{"id":"045f5166-ceb7-4d06-bd0b-6bcd8ee8b786","arxiv_id":"1908.09599","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"A new weighted fitting algorithm for stellar limb-darkening coefficients reduces transit light-curve model residuals to below 10 parts per million, and an exact light-curve generator avoids limb-darkening laws entirely.","lead":"This paper introduces an open-source Python package for modeling exoplanet transits, including a new limb-darkening coefficient calculator and an exact transit light curve generator. It claims the new calculator produces transit light-curve models accurate to a few parts per million, important for exoplanet atmospheric studies with JWST and ARIEL.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The <10 ppm claim is measured against a TRIP reference whose numerical convergence and parameter-space coverage are not demonstrated; the reported residuals may bound the fitting method only for this single test case.","rationale":"I read the paper in good faith. The algorithm is clearly described, the code is open-source, and the comparison of fitting methods is informative. The paper's core advance is a weighted-r QS least-squares fit of claret-4 coefficients, judged by light-curve residuals rather than intensity residuals. The simulations support the ranking of methods for HD209458-like parameters. The reader's CONDITIONAL verdict is appropriate. My stress-test focuses on the two places where the central quantitative claim is least secure. First, the reference standard: TRIP is called exact but is a numerical integrator; without convergence or independent validation, the sub-ppm rms values cannot be interpreted as the fitting error alone. This is checkable and not an external-model objection. Second, generality: the headline promises 'all wavelengths' and an order-of-magnitude improvement, but only one stellar atmosphere model and one transit geometry are shown; low-gravity spherical models in the advertised grids are explicitly mentioned as motivating the r-based cutoff, yet are not tested. Neither issue invalidates the method for the demonstrated case, and the paper's limitation sections are honest, so I do not move the verdict. I recommend adding the convergence and grid tests before the broad claim is used as a reference for JWST/ARIEL planning.","tokens_in":17620,"tokens_out":8378,"duration_ms":95116,"concrete_test":"Recompute the HD209458 TESS and 0.25-10 micron residual analysis with TRIP using annulus counts of 100, 1000, 10^4, and 10^5 and with both linear and cubic spline interpolation in r, keeping the SAIL coefficients fixed. If the reference flux or the resulting peak-to-peak residuals change by more than 1 ppm as N increases, or if the N=10^5 residuals differ from the published 2 ppm/<8 ppm values, then the claimed precision is not converged with respect to the reference integrator and should be restated as an upper limit pending an independent quadrature check.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central precision claim is established entirely by comparing limb-darkened light-curve models with reference light curves generated by TRIP from the same model intensity profiles. TRIP is not an analytic exact integrator: Section 2.2.3 states that Equations 16 and 17 are evaluated with the midpoint rule on a uniform partition in r, with intensities interpolated from the input profile, but the paper gives no default annulus count, no convergence study, and no independent accuracy check. If TRIP's discretization error is comparable to 1-10 ppm, then the reported 2 ppm TESS residual and <8 ppm spectral residuals measure the sum of the limb-darkening approximation error and the reference error, not the limb-darkening precision alone. Moreover, the validation uses one HD209458-like stellar model and one transit geometry; the r <= 0.99623 cutoff and weighted-r QS weights are not tested over the Teff/logg/[M/H] grid in Table 1 or across p and b, despite Equation 20 showing p^2 dependence. The paper explicitly acknowledges that real-star discrepancies are outside scope (Sections 1 and 3.3), so the issue is not external accuracy; it is that the internal numerical reference and the generality of the headline precision are unverified.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces ExoTETHyS, an open-source Python package with two subpackages. SAIL computes stellar limb-darkening coefficients by fitting parametric limb-darkening laws to passband-integrated model-atmosphere intensity profiles, using a weighted-r least-squares fit with a quasi-spherical cutoff (weighted-r QS). TRIP generates transit light curves by direct numerical integration of the full stellar intensity profile rather than through limb-darkening coefficients. The authors validate SAIL by generating noiseless reference light curves with TRIP from PHOENIX and ATLAS intensity profiles for an HD209458b-like system, then fitting those light curves with models that use coefficient sets obtained by different fitting methods and limb-darkening laws. They report that claret-4 coefficients from the weighted-r QS method reproduce the reference TESS light curve to 2 ppm peak-to-peak and the spectral light curves (20 nm bins over 0.25-10 micron) to below 8 ppm peak-to-peak, and they propose Eq. (20) to predict light-curve residuals from the intensity-profile fit residual. The central claim is that this algorithm outperforms most previously proposed fitting algorithms by about an order of magnitude for spherical stellar models.","tokens_in":17812,"tokens_out":4663,"duration_ms":48251,"significance":"If the headline precision claim is sustained, ExoTETHyS is a genuinely useful tool for high-precision transit spectroscopy, particularly for JWST and ARIEL simulations and data analysis. The paper's strengths include the release of reproducible, documented code; a direct-integration transit generator (TRIP) that sidesteps limb-darkening parameterizations entirely; and a validation strategy that correctly isolates the limb-darkening parameterization error: reference light curves and fitted coefficients both derive from the same model intensity profile, so the residuals measure how well a given parametric law reproduces the exact integral. This internal-consistency test is not circular for the paper's stated purpose, and the paper explicitly disclaims empirical validation against real stellar surfaces. The comparison of fitting methods is also useful and will help practitioners choose among algorithms. However, the generality of the <10 ppm claim rests on a single test geometry and a single stellar model, and the numerical accuracy of the TRIP reference is not demonstrated, so the strongest headline statements are not yet fully supported.","major_comments":[{"comment":"The central precision claim is measured against reference light curves generated by TRIP, but the numerical accuracy of TRIP itself is not established. Equations (16) and (17) are evaluated with the midpoint rule on a uniform partition in r, yet the paper gives no default annulus count, no convergence study, and no independent check against an analytic or high-accuracy reference. If TRIP's own discretization error is comparable to 1-10 ppm, then the reported 2 ppm TESS residual and <8 ppm spectral residuals are not a clean measurement of the limb-darkening approximation error. The authors should state the default number of annuli, demonstrate convergence with respect to that parameter, and, ideally, validate TRIP against an independent integrator or an analytic solution.","section":"Section 2.2.3 and Section 3.1"},{"comment":"The headline '<10 ppm at all wavelengths' is established from one HD209458-like PHOENIX 2012 stellar model and one transit geometry. Equation (20) explicitly introduces a p^2 dependence, and the residual magnitude should also depend on impact parameter and the stellar intensity profile, but no tests are shown over the Teff, log g, [M/H], p, or b ranges covered by the package grids in Table 1. The cutoff r <= 0.99623 is also a free choice, and no sensitivity to that cutoff is reported. The validation should be extended, or the claims should be explicitly restricted to the tested regime, before stating that the algorithm ensures <10 ppm precision for relevant transit light curves generally.","section":"Section 3.1, Section 3.3, and Table 1"},{"comment":"The statement that the algorithm outperforms 'most of the algorithms proposed in the previous literature' by one order of magnitude is not uniformly supported by the paper's own Figures 2 and 3. For the TESS simulation, weighted-mu QS and unweighted QS give peak-to-peak residuals of 3 ppm, only marginally worse than the 2 ppm of weighted-r QS, and the spectral residuals of the other QS methods remain below 13 ppm rather than a factor of ten larger. The order-of-magnitude improvement is persuasive relative to unweighted spherical fits, but the sweeping wording should be tightened to specify the comparison class and to acknowledge that the advantage over other quasi-spherical fits is modest in the tested case.","section":"Abstract and Section 5"}],"minor_comments":[{"comment":"In Eq. (6), the summation index is k but the coefficient is written as a_n; this should be a_k (or the index should be n).","section":"Section 2.1.3, Eq. (6)"},{"comment":"The text contains a typo: 'whic includes' should be 'which includes'.","section":"Section 2.1.1"},{"comment":"The sentence describing interp-r residuals says 'about 1.5 and 2 times larger residual amplitudes, respectively'; the pairing of the two interpolation grids with the two factors is ambiguous and should be clarified.","section":"Section 3.1, bottom of page 8"},{"comment":"Section 3.3 refers to 'HD20458 b' while the rest of the paper uses HD209458 b; the Figure 5 caption contains a similar typo ('HD20458 b').","section":"Section 3.3 and Figure 5 caption"},{"comment":"The value r_cut = 0.99623 is stated without explanation of where it comes from; a sentence explaining the origin of the numerical value and its precision would help users judge the robustness of the quasi-spherical cutoff.","section":"Section 2.1.4"},{"comment":"Equation (20) introduces a free factor k with k approximately 1, but the paper does not give the fitted values of k obtained in the 'different transit parameters' tests or specify the range of geometries over which the proportionality was checked; providing this information would make the formula more actionable.","section":"Section 3.3, Eq. (20)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a software methods paper, and the AJ format is appropriate. The main concern is not the internal-consistency validation concept, which is sound for measuring parameterization error, but the gap between the strong headline claims ('<10 ppm at all wavelengths', 'one order of magnitude') and the evidence from one geometry and one unverified numerical reference. These issues are fixable with additional convergence tests, parameter-space scans or explicit qualifications, so I view this as major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new piece is the weighted-r quasi-spherical (QS) fitting algorithm for limb-darkening coefficients and the systematic comparison of fitting methods on transit light-curve residuals. TRIP itself is a re-packaged version of the authors' earlier tlc algorithm, so that part is not new, but making it available in an open-source package is useful. The validation is self-consistent: reference light curves are generated with TRIP from the same model intensity profiles used to fit the coefficients, so the residuals isolate the limb-darkening law error. That is the right test for what they claim, and the result is credible: claret-4 with weighted-r QS gives about 2 ppm peak-to-peak in the TESS band and below 8 ppm across 0.25-10 microns, a clear improvement over unweighted, weighted-mu, and interpolation-based fits. The paper also does a service by comparing many algorithms on equal terms.\n\nThe soft spots are omissions rather than errors. First, TRIP's numerical integration uses the midpoint rule with no default annulus count and no convergence study. Since the absolute ppm claims are measured against TRIP, the reported residuals are upper bounds on the limb-darkening error, but the paper should demonstrate that TRIP itself is converged at the sub-ppm level. That is easy to fix and probably fine, but it should be stated. Second, the validation uses one star (HD209458-like) and one transit geometry. The fixed r-cut at 0.99623 and the p^2 scaling in Equation 20 are not tested over the parameter grid in Table 1. The abstract generalizes a bit more than the evidence: the 'order of magnitude' improvement holds against the older spherical fitting methods, but the best prior QS method gives 3 ppm versus their 2 ppm, so that part of the headline is overstated. The authors do acknowledge real-star discrepancies are out of scope, so I do not hold that against them.\n\nThe math is straightforward, the comparison is transparent, and the code is public. The citation pattern is appropriate, including the self-citation to Morello et al. 2017 for the algorithm that became TRIP. This deserves serious peer review and would be accepted; the main referee requests should be a TRIP convergence test, a broader grid of stellar parameters and transit geometries, and a toned-down abstract. I would cite it and bring it to a reading group for people working on transit spectroscopy.","headline":"Solid software paper with a credible few-ppm limb-darkening fitting method; the headline precision claim is broadly right but needs a caveat about the numerical reference and a single test case.","tokens_in":740,"tokens_out":1705,"would_cite":true,"duration_ms":50408,"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":"ExoTETHyS fits stellar limb darkening so precisely that its model transit light curves reproduce exact reference calculations to better than 10 parts per million at every wavelength.","keywords":["exoplanet transits","limb darkening","stellar atmosphere models","transit light curves","spectrophotometry","open-source software","JWST","ARIEL"],"falsifier":"Generate a noiseless reference transit light curve with TRIP using an intensity profile from an independent stellar atmosphere code not in ExoTETHyS's grids (e.g., a 3D or MESA model) for a star like HD 209458; if claret-4 coefficients from the weighted-r QS fit reproduce that curve with residuals above about 10 ppm, the claimed precision is specific to the grid rather than general.","tokens_in":17418,"feed_emoji":"🔭","tokens_out":9030,"duration_ms":76156,"temperature":0.7,"pith_summary":"The paper introduces ExoTETHyS, an open-source Python package for modeling exoplanet transits, and argues that its limb-darkening coefficient calculator, SAIL, removes a major source of systematic error in transit light-curve models. The key claim is that fitting the four-coefficient claret-4 limb-darkening law with a weighted fit in radial coordinate and a quasi-spherical cutoff yields model light curves accurate to better than 10 parts per million (ppm) at all wavelengths, an order of magnitude better than most previous algorithms for spherical stellar models. In a noiseless simulation of the HD 209458b transit in the TESS passband, peak-to-peak residuals are 2 ppm with rms below 1 ppm, and across 0.25–10 µm the residuals stay below 8 ppm. The paper also provides TRIP, an exact light-curve generator that integrates the full stellar intensity profile without any limb-darkening law, and a formula for predicting light-curve model precision from the goodness of the intensity fit. This matters because upcoming missions such as JWST and ARIEL will reach roughly 10 ppm photometric precision, where these systematic errors would otherwise dominate.","feed_headline":"Transit light-curve models drop below 10 ppm error","feed_subtitle":"New limb-darkening fitting reproduces model star transits tenfold more precisely across all wavelengths.","key_machinery":"The central object is the weighted-r quasi-spherical (QS) fit: a weighted least-squares fit of an analytic limb-darkening law to a passband-integrated stellar intensity profile, with weights proportional to the radial sampling interval in $r = \\sqrt{1-\\mu^2}$ and a cutoff discarding intensities at $r > 0.99623$. The cutoff removes the steep limb drop-off characteristic of spherical model atmospheres, which has negligible effect on the numerically integrated light curve but distorts the coefficient fit; the weights ensure the fit prioritizes the radial intervals that actually contribute to the flux integral. The procedure also rescales spherical intensity profiles to the photometric radius defined by the inflection point of the intensity gradient, so the cutoff applies uniformly across models. This fitting method, implemented in the SAIL subpackage, is what carries the sub-10 ppm precision claim; the companion TRIP subpackage generates reference light curves by direct numerical integration of the occulted flux without any limb-darkening approximation.","core_discovery":"The central discovery is that the choice of fitting algorithm for limb-darkening coefficients, not just the choice of limb-darkening law or stellar atmosphere model, controls the accuracy of transit light-curve models at the tens-of-ppm level, and that a weighted least-squares fit with weights proportional to the sampling interval in the radial coordinate $r$, combined with a cut at $r \\leq 0.99623$ to discard the steep intensity drop-off of spherical models, is the optimal procedure. With the claret-4 law, this weighted-r QS method reproduces noiseless reference light curves generated by direct numerical integration (TRIP) with peak-to-peak residuals of 2 ppm and rms below 1 ppm in the TESS passband, and below 8 ppm across 0.25–10 µm. The same coefficients retrieve the correct transit depth within 5 ppm, impact parameter within $6 \\times 10^{-4}$, and transit duration within 1 s at all wavelengths. The paper concludes that claret-4 coefficients obtained through this algorithm ensure a precision of less than or about 10 ppm in relevant transit light curves at all wavelengths, exceeding by one order of magnitude the precision of most previously proposed algorithms for spherical stellar models.","pith_inferences":["The paper validates precision against internally generated light curves; a natural extension is to test the same weighted-r QS coefficients against real ultra-precise space photometry (e.g., TESS or JWST) for stars with well-determined parameters, where residuals below about 10 ppm would corroborate the accuracy claim.","Because the QS cutoff at $r \\leq 0.99623$ is defined on the rescaled photometric radius, the method may be more robust than the original $\\mu \\geq 0.1$ cutoff for giant and supergiant stars whose intensity drop-off occurs at larger $\\mu$; one could quantify the coefficient bias for low-gravity PHOENIX models to test this.","Equation (20) suggests a cheap, instrument-independent quality metric: a user comparing any two limb-darkening laws or algorithms can predict the resulting light-curve bias without running full transit fits, and this could be extended to choose the minimal polynomial degree needed to reach a target precision for a given passband and star.","The TRIP exact integrator, being free of limb-darkening approximations, is the natural tool to assess biases in other transit parameters beyond depth; the paper's parameter-recovery tests for HD 209458b hint that these biases are sub-ppm in the QS case, but the same check could be run for grazing transits or small planets where the limb contribution is relatively larger."],"forward_implications":["For any transit modeled with claret-4 coefficients from SAIL, the systematic error contributed by limb-darkening parameterization is below about 10 ppm at all wavelengths, so it will sit under the expected photon noise of TESS, JWST, and ARIEL observations for typical targets.","Two-coefficient laws (quadratic, power-2, square-root) are shown to be unreliable below 1 µm for exoplanet spectroscopy, introducing biases of tens to hundreds of ppm, so the claret-4 law must be preferred at optical and ultraviolet wavelengths.","The proposed formula $(\\text{peak-to-peak})_{\\mathrm{ppm}} = (k \\times 10^6) \\times p^2 \\times (\\text{weighted-r QS rms})$ lets users estimate the systematic noise floor of their light-curve model from the intensity-fit residuals alone, allowing a quick check that limb-darkening systematics stay below the photon-noise limit.","Because the algorithm eliminates the degeneracy among previous fitting methods (unweighted, weighted-μ, interpolated-μ/r, and their QS variants), different analyses of the same star and passband will obtain identical limb-darkening coefficients and hence mutually consistent transit parameters."],"supporting_citations":[{"why":"Supplies the standard analytic transit formalism and the geometry of Equations 11–18 that the paper generalizes, and the baseline light-curve model that limb-darkening coefficients are meant to reproduce.","marker":"Mandel & Agol (2002)"},{"why":"Introduced the quasi-spherical fitting approach (cutoff at μ ≥ 0.1) for spherical PHOENIX models, which the paper generalizes to a rescaled r ≤ 0.99623 cutoff.","marker":"Claret et al. (2012)"},{"why":"Provides the PHOENIX 2012 spherical 1D stellar atmosphere intensity grids used in the HD 209458b validation simulations.","marker":"Claret et al. (2013)"},{"why":"Defines the four-coefficient claret-4 limb-darkening law (Equation 6) and provides the ATLAS plane-parallel intensity grids.","marker":"Claret (2000)"},{"why":"The previous limb-darkening calculator to which SAIL is conceptually similar; its alternative fitting algorithms are the comparison baselines for the residual and parameter-bias tests.","marker":"Espinoza & Jordán (2015)"},{"why":"Describes the 'tlc' algorithm for exact transit light-curve generation; TRIP with default settings is identical to it, and it is used to produce the reference light curves.","marker":"Morello et al. (2017)"},{"why":"Defines the photometric radius as the inflection point of the intensity gradient, the rescaling convention used for spherical intensity profiles before fitting.","marker":"Wittkowski et al. (2004)"},{"why":"Provides the TESS expected photometric noise floor (~60 ppm) used to contextualize the residual amplitudes of alternative fitting methods.","marker":"Ricker et al. (2014)"},{"why":"Provides the JWST/MIRI photon noise floor used to show that the sub-10 ppm precision meets the requirements of future exoplanet spectroscopy.","marker":"Beichman et al. (2014)"},{"why":"Suggests the interp-r 100 fitting method that is one of the comparison algorithms the paper's weighted-r QS method outperforms.","marker":"Parviainen & Aigrain (2015)"}],"fun_headline_variants":["ExoTETHyS cuts transit light-curve error to 10 ppm","Transit models reach 10 ppm precision with ExoTETHyS","New package nails transit models to under 10 ppm","ExoTETHyS: transit light curves accurate to 10 ppm","Transit light-curve fits now accurate to <10 ppm"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The validation is an internal consistency check: reference light curves are produced by TRIP from the same model-atmosphere intensity profiles to which the limb-darkening laws are fitted, so the claimed precision holds only if those atmosphere models (ATLAS, PHOENIX) represent real stellar surfaces.","fun_headline_variants_meta":{"raw":{"variants":["ExoTETHyS cuts transit light-curve error to 10 ppm","Transit models reach 10 ppm precision with ExoTETHyS","New package nails transit models to under 10 ppm","ExoTETHyS: transit light curves accurate to 10 ppm","Transit light-curve fits now accurate to <10 ppm"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000739,"raw_usage":{"total_tokens":3297,"prompt_tokens":941,"completion_tokens":2356,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":2265}},"tokens_in":557,"tokens_out":2356,"duration_ms":17365,"temperature":1.0,"reasoning_tokens":2265,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:06:40.450095+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Generate a noiseless reference transit light curve with TRIP using an intensity profile from an independent stellar atmosphere code not in ExoTETHyS's grids (e.g., a 3D or MESA model) for a star like HD 209458; if claret-4 coefficients from the weighted-r QS fit reproduce that curve with residuals above about 10 ppm, the claimed precision is specific to the grid rather than general.","supporting_citations":[{"cited_title":"2000, A&A, 363, 1081 —","cited_arxiv_id":null,"evidence_quote":"Defines the four-coefficient claret-4 limb-darkening law (Equation 6) and provides the ATLAS plane-parallel intensity grids."},{"cited_title":"R., Winn, J","cited_arxiv_id":null,"evidence_quote":"Provides the TESS expected photometric noise floor (~60 ppm) used to contextualize the residual amplitudes of alternative fitting methods."}],"review_version":1}