Pith. sign in

REVIEW 4 major objections 6 minor 16 references

Estimating Large Global Significances with a New Monte Carlo Extrapolation Method

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A chi-squared fit to the tail of toy-MC likelihood ratios estimates large global significances that direct counting cannot reach.

desk verdict Practical tail-extrapolation trick for global significances, validated at 4σ but with no propagated uncertainty; worth reviewing, needs a robustness analysis. read the letter →

arxiv 2412.20777 v2 pith:HQPAFZ4S submitted 2024-12-30 hep-ex

Liangliang Chen , Yufei Chen , Gerry Bauer , Leonard G. Spiegel , Zhen Hu , Kai Yi This is my paper · ORCID
classification hep-ex
keywords globalsignificancelocaltoyMonteCarloextrapolationmethodlog-likelihoodratiochi-squaredtailfitlook-elsewhereeffectmassspectrumsearch
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that very large global significances, which normally require an astronomically large number of simulated toy experiments to estimate by direct counting, can be obtained from modest toy samples by fitting a chi-squared distribution to the high tail of the log-likelihood ratio and extrapolating. It validates the method on the Y(4140) resonance observed by CDF, where direct counting gives 4.1σ and the new extrapolation gives 4.0σ, within the 0.1σ margin the authors allow. It then applies the method to cases where direct counting is impractical: the CMS observation of Y(4140), with a local significance of 7.6σ, is assigned a global significance of 6.6σ, and the ATLAS χb(3P) signal is assigned 5.7σ. A sympathetic reader would care because this gives experimental analyses a practical route to global significances that would otherwise require more than $10^{14}$ toy experiments, and a way to project how much additional data is needed to reach 5σ.

What carries the argument

The load-bearing object is the $χ^{2}$ distribution used as a tail model for the log-likelihood ratio 2×(L0−L1) from background-only toys. A single parameter n, the number of degrees of freedom, is fitted to the simulated tail above a ratio of 15; for the three test cases the fitted values are n≈2.5, 2.6, and 1.6. Because the $χ^{2}$ tail is linear on a log scale at large values, a fit to ratios between 15 and 70 can be integrated to predict the number of toys that would exceed much larger ratios such as 28 or 58, which are the values corresponding to 5.3σ and 7.6σ local significances. The p-value is the predicted count above the observed ratio divided by the total number of toys (counted below 15 plus integrated above 15), and the final significance is read off a symmetric Gaussian.

What would settle it

Generate a very large background-only toy sample for the CDF Y(4140) setup, directly count the number of toys with likelihood ratio above 28, and compare that count with the prediction from the chi-squared tail fitted above 15; a statistically significant mismatch between the direct count and the extrapolated count would show that the tail model fails at high ratios.

Watch

Extended reading notes

Core claim

The central claim is that, for background-only toy Monte Carlo experiments, the distribution of the log-likelihood ratio 2×(L0−L1) has a tail that is well described by a $χ^{2}$ distribution with a single fitted degrees-of-freedom parameter, and that this tail can therefore be extrapolated far beyond the largest values actually simulated. The method works as follows: simulate a modest number of toys, split the likelihood-ratio distribution at a threshold (15 in this paper, chosen after trial and error), count toys below the threshold, fit a $χ^{2}$ function to the tail above it, and integrate the fitted function to obtain the expected number of toys above the likelihood ratio observed in real data. That expected count divided by the total number of toys is the p-value, converted to a Gaussian significance. On CDF's Y(4140), the extrapolated p-value 2.89×$10^{{-5}}$ gives 4.0σ versus 2.50×$10^{{-5}}$ and 4.1σ from direct counting; on CMS's Y(4140), whose local significance is 7.6σ, the method yields a global significance of 6.6σ (with the alternative upcrossing method giving 6.8σ), and the same procedure on ATLAS's χb(3P) yields 5.7σ (with the alternative giving 5.6σ).

Load-bearing premise

The argument depends on the fitted chi-squared tail continuing to describe the true fluctuation distribution far above the range where it was fitted, so the extrapolation to ratios like 28 or 58 is trustworthy; if the true tail falls off differently, the estimated global significance is biased.

Editorial extensions

If this is right

  • Global significances for local significances above 5σ can be evaluated with a few hundred thousand toy experiments instead of the 10^14 or more that direct counting would require.
  • The extrapolated global significance agrees with direct counting to within 0.1σ on the CDF Y(4140) case, and with the alternate upcrossing method to within 0.2σ on the CMS and ATLAS cases.
  • The method is transferable across different experiments and mass spectra, as demonstrated on CDF, CMS, and ATLAS data with different signal and background models.
  • For signals below 5σ, the same simulation machinery lets analyzers estimate the probability of reaching 5σ with additional data, as illustrated for the CMS X(7100) candidate.
  • Because it uses only the tail of the fitted distribution, the method also works when the observed significance is so high that no single toy experiment reaches it.

Reading between the lines

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

  • The choice of the tail threshold (15) is heuristic and tuned by trial and error; a principled, data-driven way to pick it would make the method more robust, a step the paper does not take.
  • In the CMS case the two cross-checked methods agree to 0.2σ, yet their p-values differ by an order of magnitude; near decision thresholds like 3σ or 5σ such differences would matter, so the method is safest for the high-significance regime it targets.
  • One could stress-test the chi-squared tail assumption by applying the extrapolation to a search whose exact global significance is computable analytically, such as a single-channel Gaussian field, and checking whether the fitted tail remains valid as the extrapolation distance grows.
  • The same tail-fitting logic might extend beyond mass-spectrum searches to any likelihood-ratio scan, provided the asymptotic chi-squared behavior of the test statistic holds.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper describes a new 'extrapolation' method for estimating global significances in particle physics searches. The method generates a modest number of background-only toy Monte Carlo experiments, fits a chi-square distribution to the upper tail (starting at 2ΔLLR = 15) of the likelihood-ratio distribution, and integrates the fitted function to estimate the expected number of toys above a high threshold. The authors validate the method on CDF's Y(4140) by comparing with direct counting (4.0σ vs 4.1σ), cross-check it with the Gross-Vitells (G-V) method, and apply it to CMS's Y(4140) (6.6σ) and ATLAS's χb(3P) (5.7σ), where direct counting is impractical.

Significance. If the method's accuracy at large extrapolation distances were established, it would be a practically valuable, easy-to-implement alternative to brute-force toy MC for global significances beyond 5σ, avoiding samples of order 10^14. The paper includes a useful demonstration of the conventional counting method on the CMS X(7100) case and a reproducible simulation setup. The CDF validation is encouraging: the extrapolated tail expectation above 2ΔLLR = 28 (10.4 toys) agrees with the direct count (9 toys) within 0.1σ. However, the current evidence for predictive power in the advertised high-significance regime is limited: the validation is in-sample (the tail threshold is chosen on the same dataset), and the only high-significance cross-check (G-V) differs by a factor of about 2.6 in p-value for CMS. The method's usefulness depends on quantifying the uncertainty of tail extrapolation.

major comments (4)
  1. [Sec. 4.3, Table 2] No uncertainties are quoted for the extrapolated global significances in Table 2 (CMS 6.6σ, ATLAS 5.7σ). In the CMS case the extrapolated p-value is 1.58×10^-11 from an expected count of 4.30×10^-6 toys above 2ΔLLR = 58, and a small change in the fitted degrees-of-freedom parameter n = 2.6 (quoted without uncertainty) or in the tail threshold would change this expectation by orders of magnitude. The authors should propagate the fit uncertainty in n and the finite-toy uncertainty to the extrapolated p-value, and quote an uncertainty on the final significance. Without this, the claim that the method 'gives good estimation' is not quantitatively assessable.
  2. [Sec. 4.2.4] The tail-fit threshold of 15 is chosen 'after some trial and error' on the CDF toy sample, and the same 359,758-toy sample is then used to validate the method at 2ΔLLR = 28. This is an in-sample validation, not an out-of-sample prediction. The central claim of predictive power for CMS (threshold 58) and ATLAS (threshold 36) requires either an out-of-sample test, a closure test at an intermediate threshold where direct counting is feasible, or a systematic scan of the threshold and fit range demonstrating that the extrapolated significance is stable. The current presentation does not provide such evidence.
  3. [Sec. 4.2.4 and Sec. 4.3] The paper provides no theoretical or empirical justification that a chi-square distribution with a single fitted degree-of-freedom parameter describes the extreme tail of the scanned 2ΔLLR distribution over many orders of magnitude. In the CMS case the extrapolated expected count is about 10^9 below the tail-integral anchor at 15, so the tail-shape assumption is load-bearing. The authors should test this assumption concretely, for example by using a larger toy sample at an intermediate threshold, comparing with an alternative parametric tail (e.g., exponential or generalized Pareto), or varying the fit threshold over a plausible range and showing that the extrapolated significance is robust.
  4. [Sec. 4.3] In the only high-significance regime where an independent cross-check exists, the extrapolation method and the G-V method disagree by a factor of 2.6 in p-value (1.58×10^-11 vs 6.16×10^-12). The text dismisses this as a 0.2σ difference, but for a claim of 'good estimation' the relevant quantity is the p-value, and a factor 2.6 is large. Moreover, the G-V input <N(c0)> = 18.80 ± 2.75 implies a substantial uncertainty in the G-V p-value, which is not propagated either. The authors should quantify the accuracy target of the method and either reconcile the discrepancy or include it as an uncertainty.
minor comments (6)
  1. [Sec. 4.2.4] The fit range is described both as [15,70] and as 'above 15' when computing the total toy count; please clarify whether the normalization integral of the fitted chi-square distribution is taken over [15,70] or [15,∞).
  2. [Sec. 4.2.4] The quoted values of the fitted degrees-of-freedom parameter (n = 2.5, 2.6, 1.6) are given without fit uncertainties; please report them.
  3. [Sec. 4.4] The statement 'After extrapolation, we obtained a result of (3.53 ± 0.07) × 10^-04' does not define what the quantity is or where the uncertainty comes from; please state that this is the expected number of toys above 2ΔLLR = 36 and explain the uncertainty's source.
  4. [Footnote 1, Sec. 4.3] The footnote 'We became aware of a relevant paper [15] after publication, thus a comparison with it cannot be done' is a self-acknowledged missing comparison; the authors should either add the comparison or justify its omission in the main text.
  5. [Abstract and Sec. 2.3] The phrase 'assuming symmetrical Gaussian distributions' is imprecise; the conversion uses the one-sided upper tail of a standard normal distribution, not a two-sided symmetric interval.
  6. [Sec. 4.3] The sentence 'their corresponding p-values differ by an order of magnitude, making error calculation challenging' should be elaborated or removed; an order-of-magnitude difference is precisely what needs to be explained when claiming good estimation.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the extrapolated tail count is not an input to the chi-squared fit, and the CDF direct-count and G-V comparisons provide independent checks.

full rationale

The paper's central quantity—the expected number of toy experiments above a target likelihood ratio—is obtained by integrating a chi-squared distribution fitted to the observed tail above ratio 15 (Sec 4.2.4). The target thresholds (28, 58, 36) are not used as fit constraints, and the number of toys above the target is not an input to the fit; it is the output of the integral. In the CDF validation, the extrapolated expectation 10.4 is compared with the independently counted 9 events above 28, which is a separate estimator on the same toy sample and therefore a genuine cross-check, not a quantity forced by construction. The post-hoc choice of the fit threshold ('after some trial and error') and the in-sample location of the threshold 28 within the fit interval [15,70] are statistical robustness concerns, not circularity. The CMS and ATLAS results are benchmarked against the independent Gross-Vitells method, and agreement within 0.1-0.2 sigma is reported. The only self-citation is Ref [11] (Yi, Spiegel, Hu) used in Sec. 3 to describe the conventional counting method; this is not load-bearing for the new extrapolation claim or for its validation. No step in the derivation reduces by definition to its own inputs, no fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' prior work. The appropriate finding is a low circularity score reflecting the minor non-load-bearing self-citation.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The method introduces no new physical entities. Its statistical machinery is a chi-squared fit with a free degrees-of-freedom parameter and a post-hoc chosen fit threshold. The background shapes and search windows are imported from the experimental analyses.

free parameters (2)
  • chi-squared degrees of freedom n = 2.5 (CDF), 2.6 (CMS), 1.6 (ATLAS)
    Fitted to the toy MC log-likelihood ratio tail in each case; the extrapolated p-value depends on this fit.
  • tail fit threshold = 15 (likelihood ratio)
    Chosen after trial and error to make the chi-squared fit work (Section 4.2.4).
assumptions (3)
  • domain assumption The high tail of the toy MC log-likelihood ratio distribution is well described by a chi-squared distribution with unknown degrees of freedom.
    Invoked in Section 4.2.4 when fitting the tail above ratio 15; no theoretical derivation is given.
  • domain assumption The background-only toy MC samples correctly model the null hypothesis, including known resonances and phase-space backgrounds.
    Standard practice in the field; the shapes are taken from the experimental papers.
  • standard math Converting p-values to Gaussian sigmas via a one-sided symmetric Gaussian integral is appropriate.
    Common convention in particle physics; stated in Section 2.3.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Estimating Large Global Significances with a New Monte Carlo Extrapolation Method." pith.science (2026). https://pith.science/paper/HQPAFZ4S

@misc{pith2026241220777,
  author       = {Pith},
  title        = {Pith review of: Estimating Large Global Significances with a New Monte Carlo Extrapolation Method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HQPAFZ4S}},
  note         = {Machine review of arXiv:2412.20777}
}
read the original abstract

In particle physics, it is needed to evaluate the possibility that excesses of events in mass spectra are due to statistical fluctuations as quantified by the standards of local and global significances. Without prior knowledge of a particle's mass, it is especially critical to estimate its global significance. The usual approach is to count the number of times a significance limit is exceeded in a collection of simulated Monte Carlo (MC) 'toy experiments.' To demonstrate this conventional method for global significance, we performed simulation studies according to a recent Compact Muon Solenoid (CMS) result to show its effectiveness. However, this counting method is not practical for computing large global significances. To address this problem, we developed a new 'extrapolation' method to evaluate the global significance. We compared the global significance estimated by our new method with that of the conventional approach, and verified its feasibility and effectiveness. This method is also applicable for cases where only small toy MC samples are available. In this approach, the significance is calculated based on p-values, assuming symmetrical Gaussian distributions.

Figures

Figures reproduced from arXiv: 2412.20777 by the authors.

Figure 1
Figure 1. CMS experiment’s J/ψJ/ψ invariant mass spectrum fit with the no-interference model showing three structures [1]. 3.1. Example of CMS Local Significance In this CMS example, the background components are modeled with two threshold functions used in the CMS analysis [12], while the signal component is described by a standard relativistic Breit–Wigner formula [12]: BW(m; m0, Γ0) = p mΓ(m) m2 0 − m2 − imΓ(m) , Γ(m) = Γ0… view at source ↗
Figure 2
Figure 2. An example of a toy experiment simulating CMS’s J/ψJ/ψ mass spectrum along with a test fits. The upper panel shows the null-hypothesis fit to this toy, while the lower panel shows the signal-hypothesis fit the same toy. A local significance of 4.4 standard deviations for the X(7100) peak is obtained by comparing the likelihoods of the two fits. 3.2. Example of CMS Global Significance To calculate the global signific… view at source ↗
Figure 3
Figure 3. The log-likelihood difference, (L0 − L1)trial, of 13409 trials simulating the CMS case study. The red line and arrow indicate that there are 14 trials with a value greater than 8.5. 0. 0 1 0. 0 2 0. 0 3 0. 0 4 0. 0 5 N or m al di stri b uti o n 3. 1 σ [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: A normal distribution showing the 3.1σ significance of the CMS example corresponding to the p-value of 0.00104. The red arrow indicates the area with an integral value of 0.00104. 3.3. Expected Significance with Additional Data To verify if a newly found signal is real…
Figure 5
Figure 5. Figure 5: The log-likelihood difference distribution for a semi-simulated exercise to project the significance of CMS’s observed X(7100) signal. The original sample is supplemented by a simulation of 50% more signal and 50% additional background. This exercise illustrates that c…
Figure 6
Figure 6. Figure 6: The fitted Y(4140) distributions from two collaborations. Left (CDF): The mass difference (∆m = m(µ +µ −K +K −) − m(µ +µ −)) in the B + mass window [4]; Right (CMS): The number of B + → J/ψϕK + candidates as a function of ∆m = m(µ +µ −K +K −) − m(µ +µ −) [7] [PITH_FUL…
Figure 7
Figure 7. Figure 7: The null-signal likelihood ratio distribution for toy MC computed for 1 MeV binned (blue) and unbinned (red) data. Left: linear scale; right: logarithmic scale. 4.2.3. The Conventional Method—Direct Counting In order to demonstrate our extrapolation method, we generate…
Figure 8
Figure 8. Figure 8: shows a histogram of the distribution of all the log-likelihood ratio values for this sample of pseudo-experiments. 0 10 20 30 40 50 60 70 2∆NLL 1 10 2 10 3 10 4 10 Events / (0.500) Number(>28)=9 Number(FULL)=359758 2∆NLL=28 [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: The tail of the likelihood ratio distribution on a log scale and fit with a χ 2 function for the CDF case. The reduced χ 2 value of 0.69 indicates reasonably good fit quality. 4.2.5. The G-V Method As an additional cross-check, we use the G-V method to verify the effec…
Figure 10
Figure 10. Figure 10: The likelihood ratio fluctuation curve obtained by fixing the signal mass parameter in the fit to the CDF Y(4140) toy Monte Carlo sample. The horizontal axis represents the fixed signal mass, while the vertical axis represents the obtained likelihood ratios. The red s…
Figure 11
Figure 11. Figure 11: The likelihood ratio distribution on a log scale fit with a χ 2 function for CMS’s Y(4140) observation. The reduced χ 2 value of 0.98 indicates reasonably good fit quality. 4.1 4.2 4.3 4.4 4.5 4.6 4.7 m(J/ψφ) (GeV) 0 2 4 6 8 10 NLL ∆ 2 [PITH_FULL_IMAGE:figures/full_f…
Figure 12
Figure 12. Figure 12: The likelihood ratio fluctuation curve obtained by fixing the signal mass parameter in the fit to the CMS Y(4140) toy Monte Carlo sample. The red solid lines shows the c0 reference level, which crosses about 19 times with the test of statistics. For an analyzer, a glo…
Figure 13
Figure 13. Figure 13: 9.6 9.8 10 10.2 10.4 10.6 10.8 x_mass 0 10 20 30 40 50 60 70 Events / ( 0.025 ) ATLAS Data Fit Gaus1 Gaus2 Gaus3 Background 9.6 9.8 10 10.2 10.4 10.6 10.8 (GeV) Υ(1S) µ -)+m + γ)-m(µ - µ + m(µ −3 −2 −1 0 1 2 Pull [PITH_FULL_IMAGE:figures/full_fig_p014_13.png]
Figure 14
Figure 14. Figure 14: The likelihood ratio distribution on a log scale fit with a χ 2 function for ATLAS’s χb (3P) observation. The reduced χ 2 value of 0.71 indicates reasonably good fit quality. 4.5. Correlations Sometimes, we want to check if there are any correlation among the signific…
Figure 15
Figure 15. Figure 15: Upper (lower) left: signal width vs. signal mass in the CDF (CMS) case, indicating a small correlation between the signal width and signal mass. Upper (lower) middle: 2 × (L0 − L1) vs. signal mass in the CDF (CMS) case, indicating the signal mass has minimal influence…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

  1. [1]

    New Structures in the J/ψJ/ψ Mass Spectrum in Proton-Proton Collisions at √s = 13 TeV .Phys

    CMS Collaboration. New Structures in the J/ψJ/ψ Mass Spectrum in Proton-Proton Collisions at √s = 13 TeV .Phys. Rev. Lett. 2024, 132, 111901

    work page 2024

  2. [2]

    Experimental Road to a Charming Family of Tetraquarks

    Zhu, F.; Bauer, G.; Yi, K. Experimental Road to a Charming Family of Tetraquarks... and Beyond.Chin. Phys. Lett. 2024, 41, 111201

    work page 2024

  3. [3]

    Four-Charm-Quark Matter from the CMS Collaboration as a witness of the development of high- precision hadron spectroscopy

    Liu, X. Four-Charm-Quark Matter from the CMS Collaboration as a witness of the development of high- precision hadron spectroscopy. Sci. Bull. 2024, 69, 2802-2803

    work page 2024

  4. [4]

    Evidence for a Narrow Near-Threshold Structure in the J/ψϕ Mass Spectrum in B+ → J/ψϕK+ Decays

    CDF Collaboration. Evidence for a Narrow Near-Threshold Structure in the J/ψϕ Mass Spectrum in B+ → J/ψϕK+ Decays. Phys. Rev. Lett. 2009, 102, 242002

    work page 2009

  5. [5]

    Experimental review of structures in the J/ψϕ mass spectrum

    Yi, K. Experimental review of structures in the J/ψϕ mass spectrum. Int. J. Mod. Phys. A 2013, 28, 1330020

    work page 2013

  6. [6]

    Observation of the Y(4140) Structure in the J/ψϕ Mass Spectrum in B± → J/ψϕK± Decays

    CDF Collaboration. Observation of the Y(4140) Structure in the J/ψϕ Mass Spectrum in B± → J/ψϕK± Decays. Mod. Phys. Lett. A 2017, 32, 1750139

    work page 2017

  7. [7]

    Observation of a peaking structure in the J/ψϕ mass spectrum from B± → J/ψϕK± decays

    CMS Collaboration. Observation of a peaking structure in the J/ψϕ mass spectrum from B± → J/ψϕK± decays. Phys. Lett. B 2014, 734, 261-281

    work page 2014

  8. [8]

    Trial factors for the look elsewhere effect in high energy physics

    Gross, E.; Vitells, O. Trial factors for the look elsewhere effect in high energy physics. Eur. Phys. J. C 2010, 70, 525-530

    work page 2010

Show all 16 references
  1. [9]

    Observation of a new χb state in radiative transitions to Υ(1S) and Υ(2S) at ATLAS

    ATLAS Collaboration. Observation of a new χb state in radiative transitions to Υ(1S) and Υ(2S) at ATLAS. Phys. Rev. Lett. 2012, 108, 152001

    2012

  2. [10]

    Probability and Statistics in Experimental Physics ; Springer Science & Business Media: Berlin/Heidelberg, Germany, 1992

    Byron, P .R. Probability and Statistics in Experimental Physics ; Springer Science & Business Media: Berlin/Heidelberg, Germany, 1992

    1992

  3. [11]

    A global significance evaluation method using simulated events

    Yi, K.J.; Spiegel, L.; Hu, Z. A global significance evaluation method using simulated events. arXiv 2023, arXiv:2310.14317

    2023 arXiv

  4. [12]

    Observation of new structures in the J/ψJ/ψ mass spectrum in pp collisions at √s = 13 TeV

    CMS Collaboration. Observation of new structures in the J/ψJ/ψ mass spectrum in pp collisions at √s = 13 TeV . CMS-PAS-BPH-21-003. 2022. Available online: https://cds.cern.ch/record/2815336 (accessed on 9 July 2022). 17 of 17

    2022

  5. [13]

    ROOT: An object oriented data analysis framework

    Brun, R.; Rademakers, F. ROOT: An object oriented data analysis framework. Nucl. Inst. Meth. Phys. Res. 1997, 389, 81-86

    1997

  6. [14]

    Observation of New Structure in the J/ψJ/ψ Mass Spectrum in Proton-Proton Collisions at √s = 13 TeV

    CMS Collaboration. Observation of New Structure in the J/ψJ/ψ Mass Spectrum in Proton-Proton Collisions at √s = 13 TeV . 2023. Available online: https://doi.org/10.17182/hepdata.141028 (accessed on 13 July 2023)

    2023 doi

  7. [15]

    The profile likelihood ratio and the look elsewhere effect in high energy physics

    Ranucci, G. The profile likelihood ratio and the look elsewhere effect in high energy physics. Nucl. Inst. Meth. Phys. Res. A 2012, 661, 77-85

    2012

  8. [16]

    Review of Particle Physics

    Particle Data Group. Review of Particle Physics. Phys. Lett. B 2008, 667, 1

    2008

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.