{"id":"3d0fb39d-d0b4-4988-9ba0-822a298f982c","arxiv_id":"2608.07288","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The first 2D NLTE photospheric-phase simulation of a 0.02 Msun He-shell double detonation model predicts substantial NLTE and viewing-angle effects, with most lines of sight too heavily blanketed to match normal Type Ia supernovae.","lead":"Using a 2D Monte Carlo radiative transfer code with full non-local thermodynamic equilibrium physics, this paper simulates how a double detonation Type Ia supernova model looks during its photospheric phase. It finds that full NLTE treatment makes the synthetic light curves and spectra bluer, slows optical declines, and preserves strong viewing-angle dependence after peak that an approximate treatment washes out.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The post-peak viewing-angle retention claim—the paper's headline result—is demonstrated only visually, lacks Monte Carlo error bars, and is tied to an over-ionised NLTE state the paper itself shows is unphysical; its robustness is therefore unestablished.","rationale":"The paper's central contribution is the claim that full NLTE, combined with multi-dimensionality, changes observables on comparison-relevant scales and specifically retains strong spectroscopic viewing-angle dependence after peak that the approximate NLTE treatment washes out. This is a well-controlled code-to-code, same-model comparison: both simulations use the same 2D slice, the same cmfgen-based atomic data, the same 5×10^7 packets and virtual-packet scheme, and the same time stepping. Credit is due for the controlled comparison and for the physical interpretation supported by the Fe ion population maps (§3.1.1) and Ca II population maps (§3.1.2). The reader's weakest assumption is the 2D-to-3D reduction under NLTE. I find this comparatively well supported: the Gronow et al. (2021) model is axisymmetric by construction; §2.2 validates the choice with alternate slices, an azimuthally averaged model, and a 3D artis-classic run; and a 2D cylindrical grid with 3D packet transport is the standard representation of an axisymmetric ejecta for multi-D radiative transfer. The absence of a 3D NLTE run is a limitation but not, in my judgment, the load-bearing weakness. The load-bearing weakness is the robustness of the headline post-peak retention claim. Three considerations converge: (1) the claim is demonstrated visually from Figure 7 at three epochs without a quantitative variation metric or error bars; (2) the proposed mechanism—high NLTE ionisation suppressing singly-ionised species except in the He-shell-burning directions—is the same high-ionisation state the paper shows is over-ionised relative to normal SNe Ia (no secondary I-band maximum at any viewing angle, §3.3.1), so the retention is a corollary of an acknowledged deficiency, and whether it survives a more physical late-time ionisation balance is untested; (3) Appendix D concedes that the high-velocity Ca II photoionisation rates, which set the high-velocity Ca II population responsible for the retained NIR-triplet angle dependence, may be MC-noise-limited, and no seed-to-seed scatter is provided. None of this invalidates the paper; the qualitative trends are coherent and the interpretation is honest about over-ionisation. But the most distinctive claim is exactly the one that currently rests on visual inspection and on an acknowledged unphysical plasma state, which is why I would require quantitative variation metrics with Monte Carlo error bars before treating the differential retention as established. This supports the reader's CONDITIONAL verdict and does not move it; I disagree with the reader's specific choice of weakest assumption, hence 'partial' agreement.","tokens_in":29635,"tokens_out":22614,"duration_ms":201100,"concrete_test":"At 19.5 and 29.5 d, compute for both 2D simulations the RMS deviation of normalised flux across the five viewing angles in the Ca II NIR triplet (8200–8700 Å) and Ti II trough (3900–4500 Å) windows, and repeat each simulation with at least five independent Monte Carlo seeds (same packet count and virtual-packet scheme) to obtain 1σ seed scatter. If the NLTE-vs-classic difference in RMS viewing-angle variation is not larger than the seed scatter, the headline differential-retention claim is not established; if it is larger, the claim passes the statistical test, and the remaining contingency—whether the retention survives a more physical (clumping-corrected, secondary-NIR-maximum-producing) late-time ionisation—would be tested by a modified recombination run.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing new claim is that after peak the 2D NLTE simulation retains strong spectroscopic viewing-angle variation while artis-classic predicts significantly reduced variation (abstract; §3.1.3). It is demonstrated only qualitatively: Figure 7 is compared by eye at 9.5, 19.5, 29.5 d, with no quantitative variation metric and no Monte Carlo error bars. The proposed mechanism is the NLTE simulation's higher ionisation, which suppresses Fe II/Ti II absorption except in directions where He-shell burning products are intrinsically concentrated, whereas classic recombines to singly-ionised species everywhere, erasing the angle dependence. However, §3.3.1 shows this same high ionisation is over-ionised relative to normal SNe Ia—no viewing angle produces a secondary NIR maximum—so the retention claim is a corollary of an acknowledged model deficiency. If the over-ionisation were corrected (e.g., by clumping, which the paper itself suggests), the post-peak variation could move toward the classic result, weakening the central message that multi-D and NLTE must be treated together. Compounding this, Appendix D concedes that the high-velocity Ca II photoionisation rates, which control the high-velocity Ca II absorption central to the retained NIR-triplet angle dependence, 'could be somewhat impacted by Monte Carlo noise.' With 5×10^7 packets, five viewing angles, and no seed scatter reported, the claimed NLTE-vs-classic difference in post-peak variation is not quantitatively established. The reader's 2D-to-3D concern is legitimate but less load-bearing: the model is axisymmetric by construction and §2.2's slice and azimuthal-average checks plus a 3D artis-classic run adequately support the geometry.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the first multi-dimensional NLTE photospheric-phase radiative transfer simulation for a Type Ia supernova explosion model, applied to the double detonation model M10_02 (1 Msun CO core + 0.02 Msun He shell) of Gronow et al. (2021). The simulation is two-dimensional, built from a phi=90 degree slice of the 3D explosion model, and is compared with (i) a 2D simulation using the approximate NLTE treatment artis-classic, (ii) a 1D NLTE simulation based on the equatorial direction, (iii) observations of SN 2011fe and the Hicken et al. (2009) sample. The central claims are that NLTE and multi-dimensionality both matter on observational scales, that the full NLTE treatment produces a bluer SED and slower optical declines than the approximate treatment, and that after peak the 2D NLTE simulation retains strong spectroscopic viewing-angle variation whereas the approximate-NLTE simulation predicts significantly reduced variation. The paper further concludes that the 0.02 Msun He-shell mass is too large for the model to resemble normal SNe Ia over most lines of sight.","tokens_in":29899,"tokens_out":3299,"duration_ms":33383,"significance":"If the central claims hold, this is an important technical and scientific step: it is the first demonstration that multi-dimensionality and full NLTE must be treated simultaneously when evaluating explosion models against photospheric-phase observations of SNe Ia. The work has significant strengths beyond the headline result: the comparisons are carefully controlled (same ejecta model, same atomic data, same virtual-packet scheme); the 2D slice is cross-checked against other 2D constructions and a 3D artis-classic run; nothing is fitted to SN 2011fe or the Hicken sample, so the predictions are genuinely falsifiable; and the adaptive NLTE solver described in Appendix A is a valuable methodological development. The paper also makes a concrete, testable prediction, namely that the M10_02 configuration fails to produce a secondary NIR maximum for any viewing angle. For these reasons the result warrants publication once the load-bearing caveats identified below are addressed.","major_comments":[{"comment":"The headline claim that 'after peak, our 2D NLTE simulation retains a strong viewing-angle dependence in its spectra whereas our simulation of the same ejecta model using an approximate NLTE treatment predicts significantly reduced spectroscopic viewing-angle variation' is supported only by visual inspection of Figure 7. To make this claim quantitative, please provide a metric of spectroscopic variation across the five viewing angles (for example, the RMS or median absolute deviation of flux in the Ti II trough and Ca II NIR triplet regions, or a full-spectrum RMS-difference curve as a function of epoch) together with Monte Carlo uncertainties estimated from the packet seed. The absence of error bars is particularly consequential because Appendix D concedes that the high-velocity Ca II photoionisation rates, which control the high-velocity Ca II NIR triplet that is central to the retained angle dependence, 'could be somewhat impacted by Monte Carlo noise'; a quantified demonstration is needed to show that the NLTE-vs-classic difference in post-peak angle variation is larger than the Monte Carlo noise.","section":"Section 3.1.3, Figure 7"},{"comment":"The post-peak retention of viewing-angle variation is produced by the same high ionisation state that the paper itself identifies as unphysical: no viewing angle produces a secondary NIR maximum, and the text states that the ejecta 'remain over-ionised relative to observations after peak.' The retained angle variation is therefore argued from a model state that is acknowledged to be deficient. If the over-ionisation were reduced by a physically plausible mechanism such as clumping, which the paper itself suggests in Section 3.3.1, the post-peak angle variation could move toward the classic result, weakening the conclusion that multi-dimensionality and NLTE must be treated together. Please either qualify the central claim explicitly in the abstract and conclusions, or provide a test (for example, a calculation with enhanced recombination/clumping in the relevant regions) showing that the retained angle variation is not an artefact of the over-ionised state.","section":"Section 3.3.1 and Section 4"},{"comment":"The 2D cylindrical ejecta model is validated against the full 3D M10_02 model only with artis-classic simulations, not with full NLTE. Given that the paper's central claim concerns the interaction of multi-dimensional geometry with full NLTE, the possibility remains that angle-dependent radiative coupling in the true 3D structure behaves differently under full NLTE than in the reduced 2D slice. Please either present a 3D NLTE test at a few representative epochs (even with reduced packet number, clearly labelled as a convergence check), or provide a quantitative argument showing why the classic-run validation is sufficient to guarantee that the post-peak NLTE viewing-angle-retention claim is not an artefact of the slice geometry.","section":"Section 2.2 and Section 3"},{"comment":"The paper states that the excited-state Ca II photoionisation rates, which drive the high-velocity Ca II population and hence the most striking spectroscopic differences between simulations, 'could be somewhat impacted by Monte Carlo noise.' Since this is a load-bearing quantity for both the 1D-vs-2D comparison and the NLTE-vs-classic angle-dependence claim, the paper should attach a quantitative estimate of the resulting uncertainty in the Ca II photoionisation rate and in the synthetic spectra, rather than relying on the qualitative concession in the appendix.","section":"Appendix D"}],"minor_comments":[{"comment":"The preprint title contains 'T ype' with a space; this should be corrected to 'Type'.","section":"Title page"},{"comment":"Code names such as 'artis-nlte' and '1Dartis-nlte' are inconsistently spaced; please use consistent formatting such as '1D artis-nlte' and '2D artis-nlte'.","section":"Throughout"},{"comment":"The caption states 'log(normalised flux) + offset' but does not define the offset values; please state the vertical offset and normalisation used.","section":"Figure 7 caption"},{"comment":"The highlighted region for the Ca II NIR triplet is described as only covering the observed feature, with the high-velocity component appearing just blueward of it; consider extending the shading so the reader can directly see the high-velocity component in the figure.","section":"Section 3.1.2, Figure 7"},{"comment":"The derived light-curve quantities (rise time, peak magnitude, Delta m_15) would benefit from a statement of Monte Carlo noise, either as quoted uncertainties or as an assertion that the noise is negligible relative to the quoted differences; this would strengthen the 1D vs 2D comparison.","section":"Table 1"}],"recommendation":"major_revision","confidential_remarks":"The paper is a genuine technical and scientific contribution, and I would like to see it published after revision. The main risk is the discrepancy between the strength of the headline claim (post-peak NLTE retention of viewing-angle variation) and the evidence provided for it: the claim is visual, lacks Monte Carlo error bars, and is tied to an ionisation state that the paper itself demonstrates is over-ionised relative to normal SNe Ia. The reviewer's stress-test note is therefore a fair reading of the manuscript, and I agree that the robustness of the central claim is not yet established. In revision, the authors should either provide quantitative angle-variation metrics with error bars, or soften the claim to reflect that the retention is a corollary of the current model's over-ionisation and does not yet establish a general requirement to treat multi-D and NLTE together. The 2D-slice validation concern is real but secondary; it is mitigated by the 3D classic check, though not eliminated. I would not reject: the code developments and controlled comparisons are valuable, and the over-ionisation limitation is openly acknowledged in the text."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the 2D NLTE calculation itself. For the first time we get a multi-dimensional, full-NLTE look at the photospheric phase of a double detonation model, with the same ejecta model, same atomic data, and same code compared against a 2D approximate-NLTE run and a 1D NLTE run. That is a solid controlled setup, and the result that NLTE changes the SED, the light curve declines, and the secondary NIR maximum on observational scales is credible. The conclusion that 0.02 Msun of helium is too much for most lines of sight is also well supported by the overly strong high-velocity Ca II and Ti II blanketing in the northern and equatorial directions. The paper is transparent about its own over-ionisation, which is a point in its favour.\n\nThe soft spots are real but not fatal. The headline claim, that after peak the 2D NLTE run retains strong viewing-angle dependence while the approximate run does not, is demonstrated only by eye in Figure 7, with five angles, no quantitative variation metric, and no Monte Carlo error bars. Appendix D also concedes that the high-velocity Ca II photoionisation rates, central to that angle dependence, could be affected by Monte Carlo noise. More importantly, the retained variation is tied to the very over-ionisation that the paper itself says is unphysical: no viewing angle makes a secondary NIR maximum. So the post-peak retention claim is, at present, a corollary of a known model deficiency. It may survive a better treatment, but the abstract currently overstates it. The 2D-to-3D worry is less load-bearing: the model is axisymmetric by construction, and the slice and azimuthal-average checks plus a 3D classic run are reasonable support for the geometry.\n\nThis is a careful, serious paper and the central message—that multi-D and NLTE must be treated together when evaluating double detonation models—holds up. The data and inputs are not yet public, and the paper promises only spectra and light curves on HESMA, not the model and parameter files. That should be fixed.\n\nFor a referee: yes, send it out. Ask for a quantitative viewing-angle metric with uncertainty, a sharpened abstract that distinguishes the robust NLTE effects from the post-peak angle-retention claim, and a discussion of whether the retained variation would survive the clumping-style correction the paper itself suggests. With those revisions it is a strong addition to the SN Ia modelling literature.","headline":"The first 2D NLTE photospheric-phase radiative transfer simulation for a double detonation model is a real step forward, but the headline post-peak viewing-angle claim is weaker than the abstract suggests and needs a quantitative treatment and an honest caveat about the model's acknowledged over-ionisation.","tokens_in":30580,"tokens_out":1125,"would_cite":true,"duration_ms":14571,"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":"Full NLTE radiative transfer keeps double-detonation supernova spectra strongly viewing-angle dependent after peak and changes colours and light curves on observable scales.","keywords":["radiative transfer","non-local thermodynamic equilibrium","double detonation","Type Ia supernovae","photospheric phase","viewing-angle dependence","Monte Carlo methods"],"falsifier":"A 3D full-NLTE simulation of the same M10_02 ejecta, comparing post-peak spectroscopic viewing-angle variation directly to the 2D NLTE result, would settle whether the retained variation is a genuine ejecta property or an artifact of the 2D slice; if the variation collapses in 3D, the central claim fails.","tokens_in":29423,"feed_emoji":"💥","tokens_out":16134,"duration_ms":123573,"temperature":0.7,"pith_summary":"The paper carries out the first multi-dimensional, full non-local thermodynamic equilibrium (NLTE) radiative transfer simulation of the photospheric phase of a double detonation Type Ia supernova model, using the M10_02 model with a 1 $M_\\odot$ carbon-oxygen core and 0.02 $M_\\odot$ helium shell. It argues that departures from both spherical symmetry and local thermodynamic equilibrium are large enough to change synthetic observables at levels relevant to comparisons with real supernovae. The full NLTE treatment leaves the ejecta more ionised than an approximate NLTE treatment, giving bluer colours, slower optical declines, and no secondary near-infrared maximum, which moves the model closer to normal Type Ia supernovae but leaves it over-ionised after peak. The central result is that the NLTE simulation retains strong spectroscopic viewing-angle variation after peak, whereas the approximate treatment predicts markedly reduced variation. The paper concludes that multi-dimensionality and NLTE must be treated together when confronting these models with observations, and that the 0.02 $M_\\odot$ helium shell is too massive to match normal SNe Ia over most lines of sight.","feed_headline":"NLTE keeps double-detonation spectra angle-dependent post-peak","feed_subtitle":"A 2D NLTE simulation preserves post-peak viewing-angle scatter that approximate models erase.","key_machinery":"The central object is the 2D cylindrical ejecta model constructed from the $\\phi=90^\\circ$ half-plane slice of the 3D M10_02 double detonation explosion model (a 1 $M_\\odot$ carbon-oxygen core with a 0.02 $M_\\odot$ helium shell), which preserves the model's approximate axial symmetry. The argument is carried by the two plasma treatments in the Monte Carlo radiative transfer code artis: the full NLTE solver (artis-nlte) that solves the equations of statistical equilibrium and includes non-thermal ionisation, and the approximate NLTE treatment (artis-classic) that tracks only ground-state photoionisation and assumes Boltzmann level populations. The quantity that does the work is the ionisation state: the full NLTE treatment produces a higher and longer-lasting ionisation level, which shifts the spectral energy distribution blueward, removes the secondary near-infrared maximum, and keeps the post-peak spectra strongly viewing-angle dependent. An adaptive NLTE solver that strips ionisation stages with negligible populations when the matrix becomes singular is what makes the multi-dimensional early-phase simulation numerically stable.","core_discovery":"We show that a 2D NLTE radiative transfer simulation of the M10_02 double detonation model in the photospheric phase is feasible and changes the predicted observables at levels relevant to observations. Compared to an approximate NLTE treatment, the full NLTE calculation keeps the ejecta more ionised: the free electron fraction is higher, singly ionised species such as Fe II, Co II, Ti II and Ca II recombine later, and the spectrum stays bluer with slower optical light curve declines. The most striking difference is the near-infrared behaviour: the approximate treatment predicts a strong, early secondary I-band maximum, while no viewing angle in the NLTE simulation produces one, indicating that the ejecta remain over-ionised relative to normal SNe Ia after peak. Up to peak, the two treatments predict a similar scale of viewing-angle variation, but after peak the NLTE simulation continues to show strong spectroscopic dependence on viewing angle, most visibly in the Ti II absorption trough and Ca II near-infrared triplet, whereas the approximate treatment predicts markedly reduced variation as its lower ionisation state homogenises the opacities. As a result, no single 1D model reproduces the equatorial line of sight, and the model matches normal SNe Ia only for southern viewing angles; the northern and equatorial angles show excessive line blanketing from helium-shell burning products, which we interpret as evidence that a 0.02 $M_\\odot$ helium shell is too massive for the majority of lines of sight.","pith_inferences":["If the post-peak retention of viewing-angle variation is a generic property of full NLTE treatments of asymmetric explosions, then previous multi-dimensional studies using approximate NLTE may have underestimated the intrinsic spectroscopic scatter expected from double detonation models, which could affect how observed SNe Ia diversity is interpreted.","The adaptive NLTE solver that enables this 2D simulation makes it plausible to extend full NLTE to 3D for other asymmetric explosion channels, such as violent mergers or gravitationally confined detonations; the computational cost reported in the paper suggests such simulations will be expensive but within reach of current high-performance facilities.","The paper's finding that excited-state photoionisation of Ca II drives the NIR triplet differences, and that the bluest edges suffer Monte Carlo noise, suggests that packet-frequency biasing in short-wavelength regions could improve future simulations and might change the quantitative strength of the high-velocity Ca II feature.","The over-ionisation after peak, if confirmed by future 3D NLTE simulations, would strengthen the case that small-scale clumping is a missing ingredient in all current photospheric-phase SNe Ia models, not just double detonations."],"forward_implications":["To judge double detonation models against normal Type Ia supernovae, radiative transfer must combine multi-dimensional geometry with a full NLTE treatment; approximate NLTE alone can erase genuine viewing-angle diversity after peak.","The M10_02 model cannot reproduce normal SNe Ia over the majority of lines of sight because its 0.02 $M_\\odot$ helium shell produces excessive line blanketing; lower helium-shell masses are required for most viewing angles, although some southern angles already match well.","The complete absence of a secondary near-infrared maximum across all viewing angles indicates that the NLTE ejecta remain over-ionised after peak, so some additional recombination mechanism, such as density clumping, is needed in the models.","1D NLTE models constructed from a given direction do not reproduce the corresponding 2D line of sight: the 1D model gives brighter blue light curves and a different Ca II NIR triplet evolution, so 1D treatments can mislead when used for spectral interpretation.","The width-luminosity position of the model shifts with the NLTE treatment: full NLTE gives slower declines and bluer peak colours, placing the model in better agreement with the observed normal SNe Ia distribution than the approximate treatment."],"supporting_citations":[{"why":"Supplies the M10_02 double detonation ejecta model, with a 1 solar mass CO core and 0.02 solar mass helium shell, that is simulated here.","marker":"Gronow et al. (2021)"},{"why":"Provides the artis-nlte implementation of the full NLTE solver and non-thermal ionisation used for the 2D NLTE simulation.","marker":"Shingles et al. (2020)"},{"why":"Defines the artis-classic approximate NLTE treatment that the full NLTE results are compared against.","marker":"Kromer & Sim (2009)"},{"why":"Prior artis-classic simulations of the Gronow et al. sequence that establish the viewing-angle behaviour, including the 2D representation of the 3D model.","marker":"Collins et al. (2022)"},{"why":"Attributes the secondary near-infrared maximum to recombination of doubly to singly ionised iron-group elements, used to interpret the NLTE simulation's absence of the feature.","marker":"Kasen (2006)"},{"why":"Provides the SN 2011fe light curve and spectral data used as the normal SNe Ia reference for comparison.","marker":"Nugent et al. (2011)"},{"why":"Supplies the B- and V-band width-luminosity sample of normal SNe Ia against which the model's decline rates are compared.","marker":"Hicken et al. (2009)"}],"fun_headline_variants":["NLTE keeps double-detonation spectra angle-dependent post-peak","Full NLTE preserves spectral angle spread after peak in double detonation","Over-ionized ejecta in NLTE model suppress secondary NIR peak","Double detonation: NLTE keeps spectra diverse post-peak, unlike approximate","Helium shell too heavy for most lines of sight in NLTE double detonation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 2D cylindrical ejecta model, built from a single half-plane slice of the 3D M10_02 explosion, is assumed to capture the full 3D structure for NLTE radiative transfer, since no 3D NLTE simulation is performed to confirm this.","fun_headline_variants_meta":{"raw":{"variants":["NLTE keeps double-detonation spectra angle-dependent post-peak","Full NLTE preserves spectral angle spread after peak in double detonation","Over-ionized ejecta in NLTE model suppress secondary NIR peak","Double detonation: NLTE keeps spectra diverse post-peak, unlike approximate","Helium shell too heavy for most lines of sight in NLTE double detonation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000733,"raw_usage":{"total_tokens":3381,"prompt_tokens":1147,"completion_tokens":2234,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":763,"completion_tokens_details":{"reasoning_tokens":2137}},"tokens_in":763,"tokens_out":2234,"duration_ms":18126,"temperature":1.0,"reasoning_tokens":2137,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T10:59:44.504462+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A 3D full-NLTE simulation of the same M10_02 ejecta, comparing post-peak spectroscopic viewing-angle variation directly to the 2D NLTE result, would settle whether the retained variation is a genuine ejecta property or an artifact of the 2D slice; if the variation collapses in 3D, the central claim fails.","supporting_citations":[{"cited_title":"A., 2009, @doi [ ] 10.1111/j.1365-2966.2009.15256.x , http://adsabs.harvard.edu/abs/2009MNRAS.tmp.1110K 398, 1809","cited_arxiv_id":null,"evidence_quote":"Defines the artis-classic approximate NLTE treatment that the full NLTE results are compared against."}],"review_version":1}