{"id":"296d684b-acc2-42de-b05f-dd9b98382fc7","arxiv_id":"2507.17299","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"ADITYA-U tokamak data reveals a distinct disruption class, Accelerated Mode Disruption, marked by rising 2/1 mode frequency followed by a sudden frequency collapse and a faster current quench than locked-mode disruptions.","lead":"A team from India's ADITYA-U tokamak reports a new kind of disruption, called Accelerated Mode Disruption, where a magnetic instability spins faster and then suddenly crashes, causing a faster, more violent current drop than the usual locked-mode disruption. The finding offers new warning signs, such as edge safety factor and current decay, that could improve disruption prediction for tokamaks.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"AMD/LMD classification in Sec. 3.1 is qualitative; without an objective, reproducible labeling rule for the 150-shot sample, the reported thresholds and regime separation may be artifacts of manual selection.","rationale":"The reader's weakest-assumption analysis identifies the same load-bearing issue: the AMD/LMD labeling in Sec. 3.1 is qualitative, and the statistical thresholds in Sec. 3.2 depend on that labeling. I agree with this assessment, and I do not find a more fundamental objection that would change the verdict from CONDITIONAL. The representative discharges are described in detail, with Mirnov, SVD, spectrogram, bolometer, and SXR data, which gives some support to the existence of the two-phase frequency behavior; however, the generalization from two examples to a statistical separation of 150 shots requires a reproducible classification rule. Without it, the reported cutoffs could reflect selection bias rather than independent physical boundaries. I also note that the pressure-gradient derivation in Sec. 6 contains dimensional inconsistencies (e.g., '∇p_e = 2.6e4/1.6e-19 = 1.625e23 Pa/m' mixes eV and SI units and does not yield a pressure gradient), and Figure references are sometimes inconsistent (Fig. 3 vs Fig. 4 in Sec. 3.3). These are real correctness concerns, but they pertain mainly to the proposed mechanism and presentation, whereas the classification issue is more fundamental to the central claim of a distinct disruption regime. For this reason, the requested revision should prioritize a quantitative, pre-registered classification rule and full shot list; if that check fails, the regime identification is not supported. If it succeeds, the remaining mechanism concerns are addressable as corrections rather than fatal flaws.","tokens_in":13597,"tokens_out":2749,"duration_ms":31136,"concrete_test":"Obtain from the authors the full list of 150 shot numbers and an explicitly quantitative, pre-specified classification rule, e.g.: over the precursor window [-10 ms, 0] relative to Ipd, compute the linear slope of the dominant Mirnov spectral peak; label AMD if the slope is > +0.1 kHz/ms and the amplitude is within 20% of its pre-collapse maximum before the final frequency collapse, otherwise label LMD; then recompute the q_edge, current-decay, CQ-time, and CQ-rate distributions. If the reported separation thresholds are not reproduced within stated uncertainties, or if two independent blinded annotators applying the same rule achieve Cohen's kappa < 0.7 on the same traces, the claimed regime distinction is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that AMD is a distinct disruption regime separable by q_edge > 4.3, current decay > 16%, CQ time < 1.5 ms, and CQ rate > 56 MA/s—rests entirely on assigning 150 discharges to AMD or LMD in Sec. 3.1 and 3.2. The paper contrasts two representative traces (#37103 with rising frequency vs #36450 with decaying frequency) and states that a statistical analysis was performed, but it never specifies a quantitative classification rule, the inclusion/exclusion criteria for the 150-shot set, the shot list, or an inter-rater reliability check. Because the thresholds are computed on this manually labeled sample, any systematic preference for clean or extreme examples would imprint the observed separation. In particular, 'steady rise in mode frequency with nonlinearly saturated amplitude' is not defined with measurable tolerances (e.g., slope, window, saturation level), so it is unclear whether a borderline discharge would be labeled AMD or LMD, or excluded. The reported percentages (73%, 80%, 75%, 76%) and cutoffs are therefore properties of the labeling procedure unless the labels can be reproduced from an explicit rule. No independent diagnostic—such as the SXR temperature hollowing or bolometer radiation profile—is used to define the classes, so the new-regime claim lacks an external anchor to the classification. This is load-bearing because every downstream statistical statement and the proposed mitigation strategy depend on AMD being a real, consistently identifiable population.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the identification of a new disruption regime in the ADITYA-U tokamak, termed Accelerated Mode Disruption (AMD), and contrasts it with conventional Locked Mode Disruption (LMD). The distinguishing signature is a precursor phase with a steadily rising mode frequency and nonlinearly saturated amplitude, followed by a sudden frequency collapse and amplitude growth. A statistical analysis of 150 discharges is claimed to yield empirical thresholds in edge safety factor (q_edge > 4.3), normalized current decay coefficient (>16%), current quench time (<1.5 ms), and current quench rate (>56 MA/s) that separate AMD from LMD. The paper further proposes that core temperature hollowing causes steepening of pressure and current density gradients near the 2/1 rational surface, destabilizing the drift-tearing mode and producing the observed frequency behavior.","tokens_in":13867,"tokens_out":8514,"duration_ms":78038,"significance":"If substantiated, the existence of a distinct disruption class with faster and more dangerous current quenches would be valuable for disruption prediction and mitigation, particularly for future devices. The paper leverages a reasonably large dataset (150 discharges) and multiple diagnostics (Mirnov coils, SXR, bolometry, Langmuir probes), and the primary classification signal (rising versus decaying precursor frequency) is in principle independent of the variables used for the reported thresholds. These are notable strengths. However, the quantitative support is currently inadequate: the classification rule is not specified in a reproducible way, the thresholds lack error bars, and the mechanism section contains circular reasoning and a unit inconsistency. The paper's central claim is therefore not yet convincingly established, but it is potentially salvageable with a thorough revision.","major_comments":[{"comment":"The classification rule used to label the 150 discharges as AMD or LMD is not specified. The text describes AMD as exhibiting 'a steady rise in mode frequency with a nonlinearly saturated amplitude' (Sec. 3.1), but it does not define a quantitative criterion: no time window, frequency slope threshold, amplitude saturation level, or analysis method is given. The inclusion and exclusion criteria for the 150-shot sample, the shot list, and any inter-rater or independent diagnostic checks are also absent. Because the statistical thresholds in Sec. 3.2 (q_edge > 4.3, current decay > 16%) are computed on this manually labeled sample, they may be artifacts of the labeling procedure rather than independent physical boundaries. Please provide an explicit, reproducible classification algorithm or detailed quantitative criteria, and re-derive the thresholds with confidence intervals.","section":"Sec. 3.1-3.2"},{"comment":"The claim that the electron pressure gradient dominates the DTM frequency is circular. The diamagnetic contribution of 7.8 kHz is obtained as the residual after subtracting the flow contributions (1.7 kHz toroidal and 0.5 kHz poloidal) from the measured total frequency of approximately 10 kHz (Sec. 5); it is not an independent measurement of a pressure-gradient-driven term. Using that residual to conclude that 'the pressure gradient near the mode rational surface plays a dominant role' (Sec. 5) is reasoning in a circle. Furthermore, the calculation in the bullet list is dimensionally inconsistent: the text first states that ∇p_e should be 2×10^4 Pa/m, but then writes '∇p_e = ∇(n_e T_e) = 2.6×10^4 / 1.6×10^-19 = 1.625×10^23 Pa/m', which is not a valid pressure-gradient computation. Please correct the unit errors and either provide an independent measurement of the pressure gradient or explicitly frame the 7.8 kHz value as a consistency check rather than as evidence for the mechanism.","section":"Sec. 6"},{"comment":"The analysis relies on an assumed current density profile J ~ (1-(r/a)^2)^α with α=3 (Sec. 5), but no justification or sensitivity study is provided. The location of the rational surface r_s ≈ 0.12 m for shot #37103, the computed island width, and the frequency decomposition in Sec. 5 all depend on this assumption. Since the interpretation of the mode as a 2/1 drift-tearing mode and the subsequent mechanism discussion hinge on r_s, please provide a sensitivity analysis over reasonable α values or cite a direct constraint on the current density profile from the ADITYA-U diagnostics.","section":"Sec. 5-6"}],"minor_comments":[{"comment":"The text states 'Figure 3 presents the statistical analysis of these parameters across 150 disruptive shots', but the actual figures showing current quench time and rate are Fig. 4(a) and 4(b). Please correct the cross-reference.","section":"Sec. 3.3"},{"comment":"The text refers to 'Figure 5' and 'Figure 6' for the bolometer radiation profiles and chord-averaged temperature, but these data are displayed in Fig. 10(a) and 10(b). Please correct the figure citations.","section":"Sec. 6"},{"comment":"The summary contains a duplicated phrase: 'Through a comprehensive Through a comprehensive statistical study'. Please fix the typo.","section":"Sec. 7"},{"comment":"For shot #37103, the text states that the Mirnov signal shows 'a characteristic 50% increase in frequency', but the described frequency range is 7-15 kHz, which is more than a 100% increase. Please clarify the metric used for the percentage change.","section":"Sec. 3.1"},{"comment":"The paper refers to 'line-averaged temperature' estimated from SXR signals; SXR diagnostics measure line-integrated emissivity, and the temperature inference is indirect (via foil-filter ratios, as noted in Sec. 2). Please use the correct terminology and explain the inversion method.","section":"Sec. 6"}],"recommendation":"major_revision","confidential_remarks":"The observation of a potentially new disruption class in ADITYA-U is interesting and within the journal's scope, but the manuscript in its current form does not provide a reproducible classification protocol or valid quantitative support for the mechanism. I would encourage the editor to request a major revision in which the authors specify the exact labeling rule, release the shot list, add error estimates for the thresholds, and correct the unit errors in Sec. 6. The paper may then be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper probably has a real observation—a disruption precursor with rising mode frequency, distinct from the usual locked-mode decay—but it presents it as a new regime with statistical thresholds that rest on a classification rule the paper never actually states. The thresholds are post-hoc, lack error bars, and the mechanism section has unit errors that undermine the quantitative claim. That said, the empirical contrast between the two representative discharges is clear, the bolometer and SXR temperature-hollowing evidence points in the right direction, and the authors are appropriately careful to compare with JET termination tearing modes. The core observation deserves a serious look, but the paper needs a major revision.\n\nWhat is new: the statistical characterization of AMD in ADITYA-U—q_edge > 4.3, current decay > 16%, CQ time < 1.5 ms, CQ rate > 56 MA/s—is not in the prior literature. The frequency-increase precursor is a useful contrast to the usual LMD picture. The use of SVD, Mirnov, and bolometer diagnostics to show a 2/1 mode with a hollow temperature profile is reasonable and consistent with the figures.\n\nThe soft spots are real. The biggest one: the 150-shot analysis in Sec 3.1/3.2 uses an AMD/LMD label defined only by qualitative description ('steady rise' vs 'decay'). There is no shot list, no quantitative slope or window definition, no inter-rater check. The thresholds are then computed on that labeled sample, so the separation could partly be an artifact of selection. This is load-bearing and needs to be fixed. Second, the pressure-gradient derivation in Sec 6 has apparent unit errors: 2×10^4 Pa/m and then 1.625×10^23 Pa/m cannot both be the same gradient, and the '7.8 kHz diamagnetic term' is essentially the residual after subtracting measured flow contributions, so the claim that the pressure gradient dominates is not independently tested. Third, the figures are mislabeled (Fig 3 vs Fig 4) and percentages are stated without any uncertainty or cross-validation.\n\nIf the authors can supply a reproducible classification rule, a shot list, and re-derive the pressure-gradient claim, this becomes a useful contribution to the disruption database effort. It is not fatally flawed; it is under-supported. A good referee should ask for those revisions before acceptance.","headline":"Plausible new disruption class, but the reported thresholds are built on a classification rule the paper never states; needs major revision before the quantitative claims can be trusted.","tokens_in":14552,"tokens_out":2033,"would_cite":false,"duration_ms":21586,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["52.55.Fa"],"model":"deepseek-v4-flash","headline":"The paper identifies a new disruption class in the ADITYA-U tokamak, Accelerated Mode Disruption, distinguished by a rising (2,1) drift-tearing-mode frequency followed by a sudden frequency collapse and a faster, more intense current…","keywords":["Accelerated Mode Disruption","Locked Mode Disruption","drift-tearing mode","tokamak disruption","current quench","edge safety factor","temperature hollowing","ADITYA-U tokamak"],"falsifier":"Take the same 150 discharges, mask the labels, and have two independent analysts classify each shot by a pre-specified rule (for example, the sign of the linear slope of the dominant Mirnov frequency during the 10 ms before the thermal quench, or a threshold on the frequency rise). If the sharp boundaries at $q_{\\rm edge}=4.3$, 16% current decay, 1.5 ms quench time, and 56 MA/s do not reappear, the AMD/LMD separation is an artifact of the labelling procedure. A complementary check is to find a single discharge with a clear rising-frequency precursor but a current quench longer than 1.5 ms, which would break the claimed correspondence.","tokens_in":13388,"feed_emoji":"⚡","tokens_out":10284,"duration_ms":96975,"temperature":0.7,"pith_summary":"This paper claims that a sizeable fraction of disruptions in ADITYA-U form a distinct class, called Accelerated Mode Disruption (AMD), separate from the familiar Locked Mode Disruption (LMD). In AMD the precursor instability accelerates: the (2,1) drift-tearing mode frequency rises steadily while the amplitude nonlinearly saturates, and only then does the frequency collapse as the island expands and the discharge disrupts. A statistical study of 150 discharges separates the two classes with empirical thresholds: edge safety factor above 4.3, plasma current decay above 16%, current quench time below 1.5 ms, and quench rate above 56 MA/s mark AMD. The proposed mechanism is core radiation and temperature hollowing, which steepens the pressure and current-density gradients near the q=2 surface and destabilizes the drift-tearing mode. If this class is real, disruption prediction and mitigation systems would need a separate trigger for frequency-rising precursors, because AMD quenches faster and is more damaging than LMD.","feed_headline":"New disruption regime in ADITYA-U quenches current faster","feed_subtitle":"Rising (2,1) mode frequency and sub-1.5 ms current quench mark a hazard distinct from locked-mode disruptions.","key_machinery":"The carrying object is the (m/n = 2/1) drift-tearing mode (DTM), the coupled tearing-drift instability that is the sole dominant mode in AMD discharges. Its frequency is decomposed as $f_{\\rm MHD} = \\frac{m}{2\\pi r B_\\varphi}\\frac{\\nabla p_e}{e n_e} + \\frac{n v_\\varphi}{2\\pi R_0} + \\frac{m v_\\theta}{2\\pi r}$ at the rational surface; for shot #37103 the diamagnetic term contributes about 7.8 kHz of a 10 kHz frequency, with toroidal flow about 1.7 kHz and poloidal flow about 0.5 kHz. The mechanism this equation encodes is the paper's central explanation: a rising pressure gradient near the rational surface is what accelerates the mode, and the same core-temperature-hollowing that steepens that gradient also destabilizes the mode and triggers the quench. The empirical separators, $q_{\\rm edge} > 4.3$ and greater than 16% current decay, locate the q=2 surface and the current-profile evolution that set AMD apart from LMD.","core_discovery":"The central claim is that ADITYA-U hosts a previously unrecognized disruption regime, Accelerated Mode Disruption, whose precursor differs from locked-mode behavior: instead of the mode slowing toward locking, the (2,1) drift-tearing mode shows a sustained frequency rise while its magnetic island width stays roughly 4–5 cm, followed by an abrupt frequency drop and island expansion immediately before the thermal quench. The paper derives statistical boundaries between AMD and LMD from 150 discharges: 73% of AMDs occur for $q_{\\rm edge} > 4.3$ while 80% of LMDs occur below that value, and a plasma current decay greater than 16% from maximum gives a 75% chance of AMD. AMD current quenches are shorter than 1.5 ms and faster than 56 MA/s, making them more hazardous than LMD. Physically, the paper attributes AMD to core radiation and hollowing of the temperature profile, which flattens or hollows the current profile and steepens both the current-density and pressure gradients near the q=2 rational surface, driving the dominant (2,1) drift-tearing mode; roughly 78% of the observed frequency rise is assigned to the diamagnetic pressure-gradient term.","pith_inferences":["Extension, not the paper's claim: the frequency slope during the precursor could serve as a continuous danger score, with faster rises predicting shorter quench times; the paper's correlation between final frequency and quench time supports this but does not test it directly.","Extension, not the paper's claim: if the $q_{\\rm edge} > 4.3$ boundary reflects the position of the q=2 surface near the plasma edge, then similar AMD-like events should appear in other small and medium tokamaks operated at high edge safety factor with core radiation; this is testable with existing multi-device disruption records.","Extension, not the paper's claim: a practical predictor could classify AMD using only Mirnov frequency slope and soft-X-ray core hollowing, avoiding the need for full equilibrium reconstruction; the paper's diagnostics show both signals change early, but it does not test such a classifier."],"forward_implications":["Disruption prediction systems that rely on mode locking as the warning sign will miss AMD until very late; a rising (2,1) mode frequency with saturated amplitude is itself an early precursor.","Real-time monitors using $q_{\\rm edge}$ and the current decay coefficient can flag high-risk shots: values above 4.3 and above 16% decay point to the faster AMD class.","Because AMD current quenches are shorter than 1.5 ms and faster than 56 MA/s, mitigation actuators for ADITYA-U must be able to react on or before the frequency-collapse phase, not after locking.","The DTM frequency at the disruption onset correlates with quench duration, so the precursor frequency rise is not just a label but a severity measure.","A planned mitigation framework on ADITYA-U would combine software predictions with hardware actuators keyed to $q_{\\rm edge}$, current decay, and DTM frequency."],"supporting_citations":[{"why":"defines the locked-mode disruption against which AMD is contrasted","marker":"[25]"},{"why":"supplies the JET observation of tearing modes during current termination and the temperature-hollowing mechanism that AMD extends to ADITYA-U","marker":"[9]"},{"why":"establishes the drift-tearing mode as the instability class invoked to explain the (2,1) mode in AMD","marker":"[18]"},{"why":"gives the earlier ADITYA-U characterization of drift-tearing modes, including rotation direction and gas-puff effects, that this paper builds on","marker":"[19]"},{"why":"provides the drift-tearing island theory used to interpret mode saturation and growth before disruption","marker":"[20]"},{"why":"supplies the current-quench time and rate methodology the paper uses to separate AMD from LMD","marker":"[8]"},{"why":"supports the interpretation of nonlinear saturation of tearing mode amplitude before the final collapse","marker":"[26]"},{"why":"provides the MHD frequency relation from which the paper decomposes diamagnetic and flow contributions","marker":"[35]"},{"why":"gives the magnetic-island width estimate used to track the island expansion before the AMD quench","marker":"[34]"}],"fun_headline_variants":["New ADITYA-U disruption mode ends in faster current quench","Frequency rise then collapse: novel disruption regime on ADITYA-U","Drift-tearing mode dominates new two-phase disruption in ADITYA-U","ADITYA-U's new disruption: mode speeds up, then crashes fast"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire statistical separation rests on the assumption that the 150 discharges were sorted into AMD and LMD by a consistent and unbiased reading of the precursor frequency trend; the paper gives no quantitative, pre-specified sorting rule, so the empirical thresholds could partly be a product of the sorting.","fun_headline_variants_meta":{"raw":{"variants":["New ADITYA-U disruption mode ends in faster current quench","Frequency rise then collapse: novel disruption regime on ADITYA-U","Drift-tearing mode dominates new two-phase disruption in ADITYA-U","ADITYA-U's new disruption: mode speeds up, then crashes fast"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000223,"raw_usage":{"total_tokens":1515,"prompt_tokens":1059,"completion_tokens":456,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":675,"completion_tokens_details":{"reasoning_tokens":389}},"tokens_in":675,"tokens_out":456,"duration_ms":5327,"temperature":1.0,"reasoning_tokens":389,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:51:54.569172+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the same 150 discharges, mask the labels, and have two independent analysts classify each shot by a pre-specified rule (for example, the sign of the linear slope of the dominant Mirnov frequency during the 10 ms before the thermal quench, or a threshold on the frequency rise). If the sharp boundaries at $q_{\\rm edge}=4.3$, 16% current decay, 1.5 ms quench time, and 56 MA/s do not reappear, the AMD/LMD separation is an artifact of the labelling procedure. A complementary check is to find a single discharge with a clear rising-frequency precursor but a current quench longer than 1.5 ms, which would break the claimed correspondence.","supporting_citations":[{"cited_title":"Mode locking in tokamaks,","cited_arxiv_id":null,"evidence_quote":"defines the locked-mode disruption against which AMD is contrasted"},{"cited_title":"Onset of tearing modes in plasma termination on JET: the role of temperature hollowing and edge cooling,","cited_arxiv_id":null,"evidence_quote":"supplies the JET observation of tearing modes during current termination and the temperature-hollowing mechanism that AMD extends to ADITYA-U"},{"cited_title":"Drift-tearing modes in a tokamak plasma,","cited_arxiv_id":null,"evidence_quote":"establishes the drift-tearing mode as the instability class invoked to explain the (2,1) mode in AMD"},{"cited_title":"Effect of periodic gas-puffs on drift-tearing modes in ADITYA/ADITYA- U tokamak discharges,","cited_arxiv_id":null,"evidence_quote":"gives the earlier ADITYA-U characterization of drift-tearing modes, including rotation direction and gas-puff effects, that this paper builds on"},{"cited_title":"Drift-tearing magnetic islands in tokamak plasmas,","cited_arxiv_id":null,"evidence_quote":"provides the drift-tearing island theory used to interpret mode saturation and growth before disruption"},{"cited_title":"Characterization of the plasma current quench during disruptions in ADITYA tokamak,","cited_arxiv_id":null,"evidence_quote":"supplies the current-quench time and rate methodology the paper uses to separate AMD from LMD"},{"cited_title":"Observation of the bifurcation of tearing modes due to supersonic gas injected into the J-TEXT plasmas,","cited_arxiv_id":null,"evidence_quote":"provides the MHD frequency relation from which the paper decomposes diamagnetic and flow contributions"},{"cited_title":"Pre- disruption MHD activity in the LT-4 tokamak,","cited_arxiv_id":null,"evidence_quote":"gives the magnetic-island width estimate used to track the island expansion before the AMD quench"}],"review_version":1}