a hidden slow mode orthogonal to Q_CV exists, the demonstrated gains may not generalize, and the paper itself leaves open whether the smearing parameter of Q_CV needs case-by-case adjustment.

What would settle it

The central claim would be falsified if, at beta=1.15, a second PT-MetaD run using the same bias potential but a different seed fails to sample multiple sectors within the same 39,600 MDUs, or if the measured topological susceptibility is inconsistent with the value 0.127(33) when longer runs are performed. A more targeted test: add a second collective variable (e.g., topological charge at a different smearing level); if mixing does not improve, the original single CV was likely sufficient, but if it improves significantly, the single-CV assumption was incomplete.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • At beta=1.15, PT-MetaD samples multiple topological sectors and gives a nonzero topological susceptibility (chi_top V = 0.127(33)), while RHMC remains in the zero sector for 39,600 MDUs, making chi_top measurable where it was previously unattainable.
  • The swap acceptance depends only on the bias potential values, so the biased replica adds modest CPU cost (~2.3x for two replicas) and the algorithm is still more efficient than RHMC by a factor of about 3.0 (or 1.82 including the one-time bias buildup) for the topological susceptibility on the coarse lattice.
  • Because the physical replica is unbiased, standard measurements such as the smeared plaquette remain valid; at beta=1.15 the PT-MetaD plaquette differs from the RHMC value by almost 10 standard deviations, indicating the RHMC ensemble was stuck in a non-representative sector.
  • The method extends naturally to more replicas and multiple walkers, which are trivially parallelizable, offering a path to further efficiency gains.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the single-CV assumption holds, the same Q_CV with rho=0.125 may work at finer lattice spacings without retuning; the paper's stated open question—whether the smearing parameter needs case-by-case adjustment—is directly testable by scaling runs.
  • The success of a single topological-charge CV suggests that topological freezing in full QCD is dominated by a one-dimensional barrier in the charge coordinate, at least for the DBW2 action; if so, even simpler algorithms (e.g., direct biasing without replica exchange) might suffice, though parallel tempering here provides exactness insurance.
  • The bias-potential construction via singular spectrum analysis to isolate periodic barriers is a generic recipe that could be ported to other observables with metastability, such as the Polyakov loop in deconfined phases or the chiral condensate in first-order transitions.
  • The close match between predicted and observed swap rates (64.3% vs 68.6% at beta=0.95; 31.0% vs 28.6% at beta=1.15) gives a quantitative diagnostic of free-energy estimate accuracy; future applications could monitor this to decide when the bias is good enough.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper presents PT-MetaD, a hybrid of parallel tempering and metadynamics, and applies it to N_f=2 staggered QCD on a 24^4 lattice with the DBW2 action. One replica is the physical (unbiased) distribution; a second replica is biased by a static potential V(Q_CV) constructed from well-tempered metadynamics and singular spectrum analysis. Replica swaps are accepted according to Eq. (9). The central result is that at beta=1.15 the RHMC algorithm remains in the Q=0 sector for roughly 39,600 MDUs, while PT-MetaD repeatedly visits different topological sectors. At beta=0.95, where RHMC already tunnels, PT-MetaD reproduces the RHMC values of the smeared plaquette and topological susceptibility. The paper also reports chi_top V=0.127(33) at beta=1.15, a ~10-sigma shift in the smeared plaquette relative to the frozen RHMC run, and estimates an efficiency gain of about a factor of three despite the extra replica.

Significance. If the beta=1.15 results are correct, this is the first demonstration that a practical algorithm can overcome complete topological freezing in a full-QCD simulation, extending earlier pure-gauge and toy-model studies. The paper contains two genuinely useful validation ingredients: a crosscheck at beta=0.95 and an analytical swap-rate model (Appendix B) that predicts 64.3% and 31.0% swap rates, in reasonable agreement with observed 68.6% and 28.6%. The construction also has the standard detailed-balance protection for the physical replica. However, the main quantitative payload at the frozen coupling is not benchmarked against any independent unbiased estimator, so the significance is real but conditional on ruling out strong-bias implementation artifacts and CV insufficiency.

major comments (2)
  1. [Results (Table II, Fig. 2)] The central quantitative results at beta=1.15 — chi_top V=0.127(33) and the ~10-sigma plaquette shift — have no independent unbiased reference. The beta=0.95 agreement in Table II is a valuable control, but there the bias potential is small and RHMC already tunnels. At beta=1.15, V is O(10) and the biased replica is doing essentially all the sector mixing; an implementation error or an insufficient CV in this strong-bias regime would be invisible in the beta=0.95 comparison. I ask for at least one internal cross-check: (i) compare the physical-replica Q distribution with the biased-replica distribution reweighted by exp(+V); (ii) report the PT-MetaD plaquette conditioned on Q=0 and compare it with the RHMC Q=0 value; or (iii) repeat the beta=1.15 run with a substantially different bias potential or CV smearing and require agreement. Without such a check, the time-series evidence for tunn
  2. [Appendix A, Eq. (A1)] The free-energy estimate that defines V is built using an admitted biased accept/reject scheme: measurements are discarded upon trajectory rejection, and the induced bias is only stated to be 'negligible'. The production-stage physical replica is indeed protected from an approximate V, but the swap-rate prediction in Appendix B and the reported efficiency gains depend on the accuracy of this FES. Please quantify the bias by reporting buildup acceptance rates, state the SSA window length and number of retained components, and show robustness of V to e.g. using only half of the buildup data. Without these details, the validation provided by the swap-rate agreement cannot be fully assessed.
minor comments (4)
  1. [Abstract / Conclusion] The abstract says the algorithm 'unfreezes full QCD', while the Conclusion explicitly states that whether the method works without adjusting the collective-variable smearing 'remains to be seen'. This caveat should be reflected in the abstract or the framing should be softened to an existence proof at a particular action, volume, and CV.
  2. [Table III] The notation Q1 and Q2 is not defined until later in the text; please define them in the caption. Also clarify the dash in the beta=1.15 RHMC row: is tau_int(chi_top) undefined because the observable is constant, and if so, state this explicitly.
  3. [Simulation setup] Please state explicitly that the bias potential is static during the production runs and that the metadynamics deposition is turned off after the buildup. The reader currently has to infer this from the phrase 'final bias potentials'; it is an important correctness detail.
  4. [Appendix C] Figure 5 shows that the bias force can be orders of magnitude larger and more localized than the gauge and fermionic forces. Please report the acceptance rate of the biased replica, since large localized forces can cause frequent rejections that affect the practical efficiency and the mixing of the biased replica.

Circularity Check

0 steps flagged

No significant circularity: bias potentials are efficiency tools and the physical replica remains unbiased; swap-rate checks are out-of-loop consistency tests.

full rationale

The derivation is self-contained. The bias potential V(Q_CV) is fitted to a well-tempered metadynamics free-energy estimate (Appendix A), but it enters only through the biased replica's action; Eq. (6) and Eq. (9) define the two-replica distribution with p1 as the unmodified physical distribution, and the paper explicitly notes that an inexact free-energy estimate 'does not affect the correctness of our algorithm, only its efficiency' (Appendix A). Therefore the central claims—sector crossing and chi_top V at beta=1.15—are measurements from the unbiased physical replica, not predictions derived from the fitted bias. The beta=0.95 RHMC/PT-MetaD agreement is an independent crosscheck at a coupling where RHMC already moves. Appendix B's expected swap rates use the estimated p(Q) and pV(Q); the comparison with observed swap rates is a self-consistency test of the free-energy estimate, not a fitted observable relabeled as a prediction, and the small disagreement is honestly attributed to FES inaccuracy. Self-citations to [5] concern the method's prior demonstration and the SSA-based efficiency improvement, but the present unfreezing evidence is new data in this paper; no load-bearing step reduces to a self-citation. The absence of an independent unbiased benchmark at beta=1.15 is a validation limitation, not circularity.

Axiom & Free-Parameter Ledger

7 free parameters · 5 axioms · 0 invented entities

The paper introduces no new physical entities: the bias potential is a computational construction whose only role is to enhance mixing in the auxiliary replica. The fitted objects (bias potentials, gamma, discretization choices) affect efficiency and swap rates, not the unbiased measurements taken on the physical replica, which keeps the circularity burden low. The free-parameter entry 'metadynamics deposition parameters' and 'SSA window' are flagged as unstated choices that degrade reproducibility rather than soundness.

free parameters (7)
  • Bias potential V(Q_CV), beta=0.95 = O(2-3) barrier height (Fig. 1)
    Built from 12,000 well-tempered metadynamics trajectories (gamma=1.1, 4 walkers) plus SSA filtering. Affects swap rates and efficiency only; physical-replica measurements are unbiased regardless.
  • Bias potential V(Q_CV), beta=1.15 = O(10) barrier height (Fig. 1)
    Same construction with 16,000 buildup trajectories. The headline chi_top measurement does not depend on it, but the demonstrated tunneling rate does.
  • Well-tempered parameter gamma = 1.1
    Chosen for the FES buildup (Appendix A); controls the bias deposition schedule.
  • Histogram bin width / reweighting Gaussian variance = 0.01 / 0.02
    Discretization choices in the FES reconstruction (Eq. A1-A2).
  • Metadynamics deposition parameters = not stated
    Gaussian height and deposition stride during the buildup are never given; only the reweighting width is specified.
  • SSA window length / retained components = not stated
    The singular spectrum analysis that shapes the final bias potentials is not specified, so the bias construction is not fully reproducible.
  • Replica count and CV smearing depth = 2 replicas; 6 stout smears
    Hand choices; the conclusion admits the smearing parameter may need adjustment per application.
axioms (5)
  • standard math Detailed balance of PT swaps (Eqs. 2-3) with static p_i implies the physical replica samples p1 exactly.
    Standard parallel-tempering theorem; requires that the bias potential is held fixed during production.
  • domain assumption Well-tempered metadynamics with Tiwary reweighting (Eqs. A1-A2) recovers the FES of Q_CV.
    Assumes sufficient sampling of the CV range during buildup; indirectly validated by the swap-rate agreement in Appendix B.
  • ad hoc to paper Discarding measurements of rejected trajectories biases the FES estimate only negligibly.
    Admitted by the authors in Appendix A ('This introduces a bias in the estimate of the free energy surface'); argued negligible at stated acceptance rates. Affects efficiency only.
  • domain assumption Q_CV (clover charge after 6 smears) is a sufficient reaction coordinate for sector transitions.
    The central bet of the method; empirically supported at these two couplings, untested for other actions and volumes (Conclusion).
  • domain assumption Rounded Q after 30 stout smears equals the topological charge for susceptibility purposes.
    Standard practice (cf. [32]); no continuum-limit check is reported here.

pith-pipeline@v1.3.0-alltime-deepseek · 9291 in / 21573 out tokens · 193229 ms · 2026-08-01T07:02:42.724451+00:00 · methodology

0 comments
read the original abstract

We present an algorithm that addresses topological freezing in lattice QCD simulations by combining parallel tempering with collective-variable-based enhanced sampling methods, and apply it to a particularly challenging system with $N_f = 2$ staggered fermions. We find that the algorithm unfreezes the system, which is otherwise completely frozen for approximately 40000 Molecular Dynamics Units with the Rational Hybrid Monte Carlo algorithm.

Figures

Figures reproduced from arXiv: 2607.21575 by Christian Hoelbling, Gianluca Fuwa, Lukas Varnhorst, Timo Eichhorn.

Figure 3
Figure 3. Figure 3: FIG. 3. Time series of the topological charge at [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Time series of the topological charge at [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. The estimated free energy surfaces obtained via [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Time series of the average and maximum RHMC [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references · 1 canonical work pages

  1. [1]

    Aokiet al.(Flavour Lattice Averaging Group (FLAG)), FLAG review 2024, Phys

    Y. Aokiet al.(Flavour Lattice Averaging Group (FLAG)), FLAG review 2024, Phys. Rev. D113, 014508 (2026), arXiv:2411.04268 [hep-lat]

    Pith/arXiv arXiv 2024

  2. [2]

    Del Debbio, G

    L. Del Debbio, G. M. Manca, and E. Vicari, Critical slow- ing down of topological modes, Phys. Lett. B594, 315 (2004), arXiv:hep-lat/0403001

    Pith/arXiv arXiv 2004

  3. [3]

    Schaefer, R

    S. Schaefer, R. Sommer, and F. Virotta, Investigat- ing the critical slowing down of QCD simulations, PoS LA T2009, 032 (2010), arXiv:0910.1465 [hep-lat]

    Pith/arXiv arXiv 2010

  4. [4]

    Schaefer, R

    S. Schaefer, R. Sommer, and F. Virotta (ALPHA Col- laboration), Critical slowing down and error analysis in lattice QCD simulations, Nucl. Phys. B845, 93 (2011), arXiv:1009.5228 [hep-lat]

    Pith/arXiv arXiv 2011

  5. [5]

    Eichhorn, G

    T. Eichhorn, G. Fuwa, C. Hoelbling, and L. Varnhorst, Parallel tempered metadynamics: Overcoming potential barriers without surfing or tunneling, Phys. Rev. D109, 114504 (2024), arXiv:2307.04742 [hep-lat]

    Pith/arXiv arXiv 2024

  6. [6]

    Eichhorn, G

    T. Eichhorn, G. Fuwa, C. Hoelbling, and L. Varnhorst, Parallel Tempered Metadynamics, PoSLA TTICE2024, 61 (2025), arXiv:2503.09747 [hep-lat]

    Pith/arXiv arXiv 2025

  7. [7]

    Hoek, Long-time-scale correlations in topological quantities, Phys

    J. Hoek, Long-time-scale correlations in topological quantities, Phys. Lett. B186, 375 (1987)

    1987

  8. [8]

    Alles, G

    B. Alles, G. Boyd, M. D’Elia, A. Di Giacomo, and E. Vicari, Hybrid Monte Carlo and topological modes of full QCD, Phys. Lett. B389, 107 (1996), arXiv:hep- lat/9607049

    arXiv 1996

  9. [9]

    Luscher and S

    M. Luscher and S. Schaefer, Lattice QCD without topol- ogy barriers, JHEP07, 036, arXiv:1105.4749 [hep-lat]

    Pith/arXiv arXiv

  10. [10]

    A. Laio, G. Martinelli, and F. Sanfilippo, Metadynamics surfing on topology barriers: theCP N−1 case, JHEP07, 089, arXiv:1508.07270 [hep-lat]

    Pith/arXiv arXiv

  11. [11]

    Hasenbusch, Fighting topological freezing in the two- dimensionalCP N−1 model, Phys

    M. Hasenbusch, Fighting topological freezing in the two- dimensionalCP N−1 model, Phys. Rev. D96, 054504 (2017), arXiv:1706.04443 [hep-lat]

    Pith/arXiv arXiv 2017

  12. [12]

    M. S. Albergo, G. Kanwar, and P. E. Shanahan, Flow- based generative models for Markov chain Monte Carlo in lattice field theory, Phys. Rev. D100, 034515 (2019), arXiv:1904.12072 [hep-lat]

    Pith/arXiv arXiv 2019

  13. [13]

    Kanwar, M

    G. Kanwar, M. S. Albergo, D. Boyda, K. Cranmer, D. C. Hackett, S. Racani` ere, D. J. Rezende, and P. E. Shanahan, Equivariant flow-based sampling for lattice gauge theory, Phys. Rev. Lett.125, 121601 (2020), arXiv:2003.06413 [hep-lat]

    Pith/arXiv arXiv 2020

  14. [14]

    Cossu, D

    G. Cossu, D. Lancaster, B. Lucini, R. Pellegrini, and A. Rago, Ergodic sampling of the topological charge us- ing the density of states, Eur. Phys. J. C81, 375 (2021), arXiv:2102.03630 [hep-lat]

    Pith/arXiv arXiv 2021

  15. [15]

    Albandea, P

    D. Albandea, P. Hern´ andez, A. Ramos, and F. Romero- L´ opez, Topological sampling through windings, Eur. Phys. J. C81, 873 (2021), [Erratum: Eur.Phys.J.C 83, 508 (2023)], arXiv:2106.14234 [hep-lat]

    Pith/arXiv arXiv 2021

  16. [16]

    Bonanno, A

    C. Bonanno, A. Nada, and D. Vadacchino, Mitigating topological freezing using out-of-equilibrium simulations, JHEP04, 126, arXiv:2402.06561 [hep-lat]

    Pith/arXiv arXiv

  17. [17]

    Bonanno, G

    C. Bonanno, G. Clemente, M. D’Elia, L. Maio, and L. Parente, Full QCD with milder topological freezing, JHEP08, 236, arXiv:2404.14151 [hep-lat]

    Pith/arXiv arXiv

  18. [18]

    Q. Zhu, G. Aarts, W. Wang, K. Zhou, and L. Wang, Physics-conditioned diffusion models for lattice gauge theory, JHEP03, 111, arXiv:2502.05504 [hep-lat]

    arXiv

  19. [19]

    Bonanno, A

    C. Bonanno, A. Bulgarelli, E. Cellini, A. Nada, D. Pan- falone, D. Vadacchino, and L. Verzichelli, Scaling flow- based approaches for topology sampling in SU(3) gauge theory, JHEP04, 051, arXiv:2510.25704 [hep-lat]

    Pith/arXiv arXiv

  20. [20]

    Finkenrath, Review on Algorithms for dynam- ical fermions, PoSLA TTICE2022, 227 (2023), arXiv:2402.11704 [hep-lat]

    J. Finkenrath, Review on Algorithms for dynam- ical fermions, PoSLA TTICE2022, 227 (2023), arXiv:2402.11704 [hep-lat]

    Pith/arXiv arXiv 2023

  21. [21]

    Kanwar, Flow-based sampling for lattice field theo- ries, PoSLA TTICE2023, 114 (2024), arXiv:2401.01297 [hep-lat]

    G. Kanwar, Flow-based sampling for lattice field theo- ries, PoSLA TTICE2023, 114 (2024), arXiv:2401.01297 [hep-lat]

    Pith/arXiv arXiv 2024

  22. [22]

    P. A. Boyle, Advances in algorithms for solvers and 5 gauge generation, PoSLA TTICE2023, 122 (2024), arXiv:2401.16620 [hep-lat]

    Pith/arXiv arXiv 2024

  23. [23]

    Finkenrath, Future trends in lattice QCD simulations, PoSEuroPLEx2023, 009 (2024)

    J. Finkenrath, Future trends in lattice QCD simulations, PoSEuroPLEx2023, 009 (2024)

    2024

  24. [24]

    Duane, A

    S. Duane, A. D. Kennedy, B. J. Pendleton, and D. Roweth, Hybrid Monte Carlo, Phys. Lett. B195, 216 (1987)

    1987

  25. [25]

    Multi- canonical methods and lattice field theory

    with collective-variable-based enhanced sampling, in particular Metadynamics [26]. It is applicable to systems that can be characterized by a probability distribution in one or more collective variables (CVs) with several high- probability regions separated by low probability regions. In parallel tempering, we considernreplicas of a sys- tem with identica...

    Pith/arXiv arXiv 2026

  26. [26]

    R. H. Swendsen and J.-S. Wang, Replica monte carlo sim- ulation of spin-glasses, Phys. Rev. Lett.57, 2607 (1986)

    1986

  27. [27]

    Laio and M

    A. Laio and M. Parrinello, Escaping free-energy minima, Proc. Natl. Acad. Sci.99, 12562 (2002)

    2002

  28. [28]

    D. A. Kofke, On the acceptance probability of replica- exchange monte carlo trials, J. Chem. Phys.117, 6911 (2002), [Erratum: J. Chem. Phys. 117, 6911 (2002)]

    2002

  29. [29]

    Y. F. Atchad´ e, G. O. Roberts, and J. S. Rosenthal, Towards optimal scaling of metropolis-coupled markov chain monte carlo, Statistics and Computing21, 555 (2011)

    2011

  30. [30]

    Barducci, G

    A. Barducci, G. Bussi, and M. Parrinello, Well-tempered metadynamics: A smoothly converging and tunable free- energy method, Phys. Rev. Lett.100, 020603 (2008)

    2008

  31. [31]

    Valsson and M

    O. Valsson and M. Parrinello, Variational approach to enhanced sampling and free energy calculations, Phys. Rev. Lett.113, 090601 (2014)

    2014

  32. [32]

    Invernizzi and M

    M. Invernizzi and M. Parrinello, Rethinking metadynam- ics: From bias potentials to probability distributions, J. Phys. Chem. Lett.11, 2731–2736 (2020)

    2020

  33. [33]

    Bonati, The topological properties of QCD at high temperature: problems and perspectives, EPJ Web Conf

    C. Bonati, The topological properties of QCD at high temperature: problems and perspectives, EPJ Web Conf. 175, 01011 (2018), arXiv:1710.06410 [hep-lat]

    Pith/arXiv arXiv 2018

  34. [34]

    Eichhorn, G

    T. Eichhorn, G. Fuwa, C. Hoelbling, and L. Varnhorst, Enhanced Sampling Techniques for Lattice Gauge The- ory, PoSLA TTICE2025, 257 (2026), arXiv:2604.01287 [hep-lat]

    Pith/arXiv arXiv 2026

  35. [35]

    J. B. Kogut and L. Susskind, Hamiltonian Formulation of Wilson’s Lattice Gauge Theories, Phys. Rev. D11, 395 (1975)

    1975

  36. [36]

    Banks, S

    T. Banks, S. Raby, L. Susskind, J. B. Kogut, D. R. T. Jones, P. N. Scharbach, and D. K. Sinclair (Cornell- Oxford-Tel Aviv-Yeshiva), Strong Coupling Calculations of the Hadron Spectrum of Quantum Chromodynamics, Phys. Rev. D15, 1111 (1977)

    1977

  37. [37]

    Susskind, Lattice Fermions, Phys

    L. Susskind, Lattice Fermions, Phys. Rev. D16, 3031 (1977)

    1977

  38. [38]

    Morningstar and M

    C. Morningstar and M. J. Peardon, Analytic smearing of SU(3) link variables in lattice QCD, Phys. Rev. D69, 054501 (2004), arXiv:hep-lat/0311018

    Pith/arXiv arXiv 2004

  39. [39]

    Takaishi, Heavy quark potential and effective actions on blocked configurations, Phys

    T. Takaishi, Heavy quark potential and effective actions on blocked configurations, Phys. Rev. D54, 1050 (1996)

    1996

  40. [40]

    Orginos (RBC Collaboration), Chiral properties of domain wall fermions with improved gauge actions, Nucl

    K. Orginos (RBC Collaboration), Chiral properties of domain wall fermions with improved gauge actions, Nucl. Phys. B Proc. Suppl.106, 721 (2002), arXiv:hep- lat/0110074

    arXiv 2002

  41. [41]

    T. A. DeGrand, A. Hasenfratz, and T. G. Kovacs, Im- proving the chiral properties of lattice fermions, Phys. Rev. D67, 054501 (2003), arXiv:hep-lat/0211006

    Pith/arXiv arXiv 2003

  42. [42]

    Noaki (RBC Collaboration), Calculation of weak ma- trix elements in domain wall QCD with the DBW2 gauge action, Nucl

    J. Noaki (RBC Collaboration), Calculation of weak ma- trix elements in domain wall QCD with the DBW2 gauge action, Nucl. Phys. B Proc. Suppl.119, 362 (2003), arXiv:hep-lat/0211013

    Pith/arXiv arXiv 2003

  43. [43]

    Aokiet al., The Kaon B-parameter from quenched domain-wall QCD, Phys

    Y. Aokiet al., The Kaon B-parameter from quenched domain-wall QCD, Phys. Rev. D73, 094507 (2006), arXiv:hep-lat/0508011

    Pith/arXiv arXiv 2006

  44. [44]

    Borsanyiet al., Leading hadronic contribution to the muon magnetic moment from lattice QCD, Nature593, 51 (2021), arXiv:2002.12347 [hep-lat]

    S. Borsanyiet al., Leading hadronic contribution to the muon magnetic moment from lattice QCD, Nature593, 51 (2021), arXiv:2002.12347 [hep-lat]

    Pith/arXiv arXiv 2021

  45. [45]

    Borsanyi, Z

    S. Borsanyi, Z. Fodor, M. Giordano, J. N. Guenther, S. D. Katz, A. Pasztor, and C. H. Wong, Can rooted staggered fermions describe nonzero baryon density at low temperatures?, Phys. Rev. D109, 054509 (2024), arXiv:2308.06105 [hep-lat]

    Pith/arXiv arXiv 2024

  46. [46]

    Borsanyi, Z

    S. Borsanyi, Z. Fodor, J. N. Guenther, S. D. Katz, P. Parotto, A. Pasztor, D. Pesznyak, K. K. Szabo, and C. H. Wong, Continuum-extrapolated high-order baryon fluctuations, Phys. Rev. D110, L011501 (2024), arXiv:2312.07528 [hep-lat]

    Pith/arXiv arXiv 2024

  47. [47]

    Borsanyi, Z

    S. Borsanyi, Z. Fodor, J. N. Guenther, P. Parotto, A. Pasztor, L. Pirelli, K. K. Szabo, and C. H. Wong, QCD deconfinement transition line up toµB=400 MeV from finite volume lattice simulations, Phys. Rev. D110, 114507 (2024), arXiv:2410.06216 [hep-lat]

    Pith/arXiv arXiv 2024

  48. [48]

    Bors´ anyi, Z

    S. Bors´ anyi, Z. Fodor, J. N. Guenther, P. Kumar, P. Parotto, A. P´ asztor, and C. H. Wong, Finite den- sity QCD phase structure from strangeness fluctuations, Phys. Rev. D113, 054507 (2026), arXiv:2510.26455 [hep- lat]

    arXiv 2026

  49. [49]

    A. D. Kennedy, I. Horvath, and S. Sint, A New exact method for dynamical fermion computations with nonlo- cal actions, Nucl. Phys. B Proc. Suppl.73, 834 (1999), arXiv:hep-lat/9809092

    Pith/arXiv arXiv 1999

  50. [50]

    M. A. Clark and A. D. Kennedy, The RHMC algo- rithm for two flavors of dynamical staggered fermions, Nucl. Phys. B Proc. Suppl.129, 850 (2004), arXiv:hep- lat/0309084

    arXiv 2004

  51. [51]

    Omelyan, I

    I. Omelyan, I. Mryglod, and R. Folk, Symplectic ana- lytically integrable decomposition algorithms: classifica- tion, derivation, and application to molecular dynamics, quantum and celestial mechanics simulations, Computer Physics Communications151, 272 (2003)

    2003

  52. [52]

    L¨ uscher, Properties and uses of the Wilson flow in lattice QCD, JHEP08, 071, [Erratum: JHEP 03, 092 (2014)], arXiv:1006.4518 [hep-lat]

    M. L¨ uscher, Properties and uses of the Wilson flow in lattice QCD, JHEP08, 071, [Erratum: JHEP 03, 092 (2014)], arXiv:1006.4518 [hep-lat]

    Pith/arXiv arXiv 2014

  53. [53]

    Bruno and R

    M. Bruno and R. Sommer (ALPHA), On the Nf -dependence of gluonic observables, PoSLA T- TICE2013, 321 (2014), arXiv:1311.5585 [hep-lat]

    Pith/arXiv arXiv 2014

  54. [54]

    Vautard and M

    R. Vautard and M. Ghil, Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series, Physica D: Nonlinear Phenomena35, 395 (1989)

    1989

  55. [55]

    Wolff (ALPHA Collaboration), Monte Carlo er- rors with less errors, Comput

    U. Wolff (ALPHA Collaboration), Monte Carlo er- rors with less errors, Comput. Phys. Commun.156, 143 (2004), [Erratum: Comput.Phys.Commun. 176, 383 (2007)], arXiv:hep-lat/0306017

    Pith/arXiv arXiv 2004

  56. [56]

    Ramos, Automatic differentiation for error analysis, PoSTOOLS2020, 045 (2021), arXiv:2012.11183 [hep- lat]

    A. Ramos, Automatic differentiation for error analysis, PoSTOOLS2020, 045 (2021), arXiv:2012.11183 [hep- lat]

    Pith/arXiv arXiv 2021

  57. [57]

    Raiteri, A

    P. Raiteri, A. Laio, F. L. Gervasio, C. Micheletti, and M. Parrinello, Efficient reconstruction of complex free energy landscapes by multiple walkers metadynamics, J. Phys. Chem. B110, 3533 (2006), pMID: 16494409

    2006

  58. [58]

    Nishimura and D

    A. Nishimura and D. Dunson, Recycling intermediate steps to improve hamiltonian monte carlo, Bayesian Analysis15, 10.1214/19-ba1171 (2020)

    work page doi:10.1214/19-ba1171 2020

  59. [59]

    T. M. Sch¨ afer and G. Settanni, Data reweighting in meta- dynamics simulations, J. Chem. Theory Comput.16, 2042 (2020), pMID: 32192340. 6 End Matter Appendix A: Potential buildup—We perform well- tempered Metadynamics [29] simulations withγ= 1.1 to obtain initial estimates of the free energy surfaces. The simulations use four parallel walkers [56], with ...

    2042