Atmospheric neutrinos as probes of neutrino-matter interactions

Friedland, Lunardini and Maltoni have a real result: the first three-flavor treatment of atmospheric NSI, and the observation that the allowed region opens up along ε_ττ = |ε_eτ|²/(1+ε_ee), where the matter eigenstate has zero eigenvalue and the oscillations mimic vacuum. That parabola is the paper's best idea — it explains why order-one ε_ττ survives once ε_eτ is turned on, which the old two-flavor bounds missed. The analytical mapping in Eqs. (5)–(8) is clean, and the agreement with the numerical contours in Fig. 1 is convincing. The fit itself is standard: full SK-I 1489-day data, 55 points, three-dimensional scan. They are also honest that they find no evidence for new physics, only that the data do not exclude it. That framing is right and worth crediting. The soft spot is the far end of the parabola. The paper says the extent along it is 'beyond the scope of our analytical treatment' and that the fit quality is set by 'subdominant NSI effects in all energy samples.' That is an assertion, not a demonstration. The channel that should bite is ν_e appearance: along the parabola the state ν_μ oscillates into is ν_τ' = −s_β e^2iψ ν_e + c_β ν_τ, so there is a ν_e admixture of order s_β². For their own case (iv) (ε_eτ = 0.90, ε_ττ = 0.953), β ≈ 47°, so s_β² ≈ 0.5 — a large multi-GeV e-like excess unless the fit absorbs it through normalization or a shifted oscillation maximum. The paper never shows the e-like prediction at the far end. That is a specific, checkable gap, and a referee should ask for it. Second, ε_µτ = 0 is load-bearing and deferred to an in-preparation reference [7]. The two-flavor analyses put it below 10⁻²–10⁻¹; if nonzero it could shrink the allowed region. They acknowledge this, but it means the 'ε_ττ ~ 1 allowed' claim is conditional on a choice not yet justified in print. None of this kills the result. The parabola insight and the analytical mapping stand on their own, and the fit is a real fit. But the headline number — order-one NSI allowed — rests on the numerical scan, and the paper should show the e-like prediction that backs it up. Who should read it: anyone working on atmospheric neutrinos, NSI, or the interpretation of MINOS and KamLAND. It deserves a serious referee, with the e-like check and the ε_µτ sensitivity as the two things to demand.

The striking part of the central claim — that ε_ττ ~ 1 is allowed along the parabola ε_ττ = |ε_eτ|²/(1+ε_ee), extending to ε_eτ = 1 for inverted hierarchy — is supported only by the χ² fit. The analytical argument (Eqs. 5–8) only shows that the high-energy through-going muon sample is insensitive, because the leading NSI effect cancels by construction; the paper itself says 'the extent of the allowed region along the parabola is beyond the scope of our analytical treatment' and that 'the fit quality is determined by subdominant NSI effects in all energy samples.' The physical channel that should most constrain the far end is ν_e appearance. Along the parabola, Eq. (4) gives ν_τ' = −s_β e^{2iψ} ν_e + c_β ν_τ with tan 2β = 2|ε_eτ|/(1+ε_ee−ε_ττ). For the paper's own case (iv) (ε_ee = −0.15, ε_eτ = 0.90, ε_ττ = 0.953), β ≈ 46.6°, so s_β² ≈ 0.53, and Eq. (8) gives sin²2θ_m ≈ 0.65–0.8. The implied ν_μ→ν_e amplitude is s_β² sin²2θ_m ≈ 0.35–0.4, i.e. a large multi-GeV e-like excess. The paper asserts 'these effects are rather small' but never shows the e-like prediction at the far end of the parabola. If the fit accommodates this only through flux-normalization freedom or an unstated cancellation, the 'ε_ττ ~ 1 allowed' claim is less secure than presented. This is a specific, checkable tension, not a generic worry about systematics.

The reader's weakest_assumption (ε_μτ = 0, θ13 = Δm²_⊙ = 0) is a real scoping caveat, but the paper explicitly scopes its claim to the ε_eτ–ε_ττ plane and notes ε_μτ is already bounded to < 10⁻²–10⁻¹ by prior work, so the claim as stated is not undermined by that choice. The sharper load-bearing concern is internal: the paper's own text concedes that the analytical argument does not cover the region where the claim is most striking, and the numerical fit is the sole support. The ν_e appearance estimate above is a concrete reason to doubt that the far end of the parabola is genuinely consistent with the e-like data, and the paper does not present the e-like prediction there. This is a check the authors could run and report; if it passes, the claim stands, and if it fails, the 'ε_ττ ~ 1 allowed' statement needs qualification. Either way the paper's core contribution — a three-flavor reanalysis showing that the two-flavor bounds on ε_ττ and ε_eτ are relaxed — is real and the analytical/numerical agreement in Fig. 1 is convincing for the region near the origin. The verdict remains CONDITIONAL: the work is sound in its scoped claim but the most dramatic extrapolation is not independently cross-checked. I therefore recommend UNCHANGED rather than a move to ACCEPT or REJECT, and I agree only partially with the reader because the ε_μτ caveat, while acknowledged, is not the sharpest soft spot.

If it holds, the result is a useful and physically interesting extension of the two-flavor NSI literature: it identifies a flat direction in the three-flavor NSI parameter space along which atmospheric data lose sensitivity, and it correctly notes that the resulting shift in the best-fit oscillation parameters is a falsifiable signature for MINOS and for future long-baseline experiments. The analytic mapping between the zero-eigenvalue condition and the numerical contours is a genuine strength, and the paper is commendably explicit that the far end of the allowed region is not covered by the analytic treatment. The main caveats are the fixed ε_μτ = 0 and the reliance on the numerical fit for the order-one NSI region, both of which the authors acknowledge.

Atmospheric muon neutrinos are believed to oscillate into tau neutrinos, and the standard oscillation picture fits the Super-Kamiokande data well. This paper asks what happens if the tau neutrino also feels an extra force with ordinary matter, one comparable in strength to the Standard Model weak interaction. Working in three flavors and fitting the full 1489-day charged-current data set — 55 data points across contained, stopping, and through-going muon samples — the authors find that such non-standard interactions are not excluded, provided the electron–tau and tau–tau couplings lie on a particular curve. Along that curve one of the matter eigenstates has zero eigenvalue, so the muon neutrino keeps oscillating with an essentially vacuum-like probability, only into a rotated mixture of electron and tau flavor. The consequence is that a region of order-one new physics survives the fit, and if it is real, the mixing angle inferred from the data would be smaller and the mass-squared splitting larger than a standard-only analysis would report.

What carries the argument. The load-bearing object is the condition that one matter eigenvalue vanish, $|epsilon_{tau tau} = |epsilon_{e tau}|^2/(1+|epsilon_{ee})$. The matter Hamiltonian, once rotated in the electron–tau subspace, has eigenvalues $lambda_{e',tau'} = sqrt{2} G_F n_e (1 + |epsilon_{ee} + |epsilon_{tau tau} pm sqrt{(1+|epsilon_{ee}-|epsilon_{tau tau})^2 + 4|epsilon_{e tau}|^2})/2$. When the lighter of these is well below $Delta m^2/(2E_0)$ while the heavier is well above it — the hierarchy the paper calls case (a) — the muon neutrino oscillates freely into the light combination $|nu_{tau'} = -s_beta e^{2i psi} |nu_e + c_beta |nu_tau$ while the heavy state is suppressed. In that regime the survival probability keeps the vacuum form $sin^2 2theta_m sin^2[Delta m^2_m L/(4E)]$, with effective mixing and splitting given in terms of the vacuum angle and the matter angle $beta$. This is why the high-energy through-going muon sample, which normally pins down the tau-tau coupling, loses its discriminating power: the leading matter effect is cancelled by construction, and only subdominant effects in the lower-energy samples constrain the fit.

Core claim. The paper's claim is that moving from a two-flavor $|epsilon_{mu tau}|$ analysis to a three-flavor analysis with $|epsilon_{ee}|$, $|epsilon_{e tau}|$, and $|epsilon_{tau tau}|$ changes the conclusion qualitatively: interactions as strong as the Standard Model's own matter effects can be consistent with atmospheric data. The earlier two-flavor bound, roughly $|epsilon_{tau tau}| < 0.15$ at 99% C.L., is relaxed when $|epsilon_{e tau}|$ is also allowed to vary, so that even $|epsilon_{tau tau}|$ of order one is permitted. The numerical fit bears this out: at $|epsilon_{ee}| = -0.15$ the minimum sits at $|epsilon_{e tau}| = 0.07$, $|epsilon_{tau tau}| = 0.01$ with $chi^2_{min} = 48.50$, against $chi^2_{orig} = 48.57$ for no non-standard interactions at all, and the allowed region stretches along the zero-eigenvalue curve out to $|epsilon_{e tau}| = 1$ for inverted hierarchy. When only $|epsilon_{e tau}|$ is nonzero the authors quote $|epsilon_{e tau}| < 0.38$ at 99% C.L. and $< 0.5$ at $3.6sigma$. Because the data are fitted with standard oscillations only, the presence of such interactions would bias the extracted parameters: the best-fit mixing angle moves below $pi/4$ and the best-fit $Delta m^2$ moves up, a trend the authors derive analytically and confirm numerically.

The premise it rests on. The analysis sets the muon-flavored couplings, especially the muon–tau one, to zero, along with the small mass splitting and the third mixing angle, so the large allowed region depends on those terms being negligible; the authors state that corrections from a nonzero muon–tau coupling will be presented elsewhere.

MINOS, with a 735 km baseline and negligible matter effects, will measure the vacuum mixing angle; a result near $sin^2 2theta \sim 0.3$ together with a low $Delta m^2$ would be an indication of non-standard interactions rather than a failure of oscillation theory.

A future long-baseline experiment at a few thousand kilometers and $E_ν \sim 20$–$30$ GeV would be directly sensitive to the matter terms, so a disagreement between it and MINOS would be a signature of the same effect.

The solar neutrino interpretation is affected: the LMA-0 solution obtained with nonzero $|epsilon_{e tau}$ is compatible with the atmospheric data, so a joint atmospheric, solar, and KamLAND fit becomes the natural next step.

A modified neutral-current cross section from large $|epsilon_{e tau}$ and $|epsilon_{tau tau}$ could in principle show up as an excess of neutral-current events, though the size depends on axial couplings that do not enter the propagation Hamiltonian.

The allowed region is a thin surface in coupling space rather than a volume, so a modest independent handle on any one coupling — from solar data, from a long-baseline run, or from a neutral-current measurement — could collapse the whole degeneracy quickly.

Because the effect is a cancellation in the matter Hamiltonian rather than a new oscillation frequency, it would be invisible in a fit that only looked at the high-energy through-going muon sample; the subdominant multi-GeV and stopping samples are doing the real constraining here.

If the true mixing angle is measurably below maximal, the natural reading is not that the standard three-flavor picture fails but that the matter term has an unmodelled piece; this reframes a non-maximal angle as a possible positive signal rather than a tension.

The same zero-eigenvalue trick could be searched for in reactor or accelerator antineutrino data, where the sign of the matter term flips and the corresponding cancellation curve would sit at different couplings.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review