Pith. sign in

REVIEW 3 major objections 3 minor 14 references

Measurement of the Three-Flavor Composition of Astrophysical Neutrinos with Contained IceCube Events

T0 review · 3 major / 3 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read For the first time, all three astrophysical neutrino flavors are measured to be nonzero at Earth.

desk verdict First 68% CL flavor-triangle closure from IceCube is a real milestone, but the Wilks-based contours are not yet validated on this low-statistics, bounded parameter space. read the letter →

arxiv 2507.07212 v1 pith:WY3REU5J submitted 2025-07-09 astro-ph.HE hep-ex

Aswathi Balagopal V. , Vedant Basu , Albrecht Karle (for the IceCube Collaboration) This is my paper
classification astro-ph.HEhep-ex
keywords astrophysicalneutrinosflavorcompositiontaudoublecascadesIceCubeneutrinooscillationsMESEforwardfolding
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 aims to measure the flavor composition of astrophysical neutrinos at Earth, meaning the fractions of electron, muon, and tau neutrinos, using contained neutrino interactions in the IceCube detector. Its central claim is that all three flavor fractions are now constrained to be nonzero at better than 68% confidence, which is the first time such a measurement has closed its contour. The best-fit Earth ratio is $f_e:f_\mu:f_\tau = 0.30:0.37:0.33$, consistent with standard three-flavor oscillations from a pion-decay source. Zero electron neutrinos are excluded at 98.7% CL, zero tau neutrinos at 91.9% CL, and a neutron-decay-dominated source at 94.8% CL. If correct, this establishes tau neutrinos as a present component of the astrophysical flux and sharpens the case that cosmic neutrino sources produce neutrinos via pion decay.

What carries the argument

The load-bearing object is the double-cascade selection: a likelihood-based reconstruction that fits each MESE event under a two-cascade hypothesis using spatial and timing information of the deposited charge, then keeps events whose reconstructed properties, such as cascade energies, energy asymmetry, tau decay length from cascade separation, and energy near each cascade vertex, fall in ranges predefined from all-flavor simulations. This tags tau-neutrino charged-current interactions, where the tau decays after a few meters to tens of meters, with an expected yield of 7 events and 70% purity. These double-cascade events, binned by energy, zenith angle, and reconstructed length, are forward-folded together with cascade and track templates into a fit that includes atmospheric backgrounds and detector-related systematic uncertainties as nuisance parameters.

What would settle it

Re-run the flavor fit with the double-cascade events removed: if the 68% contour still closes and the zero-tau exclusion survives, the double-cascade channel is not load-bearing; if the contour opens, the claimed milestone rests entirely on those nine events. An independent cross-check would be to tag tau neutrinos in the same MESE sample with the neural-network classifier the paper cites and test whether the tau fraction remains positive.

Watch

Extended reading notes

Core claim

The paper's central claim is that the flavor composition of the astrophysical neutrino flux at Earth has now been measured with all three flavors statistically present: a fit to the MESE sample gives $f_e:f_\mu:f_\tau = 0.30:0.37:0.33$, with zero electron neutrinos excluded at 98.7% CL and zero tau neutrinos at 91.9% CL. This is the first IceCube flavor measurement to close its 68% contour on the ternary diagram. The result is achieved by adding a likelihood-based double-cascade classification that identifies tau neutrinos by the two cascades separated by the tau decay length, breaking the electron-tau degeneracy that limited earlier analyses, and by including lower-energy TeV-scale events and 11.4 years of livetime. The best-fit point lies on the line connecting the three standard source scenarios allowed by neutrino oscillations, consistent with the standard theory. A pure neutron-decay source is rejected at 94.8% CL.

Load-bearing premise

The double-cascade likelihood selection truly separates tau-neutrino events from single cascades with the simulated 70% purity, so the nine observed double-cascade events (against seven expected) carry enough information to close the 68% contour and reject a zero tau fraction.

Editorial extensions

If this is right

  • The flavor ratio at Earth becomes anchored by all three flavors, so future analyses can treat tau neutrinos as a measured component rather than a negligible or degenerate background.
  • A source dominated by neutron decay is disfavored at 94.8% CL, narrowing the allowed production mechanisms for the astrophysical neutrino flux.
  • The closure of the 68% contour means electron and tau neutrino fractions are each bounded away from zero, a qualitative step beyond earlier upper-limit-style measurements.
  • The difference in contour closure between broken and single power law spectral assumptions shows that flavor results must be paired with a correct spectral model, so improved spectral measurements will feed directly into sharper flavor constraints.
  • With more livetime and better tau tagging, the muon-damped and pion-decay source scenarios, which still overlap at 68% CL, can be separated.

Reading between the lines

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

  • An implicit consequence the paper does not draw out: if the tau fraction stays near one-third as statistics grow, the double-cascade sample itself becomes a collection of individual tau-neutrino candidates usable for point-source or cross-correlation searches.
  • Because the 95% contour closes under a single power law but not under the broken power law, one testable cross-check is to fit flavor jointly with both spectral assumptions and see whether the tau fraction shifts; a stable tau fraction would strengthen the flavor claim beyond the current fit.
  • The measurement assumes the flavor ratio is energy-independent between 1 TeV and 10 PeV, so the quoted Earth ratio should be read as a flux-weighted average if source cooling makes the flavor composition energy-dependent.
  • Independent confirmation could come from high-energy tau neutrinos that produce resolvable muon tracks in the detector: their rate should be consistent with $f_\tau \approx 0.33$, and a large deficit would point to systematics in the double-cascade purity.
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

3 major / 3 minor

Summary. The paper presents a measurement of the astrophysical neutrino flavor composition at Earth using 11.4 years of IceCube Medium Energy Starting Events (MESE). Events are classified into cascades, tracks, and double cascades, and a forward-folding likelihood fit is used to constrain the flavor fractions f_e, f_mu, f_tau under a baseline broken power-law (BPL) spectral assumption, with a single power-law (SPL) cross-check. The best-fit Earth flavor ratio is f_e:f_mu:f_tau = 0.30:0.37:0.33, and the paper claims, for the first time, that the fraction of each flavor is constrained to be greater than zero at more than 68% confidence. Additional claims are rejection of zero electron neutrinos at 98.7% CL, zero tau neutrinos at 91.9% CL, and a neutron-decay source at 94.8% CL.

Significance. If the statistical inference is valid, this is a notable step forward: a single dataset constraining all three flavors simultaneously, using a dedicated double-cascade classification to break the electron/tau degeneracy and extending to TeV-scale events. The paper is transparent about the small double-cascade sample (9 observed versus 7 expected) and provides a useful cross-check between BPL and SPL spectral assumptions. However, the central milestone claim rests on Wilks-theorem contours on a bounded ternary parameter space with sparse counts, and the manuscript does not demonstrate that these contours have the claimed frequentist coverage. The result is therefore promising but not yet fully supported as presented.

major comments (3)
  1. [§3 (Fig. 2) and §4] The confidence levels quoted in §4 (98.7% for f_e=0, 91.9% for f_tau=0, 94.8% for neutron decay) are derived from the Wilks contours shown in Fig. 2. The null hypotheses f_e=0 and f_tau=0 lie on the boundary of the ternary simplex, where the standard chi-square approximation of the profile-likelihood ratio is invalid. Boundary-aware calibration (e.g., Feldman-Cousins or Monte Carlo) is required to quote these exclusion CLs. Without such calibration, the central claim that all three flavor fractions are >0 at 68% CL is not supported by the analysis as presented.
  2. [§2 and Table 1] The sensitivity to tau neutrinos rests on only 9 double-cascade classified events (Table 1; expectation 7). The paper itself acknowledges "large statistical uncertainties" in the double-cascade histograms (§3). With such sparse bins, the asymptotic chi-square distribution of the profile-likelihood test statistic is not assured, especially when profiling over atmospheric and detector nuisance parameters. The manuscript should include a coverage study based on pseudo-experiments demonstrating that the 68% and 95% contours have the claimed frequentist coverage; this is load-bearing for the milestone claim.
  3. [§3 (Fig. 2) and §4] The 95% contour closes under the SPL assumption but not under the baseline BPL assumption, and the paper attributes this to the harder SPL spectral index. This shows that the contour geometry is sensitive to the assumed astrophysical spectrum. The paper does not propagate the uncertainty in the spectral shape into the flavor contours, for example by profiling over the BPL parameters or by presenting a combined treatment of both spectral models. The robustness of the 68% closure, which is the basis of the "first time" claim, is therefore not yet established.
minor comments (3)
  1. [§3] There is a typo in "uncertanties" in the sentence discussing the double-cascade histograms.
  2. [§2 (References)] The fit is performed with NNMFit, but reference [7] is given as "paper in preparation". A public code repository or a more detailed description of the fitting framework would help reproducibility.
  3. [§4] The claim that this analysis uses "an updated treatment of systematics and better modeling of the ice" is not supported by details in the proceedings text; a reference to a companion paper or a description of the changes would be helpful.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: the flavor fractions are free parameters fitted to MESE data; the only self-citations (companion BPL spectrum fit, NNMFit in preparation) are non-load-bearing and cross-checked.

full rationale

The analysis is a forward-folded maximum-likelihood measurement, not a derivation: f_e, f_mu, and f_tau are free fit parameters subject to f_e + f_mu + f_tau = 1, and no equation in the paper defines one fitted quantity in terms of another. The double-cascade selection uses fixed, predefined observables from simulations, and the observed nine events then constrain the tau fraction, so there is no fitted-input-renamed-as-prediction step. The BPL spectral baseline is imported from a companion IceCube paper (Ref [3]) and other IceCube measurements (Ref [6]); this is a minor self-citation, but it is not load-bearing because the paper explicitly cross-checks with an SPL assumption under which the central 68% closure is also reported, and the best fits and contour shapes remain comparable. The software citation to NNMFit is "paper in preparation" (Ref [7]), a reproducibility gap rather than a circular argument. The manuscript itself flags "large statistical uncertanties in the double cascades histograms" (Section 3) and notes that the 95% contour does not close under BPL; these are Wilks-coverage and model-dependence concerns (especially in the bounded ternary flavor space), which are risks to the claimed 68% milestone but not circularity. No self-citation is invoked to forbid alternative flavor compositions, and no uniqueness or equivalence theorem is imported from the authors' prior work. The result is therefore self-contained as a measurement; score 2 reflects only the non-load-bearing self-citations in the spectral assumption and software chain.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The fit is a measurement of flavor fractions as free parameters, not a first-principles derivation. The main unstated costs are the assumed spectral shape, the validity of Wilks' confidence levels near the boundaries, and the simulated purity of the tiny double-cascade sample.

free parameters (4)
  • Astrophysical flavor fractions f_e and f_tau (f_mu = 1 - f_e - f_tau) = f_e=0.30, f_tau=0.33, f_mu=0.37
    Parameters of interest measured via forward-folded maximum-likelihood fit to the 2D/3D energy-zenith-length histograms.
  • BPL astrophysical spectrum parameters = phi=2.72e-18 /GeV/cm2/s/sr, gamma1=1.76, gamma2=2.81, log10(E_break/GeV)=4.5
    Baseline spectral shape taken from the MESE spectrum fit [3] or refit here; quoted as best-fit values. They affect the predicted event distributions and hence the flavor contours.
  • Atmospheric flux nuisance parameters = not quoted
    Cosmic ray flux and neutrino production uncertainties included as nuisance parameters in the fit.
  • Detector systematics (ice model, DOM optical efficiency) = not quoted
    Nuisance parameters allowing light-acceptance variations in the fit.
assumptions (5)
  • domain assumption Standard three-flavor neutrino oscillations averaged over astrophysical baselines map source flavor ratios to Earth ratios as in [10].
    Used to interpret the Earth-measured ratio and to draw the source-scenario lines in Figure 2.
  • ad hoc to paper The BPL shape with the quoted parameters adequately describes the astrophysical spectrum over 1 TeV to 10 PeV, and the flavor ratio is energy-independent in that range.
    Stated in Section 2: 'This fit is performed under the assumption that the flavor ratio remains the same across the energy scales considered here.' The BPL baseline follows [3].
  • ad hoc to paper Wilks' theorem applies to the likelihood ratio contours on the constrained ternary flavor-fraction space, including near the physical boundaries.
    The confidence levels (68%, 95%, rejection CLs) are computed as Wilks' contours from the maximum-likelihood fit without showing the small-sample or boundary behavior of the test statistic.
  • domain assumption The double-cascade classification variables and cuts, predefined using simulations, have the modeled 70% purity and correctly separate tau-neutrino double cascades from single cascades and tracks.
    The paper reports an expected 7 double-cascade events with 70% purity and observes 9; the tau sensitivity depends on this classification being accurate.
  • domain assumption MESE event selection and the deep neural network morphology classification (88% cascade, 97% track efficiency) are reliable.
    Invoked in Section 1, supported by refs [2] and [4], but not re-derived here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Measurement of the Three-Flavor Composition of Astrophysical Neutrinos with Contained IceCube Events." pith.science (2026). https://pith.science/paper/WY3REU5J

@misc{pith2026250707212,
  author       = {Pith},
  title        = {Pith review of: Measurement of the Three-Flavor Composition of Astrophysical Neutrinos with Contained IceCube Events},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WY3REU5J}},
  note         = {Machine review of arXiv:2507.07212}
}
read the original abstract

The IceCube Neutrino Observatory at the South Pole detects neutrinos from the entire sky, both of astrophysical and atmospheric origin, via the Cherenkov light emitted when these neutrinos interact in the ice, giving rise to rapidly moving charged particles. Neutrino events with vertices contained within the detector volume are useful for studying the neutrino flavor ratio, as they allow for a better reconstruction of the event morphology. The Medium Energy Starting Events (MESE) data sample is a selection of such events with energies of at least 1 TeV. This sample includes electron-, muon-, and tau-neutrino events, processed consistently. We use it to constrain the flavor ratio of astrophysical neutrinos at Earth, which in turn informs us of the flavor composition at the source itself. In this talk, we will present the results of this study, based on 11.4 years of IceCube data.

Figures

Figures reproduced from arXiv: 2507.07212 by the authors.

Figure 1
Figure 1. The 1D projections of the reconstructed energy and reconstructed cosine zenith distributions of cascade, track, and double cascade classified events in MESE. The reconstructed length distribution for double cascade classified events is also shown. Data is shown in black and total best-fit MC is shown in blue. The individual components of the total MC are also shown in the figure. The prompt atmospheric flux fits to … view at source ↗
Figure 2
Figure 2. Fits to the astrophysical flavor ratio with the MESE sample for the baseline assumption of a BPL flux (blue) and for the cross-check with an SPL fit (orange). 68% CL and 95% CL Wilks’ contours are shown here along with the best fit. The best fit from both flux assumptions are consistent with each other. Assumptions of flavor ratio at source, after undergoing oscillations during their propagation towards Earth are al… view at source ↗
Figure 3
Figure 3. A comparison of the flavor ratio measured with the MESE sample to previous measurements made with IceCube [8] [9] [5] is shown. All curves show the 68% CL Wilks’ contours, along with the respective best fit. Previous measurements with IceCube assumed an SPL flux when measuring the flavor ratio, while this work assumes the flux to be a BPL, consistent with IceCube’s latest spectral measurements [3]. Including the dou… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

14 extracted references · 6 canonical work pages

  1. [1]

    IceCube Collaboration, M. G. Aartsen et al. , http://dx.doi.org/10.1088/1748-0221/12/03/P03012 JINST 12 no. 03, (2017) P03012

    doi:10.1088/1748-0221/12/03/p03012 2017

  2. [2]

    Basu and A

    IceCube Collaboration, V. Basu and A. Balagopal V., http://dx.doi.org/10.22323/1.444.1007 PoS. (ICRC2023) 1007

    work page doi:10.22323/1.444.1007

  3. [3]

    IceCube Collaboration, PoS ICRC2025 (these proceedings) 985

    work page

  4. [4]

    Kronmueller and T

    IceCube Collaboration, M. Kronmueller and T. Glauch, http://dx.doi.org/10.22323/1.358.0937 PoS (ICRC2019) 937

    doi:10.22323/1.358.0937

  5. [5]

    Abbasi et al

    IceCube Collaboration, R. Abbasi et al. , http://dx.doi.org/10.1140/epjc/s10052-022-10795-y Eur. Phys. J. C 82 no. 11, (2022) 1031

    doi:10.1140/epjc/s10052-022-10795-y 2022

  6. [6]

    IceCube Collaboration, R. Naab, E. Ganster, and Z. Zhang, http://dx.doi.org/10.22323/1.444.1064 PoS (ICRC2023) 1064

    work page doi:10.22323/1.444.1064

  7. [7]

    Haack, E

    C. Haack, E. Ganster, et al. , `` NNMFit - an aesara-based analysis framework for diffuse analyses.'' paper in preparation

    work page

  8. [8]

    IceCube Collaboration, M. G. Aartsen et al. , http://dx.doi.org/10.1088/0004-637X/809/1/98 Astrophys. J. 809 no. 1, (2015) 98

    doi:10.1088/0004-637x/809/1/98 2015

Show all 14 references
  1. [9]

    IceCube Collaboration, M. G. Aartsen et al. , http://dx.doi.org/10.1103/PhysRevD.99.032004 Phys. Rev. D 99 no. 3, (2019) 032004

    2019 doi

  2. [10]

    N. Song, S. W. Li, C. A. Arg\"uelles, M. Bustamante, and A. C. Vincent, http://dx.doi.org/10.1088/1475-7516/2021/04/054 JCAP 04 (2021) 054

    2021 doi

  3. [11]

    Abbasi et al

    IceCube Collaboration, R. Abbasi et al. , http://dx.doi.org/10.1103/PhysRevLett.132.151001 Phys. Rev. Lett. 132 no. 15, (2024) 151001

    2024 doi

  4. [12]

    R. Clark, http://dx.doi.org/https://kclpure.kcl.ac.uk/portal/en/studentTheses/measuring-high-energy-diffuse-fluxes-at-the-south-pole-with-the-i Measuring high energy diffuse fluxes at the South Pole with the IceCube Neutrino Observatory and DM-Ice17 . PhD thesis, 2025

    2025

  5. [13]

    write newline

    " write newline "" before.all 'output.state := FUNCTION blank.sep after.quote 'output.state := FUNCTION fin.entry output.state after.quoted.block = 'skip 'add.period if write newline FUNCTION new.block output.state before.all = 'skip output.state after.quote = after.quoted.blo...

  6. [14]

    write newline

    " write newline "" before.all 'output.state := FUNCTION blank.sep after.quote 'output.state := FUNCTION fin.entry output.state after.quoted.block = 'skip 'add.period if write newline FUNCTION new.block output.state before.all = 'skip output.state after.quote = after.quoted.blo...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.