{"id":"e3a7ff35-4059-4614-99c3-25c5b058d845","arxiv_id":"2412.13696","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"New eta-based parametrisations of the ISS(3,3) show that Z-penguin flavour violation can reveal non-degenerate heavy sterile mixing even when radiative decays are absent.","lead":"This paper develops new mathematical forms of the inverse seesaw model that directly encode low-energy flavour data, then uses them to map which rare lepton decays future experiments should watch for. The central lesson: seeing muon-to-three-electron decay without muon-to-electron-photon decay would hint at non-degenerate, mixed heavy sterile neutrinos.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Residual V2-induced dipole from non-degenerate heavy masses is asserted rather than computed; the headline pattern survives only if this residual BR(μ→eγ) stays below MEG II sensitivity wherever μ→3e or μ-e conversion is visible.","rationale":"The genuinely new result of the paper is not the parametrisation itself, which the authors correctly state is formally equivalent to earlier fixed-texture approaches, but the claimed observable pattern: null μ→eγ together with visible μ→3e or μ-e conversion selects a mixed, non-degenerate heavy spectrum. That pattern requires the dipole amplitude to decouple from V2 and from heavy mass splittings. The authors are explicit that this is approximate (Section 4.2), so the open question is numerical: is the residual dipole small enough compared with future sensitivities? A rough estimate for the 10%-splitting benchmark suggests the residual is suppressed relative to the Z-penguin combination by roughly (Δx/x)(log x)/x, i.e. of order a few percent for M0 around 1 TeV, which would put BR(μ→eγ) near 10^-14 for large θ12—below MEG II but not by a huge margin. Because the claim is framed as a generic hint rather than a theorem, a dedicated exact calculation for the benchmark and for larger splittings is the load-bearing check. If it confirms the suppression, the conditional verdict stands; if it finds overlap between visible μ→3e/conversion and residual μ→eγ above 6×10^-14, the inference in the abstract must be qualified to small splittings. Secondary weaknesses noted by the reader—overlapping authorship in numerical input references, the near-tautological relation between η_eμ and the μ→eγ bound, and the absence of released code—do not independently threaten the central claim, so the verdict should remain CONDITIONAL rather than move further.","tokens_in":18986,"tokens_out":9192,"duration_ms":85460,"concrete_test":"Compute the exact one-loop BR(μ→eγ) for the ISS(3,3) using the physical heavy mass eigenstates and the full Gγ(x) of Eq. (28), not the degenerate approximation Eq. (33). Set V1=1, off-diagonal η=0, diagonal ηii at their Eq. (32) maxima, MR=diag(0.9,1,1.1)M0, and V2 a real 12-rotation with angle θ12. Scan M0 ∈ [0.1,10] TeV and sinθ12 ∈ [10^-4,1], and overlay BR(μ→eγ) with the μ→3e and μ-e conversion contours of Figure 4. Repeat for splittings e.g. diag(0.5,1,1.5)M0 and diag(0.1,1,10)M0 to map where the approximation degrades. The concern is settled if no region with BR(μ→3e) or CR(μ-e) above the quoted future sensitivities also has BR(μ→eγ) above the MEG II future sensitivity 6×10^-14.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central inference—that with off-diagonal η_ij=0 the photon-dipole operator is switched off while Z-penguin operators remain active—is only controlled in the degenerate-mass limit. For the non-degenerate benchmark MR=diag(0.9,1,1.1)M0, a non-trivial V2 generates a residual dipole even when η_eμ=0. With V1=1 and a 12-rotation in V2, the dipole combination entering BR(μ→eγ) is approximately (1/2) y_e y_μ sinθ cosθ [Gγ(x1)-Gγ(x2)] in place of the zero one would get from Eq. (33). The paper states in Section 4.2 that this is true 'to a very good approximation' and that μ→eγ is then 'negligible by construction', but no exact contour for BR(μ→eγ) is shown in Figure 4 and no estimate of the residual rate is provided. Since Gγ(x)→1/2 with corrections of order (3/2) log(x)/x, the residual is not identically zero; its size relative to the MEG/MEG II sensitivity and relative to the μ→3e and μ-e conversion rates determines whether the proposed smoking-gun pattern actually holds. If the residual dipole lies above 6×10^-14 anywhere that μ→3e or CR(μ-e) is visible, the claimed disentanglement fails in that region; if it lies below over the whole visible region, the claim is confirmed.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the charged-lepton flavour phenomenology of the Inverse Seesaw with three generations of sterile fermions (ISS(3,3)). Its technical core is a set of new parametrisations of the Dirac Yukawa matrix Y_D based on polar (Eq. 12), QR (Eq. 18) and singular-value (Eq. 21) decompositions, in which the deviation from unitarity of the effective PMNS matrix (the η-matrix of Eq. 5) is imposed directly as input, while the mixing within the heavy sterile sector is confined to a 'hidden' unitary matrix V_2 (Eqs. 25-27). After deriving the algebraic identities connecting η to the Yukawa couplings, the authors use the new parametrisations to scan the ISS(3,3) parameter space and compare the reach of low-energy cLFV searches (μ→eγ, μ→3e, μ-e conversion, τ decays), Z-pole LFUV observables, and the proposed μTRISTAN collider. The central result is a proposed disentanglement: with off-diagonal η_ij set to zero, radiative decays ℓα→ℓβγ are controlled solely by η_αβ via BR(ℓα→ℓβγ) ≈ (3α/2π)|η_αβ|² (Eq. 33), while Z-penguin dominated processes (μ→3e, μ-e conversion in nuclei) can still receive sizeable contributions from V_2 mixing combined with a non-degenerate heavy spectrum (benchmark M_R = diag(0.9, 1, 1.1) M_0, Figure 4). The paper concludes that in the absence of radiative decays, sizeable rates for Z-penguin dominated observables could hint at a non-trivially mixed and non-degenerate heavy spectrum (Abstract and Section 5).","tokens_in":19288,"tokens_out":55793,"duration_ms":446102,"significance":"The disentanglement proposal is the paper's most valuable contribution. If it survives quantitative scrutiny, it provides a falsifiable diagnostic—measurable by MEG II, Mu3e, COMET and Mu2e—that separates active-sterile flavour mixing (η) from mixing within the heavy sterile sector (V_2). The algebraic identities of Section 2 are standard matrix decompositions that can be verified by hand, and the SVD parametrisation is a practical improvement over the modified Casas-Ibarra form: the hidden parameters enter through bounded angles rather than hyperbolic functions, and the global η bounds of Ref. [40] can be imposed directly. The phenomenological analysis treats a broad set of observables with consistent current and projected sensitivities, and the V_2-induced predictions of Section 4.2 are concrete and testable. The main caveats are that no numerical code is provided, the residual V_2-induced photon-dipole contribution is asserted rather than computed (Major Comment 1), and the normalisation of the parametrisations relative to Eq. (6) is ambiguous as printed (Major Comment 2).","major_comments":[{"comment":"The claim that setting off-diagonal η_ij = 0 switches off the photon dipole is exact only in the degenerate-mass limit and is left unquantified for the benchmark M_R = diag(0.9, 1, 1.1) M_0, where the paper states that 'the rate of the radiative decay μ → eγ is negligible by construction'. With V_1 = 1 and a 12-rotation in V_2, the GIM-violating dipole amplitude is (y_e y_μ/2) sinθ12 cosθ12 [Gγ(x2) − Gγ(x1)], with y_i = √(2η_ii) as stated after Eq. (24) and Gγ given by Eq. (28). Using Gγ(x) ≈ 1/2 − (3/2) log(x)/x and the maximal diagonal η values of Eq. (32), I estimate the residual BR(μ→eγ) ≈ (3α/2π)|(y_e y_μ/2) sinθ12 cosθ12 [Gγ(x2) − Gγ(x1)]|² to be of order 10^-14 at M_0 = 1 TeV, rising to order 10^-13 at M_0 ≈ 300 GeV for sinθ12 ≈ 0.3, i.e., at or above the MEG II projection of 6 × 10^-14 in the lower part of the mass range of Figure 4. For the 9-11 TeV masses of Figure 5 the residual is instead negligible (order 10^-17), so the concern is specific to the low-mass end of Figure 4, where the displayed μ→3e and μ-e conversion constraints are nevertheless sizable. The authors should add the μ→eγ contours, or an analytic estimate of the residual, over the full M_0 range of Figure 4, and should restrict the disentanglement claim in the Abstract, in the Introduction ('only observables that depend on Z-penguin transitions can be affected'), and in Section 5 to the region where the residual is below the quoted sensitivity. The logarithmic enhancement of the Z-penguin relative to the asymptotically flat dipole makes the proposed pattern plausible for M_0 ≳ 1 TeV, so I regard this as a missing calculation rather than a disproof of the claim.","section":"Section 4.2, Figure 4"},{"comment":"As printed, the parametrisations are not consistent with Eq. (6) under the stated relation m_D = v Y_D/√2. Inserting Eq. (27) into Eq. (6) gives η = (1/4) V_1^* diag(y²) V_1^T, a factor of two below Eq. (24); likewise, Eqs. (25) and (26) yield η = η_input/2 when evaluated through Eq. (6). The identities in Eqs. (13), (14), (19) and (24) are internally consistent; the discrepancy appears only when Eq. (6) is applied to the explicit Y_D displayed in Eqs. (25)-(27). The stated translation y_i = √(2η_ii) is compatible with Eq. (24), but requires either doubling the coefficients in Eqs. (25)-(26) (√2/v → 2/v in Eq. (25); 1/v → √2/v in Eqs. (26) and (27)) or dropping the 1/√2 factor in the m_D definition for these parametrisations. Because y_i enters the perturbativity bounds in Figures 1-5 and the V_2-induced amplitudes in Figures 4-5, and because the plotted η values must correspond to the physical η of Eq. (6) rather than half of it, the convention should be stated unambiguously and the numerical implementation checked against it.","section":"Section 2, Eqs. (6), (24)-(27)"}],"minor_comments":[{"comment":"Since BR(ℓα→ℓβγ) ≈ (3α/2π)|η_αβ|² is the defining relation of the scan variable, the μ→eγ exclusion contour in the left panel of Figure 3 is a direct translation of the experimental bound onto |η_eμ| and carries no model information beyond Eq. (33); the informative content of the figure lies in the comparison with the Z-penguin and box observables. I suggest stating this explicitly in the text and drawing the μ→eγ curve as a reference rather than as an independent constraint.","section":"Section 4.1, Eq. (33), Figure 3"},{"comment":"The inequality in Eq. (15) and the discussion around it are attributed to 'Schwartz'; the correct spelling is 'Schwarz' (Hermann Schwarz). The same typo appears near Eq. (17).","section":"Section 2, Eq. (15)"},{"comment":"The algorithm in Appendix A is attributed to 'Cholesky-Banachiwiecz'; the correct spelling is 'Cholesky-Banachiewicz'.","section":"Appendix A"},{"comment":"The phases δij appearing in the parametrisation of V_2 (Eq. (22)) are not specified for the scans of Figures 4 and 5. Please state whether they are all set to zero, since the interference effects discussed in Figure 5 (e.g., the mutual suppression of the μ-e rates for large sinθ13) depend on these phases.","section":"Figures 4-5"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the manuscript is within the journal's scope and the reference practice is appropriate. The μTRISTAN projections are taken from the companion paper [55] by the same group and have not been independently checked here. The two major points above are repairable: (i) the authors should add the μ→eγ contours, or an analytic bound on the residual dipole, to Figure 4 (and state where the residual falls below the quoted sensitivity) and recalibrate the wording of the Abstract and conclusions accordingly; (ii) they should fix or explicitly define the √2 normalisation of Eqs. (25)-(27) relative to Eq. (6) and verify that their numerical implementation realises the stated η rather than η/2. Since no scan code or analytic amplitude files are provided, I would encourage asking for the relevant implementation as supplemental material in the revision, which would allow the residual-dipole estimate and the normalisation issue to be checked directly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the paper does something genuinely useful: it writes down three explicit parametrisations of the ISS(3,3) Yukawa couplings (polar, QR/Cholesky, SVD) that encode the non-unitarity matrix η directly and isolate the heavy-sector mixing V2. That is clean, internally consistent, and a real technical step up from the fixed textures used in earlier work. Second, the headline phenomenological claim — that with η_ij = 0 you can have sizeable μ→3e and μ-e conversion while μ→eγ is absent, signalling a mixed non-degenerate heavy spectrum — is not actually established. The dipole suppression is only exact in the degenerate-mass limit; for the benchmark MR = diag(0.9, 1, 1.1) M0, a residual μ→eγ amplitude appears from the spread in Gγ(x_k). The paper says this is negligible \"to a very good approximation\" and \"by construction\", but shows no contour and gives no bound on the residual relative to MEG II. That is the real gap.\n\nWhat the paper does well: the algebraic identities (Eqs. 13, 19, 24) are standard and check out. The parametrisations are genuinely new in the systematic treatment of V2 and in making the connection to η explicit. The scan results in Figures 4 and 5 clearly illustrate the Z-penguin enhancement and the interplay between the V2 angles. The paper is honest that the parametrisations are formally equivalent to earlier approaches, and the use of overlapping references for numerical inputs (Z-pole observables from [23], μTRISTAN reach from [55]) is not by itself a flaw.\n\nSoft spots, in proportion: the μ→eγ constraint in Figure 3 is essentially a restatement of the η_eμ scan variable via Eq. (33), so it is not independent of the input. No code or data files are provided, which is minor but would help. The missing residual-dipole estimate is the one substantive issue: without it, the central claim is a conjecture backed by an approximation. The stress-test note is on point here.\n\nWho this is for: flavour phenomenologists working on low-scale seesaw models and cLFV. A serious referee should engage with it; the parametrisations alone justify the effort. The paper should be sent to peer review, with the residual dipole required as an explicit contour or bound before acceptance.","headline":"Useful new parametrisations of the ISS(3,3), but the headline 'no μ→eγ with sizeable μ→3e' rests on an unquantified residual-dipole approximation that needs a numerical check.","tokens_in":19994,"tokens_out":10491,"would_cite":true,"duration_ms":92182,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"In the inverse seesaw, the absence of muon-to-electron gamma decay together with sizeable muon-to-3e or conversion rates would be a sign of a non-degenerate, mixed heavy sterile spectrum.","keywords":["inverse seesaw","lepton flavour violation","non-unitarity","heavy neutral leptons","Z-penguin","radiative decays","muon to electron conversion","neutrino mass model"],"falsifier":"Compute the exact one-loop photon-penguin amplitude for the benchmark heavy spectrum $M_R=\\mathrm{diag}(0.9,1,1.1)M_0$ with a non-trivial $V_2$; if the resulting $\\mathrm{BR}(\\mu\\to e\\gamma)$ is not suppressed by several orders of magnitude relative to $\\mathrm{BR}(\\mu\\to3e)$ and the $\\mu$--$e$ conversion rate, the clean separation fails. Experimentally, a measurement of $\\mu\\to e\\gamma$ at future sensitivity in a parameter region the model predicts to be silent would falsify the disentanglement claim.","tokens_in":18628,"feed_emoji":"⚛️","tokens_out":16072,"duration_ms":118380,"temperature":0.7,"pith_summary":"The paper establishes that, within the inverse seesaw with three generations of sterile fermions, the sources of charged lepton flavour violation can be disentangled. It develops new parametrisations of the Yukawa couplings that encode the non-unitarity of the effective PMNS matrix (the $\\eta$-matrix) directly, so that low-energy bounds can be imposed by construction. Using these, it shows that radiative decays $\\ell_\\alpha\\to\\ell_\\beta\\gamma$ are, to a very good approximation, controlled only by the off-diagonal $\\eta_{\\alpha\\beta}$, while Z-penguin observables such as $\\mu\\to 3e$ and $\\mu$--$e$ conversion also depend on the mixing $V_2$ among the heavy states and on the heavy mass splittings. A future experimental pattern with no $\\mu\\to e\\gamma$ but sizeable $\\mu\\to 3e$ or conversion would therefore indicate a non-degenerate, non-trivially mixed heavy sector.","feed_headline":"Absent mu-to-e-gamma, mu-to-3e could expose heavy neutrino mixing","feed_subtitle":"If true, a future pattern of no mu-to-e-gamma but sizeable mu-to-3e singles out a non-degenerate heavy spectrum","key_machinery":"The key machinery is a set of three equivalent parametrisations of the Yukawa matrix $Y_D$ (polar decomposition, Cholesky/QR decomposition, and singular value decomposition) in which the active-sterile mixing is encoded in the $\\eta$-matrix, the deviation from unitarity of the effective PMNS matrix, and the mixing among heavy states is isolated in a unitary factor $V_2$ (Eqs. (25)--(27)). This algebraic separation, combined with the asymptotic loop functions $G_\\gamma(x)\\to 1/2$ and $F_Z(x)\\to 5/2-\\frac{5}{2}\\log x$ for $x\\gg1$, is what splits observables into those that only feel $\\eta_{\\alpha\\beta}$ (radiative decays) and those that additionally feel $V_2$ and the heavy mass splittings (Z-penguin processes). The split is exact in the degenerate limit and approximate for a non-degenerate spectrum.","core_discovery":"The central claim is that the inverse seesaw with three generations of sterile fermions, the ISS(3,3), can generate sizeable rates for Z-penguin dominated processes $\\mu\\to3e$, neutrinoless $\\mu$--$e$ conversion, and some collider signals even when the off-diagonal entries of the $\\eta$-matrix are set to zero, provided the heavy pseudo-Dirac pairs are non-degenerate and their mixing $V_2$ is non-trivial. In that configuration the radiative decays $\\ell_\\alpha\\to\\ell_\\beta\\gamma$ remain negligible, because the photon-dipole amplitudes depend essentially only on $\\eta_{\\alpha\\beta}$. This is demonstrated with a benchmark $M_R=\\mathrm{diag}(0.9,1,1.1)M_0$ and one or two non-zero angles in $V_2$, where $\\mu\\to3e$ and $\\mu$--$e$ conversion lie within future reach while $\\mu\\to e\\gamma$ is switched off by construction. The paper concludes that a future null result for $\\mu\\to e\\gamma$ combined with a positive signal in the Z-penguin observables would hint at a non-trivially mixed and non-degenerate heavy spectrum.","pith_inferences":["A direct extension of the disentangling logic suggests that a signal in $\\tau\\to3\\ell$ without $\\tau\\to\\ell\\gamma$ at a future tau-factory or high-luminosity Z-factory would similarly point to heavy-sector mixing, giving a flavour-universal diagnostic.","The approximate nature of the 'switched-off' radiative decay invites a dedicated higher-order computation of the mass-splitting corrections to the dipole amplitude; if those corrections are comparable to the leading Z-penguin rates for splittings around 10%, the clean experimental discriminant would be blurred.","Model builders aiming to evade $\\mu\\to e\\gamma$ bounds while predicting observable $\\mu\\to3e$ could use this parametrisation as a target: arrange for a non-degenerate, mixed heavy sector with small off-diagonal $\\eta$.","The ratio $\\mathrm{BR}(\\mu\\to3e)/\\mathrm{BR}(\\mu\\to e\\gamma)$ measured in a single future run could serve as a quantitative probe of the heavy mass splitting in the ISS(3,3), should the model be realised."],"forward_implications":["If the central claim is correct, a future observation of $\\mu\\to3e$ or $\\mu$--$e$ conversion without $\\mu\\to e\\gamma$ would be a smoking gun for a non-degenerate, mixed heavy sterile spectrum in the ISS(3,3).","The new $\\eta$-based parametrisations allow a controlled exploration of the ISS(3,3) parameter space that automatically respects low-energy universality bounds, avoiding the numerical instability of the standard seesaw parametrisation.","A high-luminosity $\\mu^+e^-$ collider would provide the strongest future constraints on $\\tau$--$e$ and $\\tau$--$\\mu$ flavour violation, outperforming Z-pole searches, while dedicated muon facilities remain the best probes of the $\\mu$--$e$ sector.","Simultaneous presence of two heavy-mixing angles in $V_2$ can induce new flavour-violating directions (for example, $\\theta_{13}$ together with $\\theta_{23}$ generates $\\mu\\to3e$), so the pattern of future cLFV signals encodes information about the heavy mixing matrix."],"supporting_citations":[{"why":"Supplies the model-independent bounds on the diagonal entries of the non-unitarity matrix used to fix the $\\eta_{ii}$ to their maximal allowed values.","marker":"[40]"},{"why":"Provides the loop functions $G_\\gamma$ and $F_Z$ and their asymptotic limits that separate radiative decays from Z-penguin observables.","marker":"[41]"},{"why":"Gives the one-loop predictions for Z-pole observables in low-scale seesaw realisations, used for the future collider projections and the rescaled parametrisation.","marker":"[23]"},{"why":"Gives the projected cLFV sensitivities at a high-luminosity $\\mu^+e^-$ collider, used for the heavy-mixing benchmark plots.","marker":"[55]"},{"why":"The seesaw parametrisation that incorporates neutrino oscillation data; serves as the baseline whose sources of flavour violation are entangled.","marker":"[30]"},{"why":"Introduces the $\\eta$-matrix as the deviation from unitarity of the effective PMNS matrix, the central object of the new parametrisations.","marker":"[32]"},{"why":"Proposes a rescaled parametrisation of the Dirac mass matrix in terms of a unitary matrix and the heavy mass matrix, a direct precursor to the parametrisations developed here.","marker":"[36]"},{"why":"Earlier construction of textures for the Dirac mass matrix aimed at maximising specific flavour-violating directions, generalised in the present work.","marker":"[35]"}],"fun_headline_variants":["No mu-to-e-gamma, but mu-to-3e? Heavy neutrino mixing exposed","Mu-to-3e without mu-to-e-gamma: probe of non-degenerate heavy neutrino sector","If mu->e gamma stays silent, mu->3e exposes heavy neutrino mixing","No radiative decay? Mu-to-3e could signal non-degenerate heavy neutrinos","Mu->3e as Z-penguin probe when mu->e gamma is absent"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the photon-loop piece of radiative decays is controlled by the active-heavy mixing alone, so setting that mixing to zero silences those decays while Z-penguin processes stay sensitive to the heavy states' internal mixing and mass differences.","fun_headline_variants_meta":{"raw":{"variants":["No mu-to-e-gamma, but mu-to-3e? Heavy neutrino mixing exposed","Mu-to-3e without mu-to-e-gamma: probe of non-degenerate heavy neutrino sector","If mu->e gamma stays silent, mu->3e exposes heavy neutrino mixing","No radiative decay? Mu-to-3e could signal non-degenerate heavy neutrinos","Mu->3e as Z-penguin probe when mu->e gamma is absent"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001445,"raw_usage":{"total_tokens":5850,"prompt_tokens":999,"completion_tokens":4851,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":615,"completion_tokens_details":{"reasoning_tokens":4737}},"tokens_in":615,"tokens_out":4851,"duration_ms":30440,"temperature":1.0,"reasoning_tokens":4737,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T12:54:38.711984+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the exact one-loop photon-penguin amplitude for the benchmark heavy spectrum $M_R=\\mathrm{diag}(0.9,1,1.1)M_0$ with a non-trivial $V_2$; if the resulting $\\mathrm{BR}(\\mu\\to e\\gamma)$ is not suppressed by several orders of magnitude relative to $\\mathrm{BR}(\\mu\\to3e)$ and the $\\mu$--$e$ conversion rate, the clean separation fails. Experimentally, a measurement of $\\mu\\to e\\gamma$ at future sensitivity in a parameter region the model predicts to be silent would falsify the disentanglement claim.","supporting_citations":[{"cited_title":"Charged lepton-flavor violating processes and suppression of nonunitary mixing effects in low-scale seesaw models","cited_arxiv_id":"2302.07379","evidence_quote":"Proposes a rescaled parametrisation of the Dirac mass matrix in terms of a unitary matrix and the heavy mass matrix, a direct precursor to the parametrisations developed here."}],"review_version":1}