Pith. Sign in

REVIEW

Testing and tuning symplectic integrators for Hybrid Monte Carlo algorithm in lattice QCD

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
Tetsuya Takaishi, Philippe de Forcrand

High Energy Physics - Lattice · Submitted 18 May 2005 · 25 references

Pith's verdict: A solid, modest methods paper: the Omelyan 2MN integrator really is about 50% cheaper than leapfrog for HMC, and the retuning of lambda is a genuine correction to Omelyan's prescription — but the headline number is cashed out at the edge of the verified scaling regime and nothing has error bars.

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 · 9 minor · Ask about this paper · The compiled record · Read the full review →
Challenge this review Comment (0) Read on arXiv ↗
Cite ▾
arXiv IDhep-lat/0505020 (version 3)
Pith IDRMX7PWIM, Pith's permanent record of this paper and its review
CitationPhilippe de Forcrand, Tetsuya Takaishi. "Testing and tuning symplectic integrators for Hybrid Monte Carlo algorithm in lattice QCD." arXiv:hep-lat/0505020 (2005). https://pith.science/paper/RMX7PWIM
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 paper argues that a tuned second-order minimum-norm integrator is about 50% more efficient than the standard leapfrog for Hybrid Monte Carlo in lattice QCD.

The reading

The paper asks whether a second-order integrator built to minimize the norm of its leading error coefficients can beat the standard second-order leapfrog in Hybrid Monte Carlo for lattice QCD. The authors measure the root-mean-square energy change at the end of a unit-length trajectory and find it about ten times smaller for the new integrator, which allows a step size about three times larger. Because each step costs two force evaluations instead of one, the net gain is about 50%, not 300%. They also find that the internal parameter $��$ should be tuned slightly above the value recommended by the integrator's designers and depends on the quark mass, and that starting with positions rather than momenta gives a small additional advantage with one fewer force evaluation per trajectory.

What carries the argument

The load-bearing object is the two-force-evaluation symmetric decomposition $I_{2MN}(�� t)=e^{���� t T}e^{(�� t/2)V}e^{(1-2��)�� t T}e^{(�� t/2)V}e^{���� t T}$, where $T$ and $V$ are the kinetic and potential Lie operators. Its leading error is $��(��)[T,[V,T]]+��(��)[V,[V,T]]$ with $��(��)=(1-6��+6��^2)/12$ and $��(��)=(1-6��)/24$. The tuning works by choosing $��_1=(1-1/��3)/2$, where $��=0$, and $��_2=1/6$, where $��=0$, to measure the two unknown error magnitudes $f$ and $g$, then fitting $����H^2��^{1/2}�� ��{��^2 f^2+��^2 g^2}�� t^2$ and minimizing over $��$.

Extended reading notes

Core claim

The central claim is that the second-order minimum-norm (2MN) integrator of Omelyan et al. is about 50% more efficient than the standard second-order leapfrog (2LF) for Hybrid Monte Carlo in lattice QCD. At $��=5.00$ and $��=0.160$ on $4^4$ lattices, the RMS energy error $�� ��H^2 ��^{1/2}$ scales as $�� t^2$ for both integrators, and the ratio $C_{2LF}/C_{2MN}$ is about 10 across $��$ (Fig. 2). Since the 2MN integrator costs two force evaluations per step, the efficiency gain is $��{10}/2 �� 1.5$. The authors also report that the optimal internal parameter $��$ is slightly larger than the Omelyan value $0.1931833275037836$ and varies with $��$, and that the position-first version gives a slightly lower minimum error and saves one force evaluation at the end of a trajectory.

Load-bearing premise

The 50% efficiency claim assumes that the step size can be scaled up by about $��{10}$ before the integrator becomes unstable, so that energy-error accumulation, not a stability wall, sets the maximum step; the supporting test is a $4^4$ lattice at $��=5.00$, $��=0.160$.

Editorial extensions

If this is right

  • Hybrid Monte Carlo simulations at the tested parameters can use the 2MN integrator with position-first ordering to gain about 50% in work per accepted trajectory over the 2LF integrator.
  • The optimal $��$ can be tuned per simulation parameter set; it lies slightly above the Omelyan value and moves with the quark mass.
  • The position-first version is the recommended choice because it gives a slightly lower minimum RMS error and requires one fewer force evaluation per trajectory.
  • The tested fourth-order minimum-norm integrators are unattractive at zero temperature because higher-order terms dominate already in the practical acceptance range, but at finite temperature the 4MN5FV integrator may give a gain of about 2.
  • The same decomposition could improve Trotter-Suzuki quantum Monte Carlo simulations when continuous imaginary time is not practical.

Reading between the lines

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

  • Because the optimal $��$ is $��$-dependent, a short calibration run measuring $f$ and $g$ at $��_1$ and $��_2$ could set $��$ automatically before production; the paper implies this but does not implement it as a workflow.
  • The fit neglects the cross term $��F G+GF��$ in $��H^2$; if that term grows near $��_c$, the two-parameter tuning may need a third measurement to remain accurate.
  • The finite-temperature gain suggests that integrator order should be chosen as a function of temperature and lattice size rather than fixed once for all simulations.
  • If instability at small quark mass is the binding constraint, adaptive step-size or force-gradient integrators are the natural next comparison against the 2MN integrator.
Original abstract

We examine a new 2nd order integrator recently found by Omelyan et al. The integration error of the new integrator measured in the root mean square of the energy difference, $\bra\Delta H^2\ket^{1/2}$, is about 10 times smaller than that of the standard 2nd order leapfrog (2LF) integrator. As a result, the step size of the new integrator can be made about three times larger. Taking into account a factor 2 increase in cost, the new integrator is about 50% more efficient than the 2LF integrator. Integrating over positions first, then momenta, is slightly more advantageous than the reverse. Further parameter tuning is possible. We find that the optimal parameter for the new integrator is slightly different from the value obtained by Omelyan et al., and depends on the simulation parameters. This integrator could also be advantageous for the Trotter-Suzuki decomposition in Quantum Monte Carlo.

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, 5,605 input tokens, $0.0002 reported.

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.53 · human participants 0.02 · controlled comparison 0.84 · randomized assignment 0.01 · states quantitative result 0.98 · compares with prior results 0.98 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v3. Results and the main claim are underlined; .

We examine a new 2nd order integrator recently found by Omelyan et al. The integration error of the new integrator measured in the root mean square of the energy difference, $\bra\Delta H^2\ket^{1/2}$, is about 10 times smaller than that of the standard 2nd order leapfrog (2LF) integrator. As a result, the step size of the new integrator can be made about three times larger. Taking into account a factor 2 increase in cost, the new integrator is about 50% more efficient than the 2LF integrator. Integrating over positions first, then momenta, is slightly more advantageous than the reverse. Further parameter tuning is possible. We find that the optimal parameter for the new integrator is slightly different from the value obtained by Omelyan et al., and depends on the simulation parameters. This integrator could also be advantageous for the Trotter-Suzuki decomposition in Quantum Monte Carlo.

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 examine a new 2nd order integrator recently found by Omelyan et al. The integration error of the new integrator measured in the root mean square of the energy difference, $\bra\Delta H^2\ket^{1/2}$, is about 10 times smaller than that of the standard 2nd order leapfrog (2LF) integrator. result 0.87 0.74 0.38 asserted 1.00
s1 As a result, the step size of the new integrator can be made about three times larger. result 0.76 0.79 0.16 asserted 0.80
s2 Taking into account a factor 2 increase in cost, the new integrator is about 50% more efficient than the 2LF integrator. result 0.89 0.87 0.53 asserted 1.00
s3 Integrating over positions first, then momenta, is slightly more advantageous than the reverse. result 0.99 0.85 0.04 asserted 0.95
s4 Further parameter tuning is possible. future work 0.25 0.29 0.04 hedged possibility 0.73
s5 We find that the optimal parameter for the new integrator is slightly different from the value obtained by Omelyan et al., and depends on the simulation parameters. result 1.00 0.96 0.09 asserted 1.00
s6 This integrator could also be advantageous for the Trotter-Suzuki decomposition in Quantum Monte Carlo. interpretation 0.81 0.35 0.05 hedged possibility 1.00

What kind of paper the classifier reports this is: simulation study 0.86 · evidence from simulation or computation 0.99.

What was examined, and what was not

sources read
  • arXiv listing abstract v3
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. h∆H2 i 1/2 as a function of ∆t. Simulations are performed at β = 5.00 and κ = 0.160 on 44 lattices. The line proportional to ∆t 2 is drawn to guide the eye. 3.4 Performance of 2nd order MN (2MN) integrator Here we compare the efficiency of the 2MN integrator with that of the 2LF integrator. For 2nd order integrators, from eq.(38) h∆H2 i 1/2 at small ∆t is expected to be C2V 1/2∆t 2 . We measure the coefficient C2 for both integrators at small enough ∆t, and by comparing the coefficients we obtain the performance of the 2MN integrator relative to the 2LF integrator
Figure 2. C2LF /C2MN as a function of κ. Simulations at β = 5.00(5.60) are performed on 44 (84 ) lattices. be done by following the analysis of [4]. However we found a problem with the 4MN integrator. Namely the error of the Hamiltonian h∆H2 i 1/2 is not simply described by eq.(38) but is dominated by higher order terms in ∆t already at small h∆H2 i 1/2
Figure 3. h∆H2 i 1/2 as a function of ∆t. Simulations are performed on 84 lattices. 4RC stands for the 4th order integrator obtained by eq.(12). The lines proportional to ∆t n are drawn to guide the eye. where κ42 is the relative cost factor and κ42 = 2.5 for the 2MN and 4MN5FV integrators. Sub￾stituting the measured values into G, G is calculated to be ≈ 2, which shows that the 4MN5FV integrator is more effective. Thus at finite temperature there is room to use a 4MN integrator depending on the simulation parameters. 4 Tuning the 2MN integrator The strategy to minimize eqs.(18) and (19) is…
Figure 4. h∆H2 i 1/2 as a function of λ. The right figure is a zoom of the left. Simulations are performed at β = 5.00 and κ = 0.160 on 44 lattices with ∆t = 0.05. The lines are determined from simulations of the position version integrator at λ1 and λ2. The position version has a small advantage over the velocity version, since it gives a slightly reduced minimum RMS error (right). 2LF integrator at λ = 0, and vice versa. The position version of the 2MN integrator gives a slightly smaller minimum. At λ = 0 the velocity version of 2MN integrator has a smaller error than the position version,…
Figure 5. f|∆t 2 , |g|∆t 2 and |f|/|g| as a function of 1/κ. Simulations are performed at β = 5.00 on 4 4 lattices with ∆t = 0.05. The dashed line indicates κc = 0.187[19]. 5 6 7 8 9 10 1/κ 0.185 0.19 0.195 0.2 0.205 0.21 λ λ=0.193183327503783 κ c
Figure 6. Optimal λ as a function of 1/κ. Simulations are performed at β = 5.00 on 44 lattices with ∆t = 0.05. The dashed line indicates κc = 0.187. 12

Figures reproduced from arXiv:hep-lat/0505020 by the authors.

References
How later papers used it

61 later papers cite it

Evidence: substantial. 60 of the citing papers share no author with it, and 49 of those carry a confident label from Pith's citation classifier.

Pith read 24 of the 72 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 18 of 47 papers with this label
How the literature uses it

51 arXiv papers cite this in 59 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 · 16 sentences without a confident label

Uses its method or tool 40 papers
Uses its data or values 1 paper
Compares with it 1 paper
Cites it as background 1 paper
  • They range from preconditioning the solver (e.g. even-odd [ref] , domain decomposition [ref] , deflation [ref] , multigrid [ref] ) to preconditioning the action (ILU [ref] , UV-filtering [ref] ) to tuning the integrator ( [ref] ).

    RHMC with Block Solvers and Multiple Pseudofermions · Introduction · 0.97

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.