{"id":"7e86e786-8aa6-4f4f-ac29-3113c4240990","arxiv_id":"2506.08170","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"The authors present a scalable online portal, udel.edu/atom, that serves high-precision atomic properties for 28 atoms and ions, generated by coupled-cluster and CI+all-order codes with estimated uncertainties.","lead":"This paper introduces an open-access web portal that provides computed atomic data (energies, transition rates, polarizabilities, and more) for 28 atoms and ions, with an automated pipeline for adding new data. It matters because atomic physics, astrophysics, and quantum technology groups need reliable, easy-to-access numerical data with uncertainties.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Uncertainty model (Eq. 20) is uncalibrated against independent measurements, so the 'high-precision' claim is not yet evidenced; reader's conditional verdict stands.","rationale":"The paper's central deliverable is a database of atomic properties with estimated uncertainties; the title's 'high-precision' and the abstract's 'All calculated values include estimated uncertainties' make the uncertainty model the load-bearing element. The model (Sec. 4.2, Eq. 20) sets Δ = sqrt(Δ0^2 + Δmin^2), with Δ0 the CI+all-order vs CI+MBPT difference and Δmin a hand-set floor. This is a standard heuristic in the group's prior work, but it is not calibrated here against independent measurements of the quantities the portal sells (matrix elements, rates, lifetimes, polarizabilities). The only external comparison is with NIST energy levels (Sec. 5.1), which validates state assignments and energies, not transition amplitudes. Two more specific weaknesses sharpen this: (1) CI+all-order and CI+MBPT share the same valence CI model space and one-electron basis, so their difference may not capture common systematic errors; the floor is intended to compensate but is chosen by judgment, not derived. (2) Appendix A.1's config parameter min_uncertainty: 1.5 is described as a percentage, while Sec. 4.2 quotes absolute floors in a.u. (0.015, 0.003, 0.013), leaving the propagation rule underdocumented. These do not demonstrate that the uncertainties are wrong, but they mean the high-precision claim is not yet evidenced. The reader's conditional verdict is appropriate; my concern does not change it. The proposed benchmark comparison (portal values vs. high-precision measured lifetimes/A-values) would settle whether the uncertainty intervals are calibrated.","tokens_in":12367,"tokens_out":10027,"duration_ms":118943,"concrete_test":"Compile a benchmark of 20–30 transitions for the portal's divalent systems (Sr, Ca, Mg) and selected monovalent ions that have high-precision measured A-values or lifetimes (e.g., Sr 5s5p 3P1, Ca 4s4p 3P1, Mg 3s3p 3P1, and alkali D lines). For each transition, compute r = (theory − experiment)/quoted_uncertainty and count the fraction with |r| > 2. If that fraction exceeds 5%, the uncertainty model is miscalibrated. As a second check, verify the Sr 5s5p 3P1 matrix element against the lifetime-derived value: the quoted uncertainty is dominated by Δmin = 0.015 a.u. and should bracket the measured value if the floor is honest.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the portal delivers high-precision atomic data with reliable estimated uncertainties. The load-bearing assumption is the uncertainty model of Sec. 4.2, Eq. (20): Δ = sqrt(Δ0^2 + Δmin^2), where Δ0 is the difference between CI+all-order and CI+MBPT results and Δmin is a hand-set floor (e.g., 0.015 a.u. for Sr). This is not calibrated against independent measurements of matrix elements, rates, or lifetimes; the only external check, the NIST comparison in Sec. 5.1, validates energy-level labeling, not transition amplitudes. Since CI+all-order and CI+MBPT share the same valence CI space and basis, their difference may miss common systematic errors, and the floor values are expert judgments, not derived bounds. Appendix A.1 adds ambiguity: config's min_uncertainty: 1.5 is called a percentage, while Sec. 4.2 lists absolute floors in a.u. If the uncertainty intervals are not calibrated, every derived quantity (rates, lifetimes, polarizabilities, magic wavelengths) inherits an unverified error bar, so the 'high-precision' claim is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript describes the University of Delaware Atom Portal, a web resource delivering energies, electric and magnetic multipole matrix elements, transition rates, radiative lifetimes, branching ratios, hyperfine constants, and scalar/tensor dynamic polarizabilities for 28 atoms and ions. Data are generated by automated workflows built on the authors' all-order coupled-cluster and CI+all-order/CI+MBPT codes (pCI), with a five-step correspondence algorithm that maps computed states to NIST energy levels and prefers NIST energies when available. Uncertainties are assigned by combining the difference between CI+all-order and CI+MBPT results with a hand-set minimum floor. The paper also presents the portal architecture, cloud data pipeline, testing framework, and interactive polarizability plotting interface.","tokens_in":12699,"tokens_out":8150,"duration_ms":99065,"significance":"If the claimed precision and uncertainty estimates are reliable, the portal is a useful community resource for atomic physics, plasma physics, astrophysics, and quantum technology applications, because it makes a broad set of state-of-the-art many-body calculations accessible through a searchable, interactive interface. Strengths include the use of well-established relativistic many-body methods, an automated pipeline with internal consistency checks, NIST-based energy correspondences, and public availability of the underlying pCI software. The core formulas for transition rates, lifetimes, branching ratios, and polarizabilities are standard and correctly presented. However, the central reliability claim is not yet fully evidenced: the uncertainty model is not calibrated against independent measurements of transition amplitudes, and the hand-set floors are not derived from a documented procedure. If the uncertainty estimates are accurate, the portal meets its stated goal; if not, the 'high-precision' claim overstates the data quality. The load-bearing issue is therefore the uncertainty model, not the physics formulas or the software architecture.","major_comments":[{"comment":"The central 'high-precision' claim rests on the uncertainty model of Eq. (20), but the model is not calibrated against independent measurements. The difference Δ0 between CI+all-order and CI+MBPT is a method-spread estimator for two calculations that share the same valence CI space and one-electron basis; common systematic errors can cancel, and the hand-set floors Δmin are expert judgments rather than derived bounds. The NIST comparison in Sec. 5.1 validates energy-level correspondences and state labels, not transition amplitudes. I recommend adding a validation subsection that compares representative matrix elements, lifetimes, and magic or tune-out wavelengths to precision measurements, or explicitly relabeling all quoted uncertainties as method-difference spreads throughout the abstract, Sec. 4.2, and Sec. 7.","section":"Secs. 4.2, 5.1; Eq. (20)"},{"comment":"The pipeline description says that when only one of CI+all-order or CI+MBPT results is present, 'uncertainties are set to 0.' This appears to contradict the abstract's claim that all calculated values include estimated uncertainties. Please state whether any final portal entries are displayed with zero uncertainty, describe how the config-level min_uncertainty interacts with this zero step, and, if zero-error entries exist, qualify the abstract and Sec. 7 claims accordingly.","section":"Appendix A.1, step 1"},{"comment":"The minimum-uncertainty parameters are defined inconsistently: Sec. 4.2 gives Δmin in absolute atomic units (0.015, 0.003, 0.013 a.u. for Sr, Mg, Ca), while Appendix A.1 calls portal.min_uncertainty: 1.5 a minimum uncertainty 'in percentage.' The manuscript does not specify the reference value for the percentage or how the per-system absolute floors are reconciled with the global percentage default. This must be clarified for the uncertainty pipeline to be reproducible.","section":"Sec. 4.2 and Appendix A.1"},{"comment":"The propagation of matrix-element uncertainties into derived quantities is not documented. Eq. (20) defines Δ for matrix elements, but no formulas are given for the uncertainty of transition rates, lifetimes, polarizabilities, or magic wavelengths; the text only states that these are obtained from Δ. In particular, rates are computed with NIST energies when available, and the manuscript does not state whether NIST energy uncertainties are propagated into wavelengths. Please supply the full uncertainty-propagation chain or state the assumptions under which neglected terms are small.","section":"Secs. 2.2, 2.3"}],"minor_comments":[{"comment":"The text contains numerous typographical errors, including 'di fferent' (Secs. 2.2 and 4.1), 'o ffer' (Sec. 2.4), 'V ol.' (Ref. [1]), and the informal 'We've' in the abstract; these should be corrected.","section":"Throughout"},{"comment":"The magic-wavelength uncertainty procedure is described only verbally; please specify precisely which intersections of the α±δ curves are used to set the upper and lower bounds.","section":"Sec. 2.3 and Fig. 2"},{"comment":"For magnetic multipole transitions, the line strength S(Tk) should be defined with the relevant units for matrix elements expressed in μ0, because Eqs. (3)–(8) are written without an explicit unit convention.","section":"Sec. 2.2"},{"comment":"The NIST ASD reference contains an informal access date ('Mon Jun 20 2022'); please format it according to the journal style.","section":"Ref. [21]"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the scope of a software/database paper in computational physics. The uncertainty-validation issue is the main risk; it can be addressed in revision by adding a calibration section and clarifying the pipeline, and the thesis 'high-precision' claim can be softened if independent calibration is not yet available. I do not see grounds for rejection, but the current wording of the abstract and conclusions overstates what Eq. (20) and the NIST correspondence procedure actually establish."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The real contribution is the portal and its automation: a live, open, no-registration atomic data site covering 28 atoms and ions, an automated pipeline that turns pCI output into PostgreSQL tables, and a five-step NIST correspondence algorithm for mapping computed states to experimental levels. That is genuinely useful, and the paper is honest about being a resource paper rather than a new-physics claim. The testing framework—internal CSV consistency checks plus Selenium-based inter-version tests—is a real plus, and the code lives in the pCI GitHub repo, which is more than many portal papers offer.\n\nThe soft spot is the uncertainty model. Equation (20) composes a difference between CI+all-order and CI+MBPT results with hand-set floors (0.015 a.u. for Sr, 0.003 for Mg, 0.013 for Ca). Those floors are expert judgment, not derived bounds, and the two methods share the same valence CI space and basis, so common systematic errors may not appear in their difference. The NIST comparison in Sec. 5.1 validates energy-level labeling, not transition amplitudes, so it does not calibrate matrix-element uncertainties. Also, Appendix A.1's config.yml shows min_uncertainty: 1.5 as a percentage, while Sec. 4.2 gives absolute floors in a.u. That inconsistency should be fixed. If the error bars are off, every derived quantity—rates, lifetimes, polarizabilities, magic wavelengths—inherits an unverified uncertainty, which matters for a portal marketed as high-precision.\n\nThat said, the underlying methods are established in prior work, and the portal is transparent about data provenance. This is a calibration/validation gap, not sloppy calculation. For a resource paper, I'd want a commitment to archival versioning (commit hash or DOI) and, ideally, a comparison against independent experimental matrix elements or lifetimes for a few key transitions.\n\nWho benefits: atomic physicists, optical clock people, plasma and astrophysics data users. It deserves a serious referee—it is a worthwhile community resource—but the referee should push on uncertainty calibration and the Appendix inconsistency. I would accept it after those revisions.","headline":"A genuinely useful community portal with an honest resource-paper framing, but the uncertainty model needs calibration before the 'high-precision' claim is taken at face value.","tokens_in":13193,"tokens_out":1734,"would_cite":true,"duration_ms":21047,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper reports a free, open-access web portal that packages high-precision atomic data for 28 atoms and ions, produced by an automated pipeline in which every calculated value carries an estimated uncertainty.","keywords":["atomic data portal","CI+all-order method","coupled-cluster all-order","transition matrix elements","polarizabilities","magic wavelengths","NIST Atomic Spectra Database","uncertainty quantification"],"falsifier":"Download the portal's E1 matrix elements or lifetimes for a specific element such as Sr, and compare them transition by transition with the most precise experimental measurements available for those same transitions; if the measured values fall outside the quoted one-sigma uncertainties for a substantial fraction of transitions, the uncertainty model fails. A cleaner version is to recompute one portal system with an independent method family, for example a B-spline or multiconfiguration Dirac-Fock code, and check whether the two calculations agree within the portal's quoted error bars on every published value.","tokens_in":1744,"feed_emoji":"⚛️","tokens_out":2811,"duration_ms":99492,"temperature":0.7,"pith_summary":"This paper reports a working web resource that turns high-precision atomic structure calculations into an automated, steadily growing data service. The portal currently covers 28 atoms and ions and provides energies, transition matrix elements, rates, lifetimes, branching ratios, polarizabilities, and hyperfine data, with an estimated uncertainty attached to every calculated value. The authors' central claim is that the underlying pipeline, from coupled-cluster and configuration-interaction computations to database ingestion and display, can run without hand intervention, including the step that matches computed states to the NIST Atomic Spectra Database and flags what theory provides that NIST does not. If the claim holds, the atomic, plasma, and astrophysics communities gain a single open site where precision data and their error bars can be browsed, plotted, and downloaded, and where new systems can be added on demand.","feed_headline":"Open portal hosts high-precision atomic data for 28 atoms and ions","feed_subtitle":"Every calculated value carries an estimated uncertainty, checked automatically against the NIST database.","key_machinery":"The load-bearing mechanism is the automated portal pipeline built around the pCI software package. Two new scripts, gen_portal_csv.py and calc_lifetimes.py, take raw output from pCI computations, attach uncertainties by differencing CI+all-order and CI+MBPT results subject to a minimum-uncertainty floor, $\\Delta = \\sqrt{\\Delta_0^2 + \\Delta_{\\rm min}^2}$, correct misidentified configurations through the five-step NIST correspondence procedure, and emit CSV files that the portal ingests into a PostgreSQL database. A separate Flask backend computes dynamic polarizabilities, magic wavelengths, and tune-out wavelengths on demand via the Sternheimer or Dalgarno-Lewis inhomogeneous-equation approach, so the portal never has to store millions of precomputed polarizability rows. The pipeline's claim to scalability rests on this separation: computation, ingestion, validation, and display are each automated steps that a new element's data pass through without human editing.","core_discovery":"The paper's discovery is that precision atomic data production can be industrialized without losing per-value uncertainty estimates. Using the relativistic all-order (coupled-cluster) method for monovalent systems and the CI+all-order method for multivalent ones, with CI+MBPT run alongside as an error benchmark, the authors generate energies, reduced matrix elements, transition rates, lifetimes, branching ratios, polarizabilities, and hyperfine constants for 28 systems. A five-step state-correspondence algorithm matches each computed level to the NIST Atomic Spectra Database by angular momentum, term symbol, multiplicity (allowing plus or minus one), and configuration, in that order, so that misidentified computational output is corrected automatically and NIST energies are used wherever they exist. Uncertainties on matrix elements are computed in quadrature from the CI+all-order versus CI+MBPT difference plus a per-system floor, and every portal value is displayed as value(uncertainty).","pith_inferences":["A testable extension is to run a third, independent method, for example a different all-order variant or high-precision experimental transition rates, over the portal's full element list; if deviations exceed the quoted one-sigma uncertainties for even one element family of matrix elements, the floor-based uncertainty model is under-covering systematic error.","The correspondence algorithm's preference ordering could silently misassign states in dense, strongly mixed spectra where the primary configuration label in NIST differs from the dominant configuration-interaction eigenvector; a natural safeguard would be to expose the residual energy disagreement of every match, not just the final state.","The same architecture likely generalizes beyond atoms: the ingest-validate-display pipeline is agnostic to the physics, so molecular or nuclear data with the same CSV schemas could ride the same portal infrastructure.","The paper's version-4 target of 100 atomic systems implies that the hand-set minimum uncertainties (for example 0.015 a.u. for Sr) will eventually need to become data-driven rather than per-element manual inputs if the quoted error bars are to stay meaningful at scale."],"forward_implications":["If the pipeline works as claimed, adding a new atomic system reduces to running pCI, uploading CSVs to a shared folder, and letting ingestion scripts validate and publish the pages.","Every portal value carrying a stated uncertainty means users such as clock builders, astrophysicists, and plasma modelers can propagate error bars into their own calculations instead of treating the dataset as exact.","The five-step NIST correspondence lets theory supplement the NIST database: states pCI finds that NIST lacks are kept with estimated theoretical uncertainties, extending coverage beyond measured spectra.","The on-demand polarizability service with magic-wavelength and tune-out crossing finding makes laser-wavelength selection for optical clocks and traps a browser-side query rather than a bespoke computation."],"supporting_citations":[{"why":"pCI, the parallel configuration-interaction software package that the whole computation and automation pipeline runs on.","marker":"[16]"},{"why":"Introduces the CI+all-order method used for multivalent systems on the portal.","marker":"[9]"},{"why":"Source of the uncertainty procedure that differences CI+all-order and CI+MBPT results and applies a minimum floor, as developed for Sr clock shifts.","marker":"[10]"},{"why":"NIST Atomic Spectra Database, the experimental reference every computed level is matched against and the source of energies where available.","marker":"[21]"},{"why":"The all-order (coupled-cluster) methods review that supplies the monovalent computation framework.","marker":"[1]"},{"why":"Sternheimer's inhomogeneous-equation method used for on-the-fly valence polarizability calculations.","marker":"[23]"},{"why":"Dalgarno-Lewis perturbation approach, the companion method for the polarizability solver.","marker":"[24]"}],"fun_headline_variants":["Atomic data portal: 28 systems, every value with uncertainty","Automated precision for 28 atoms: coupled-cluster plus NIST checks","New atomic data hub: 28 species, auto-validated against NIST","Coupled-cluster data for 28 ions, each with estimated error","Industrial-scale atomic data: 28 systems, uncertainties built-in"],"cache_read_input_tokens":15360,"weakest_assumption_plain":"The 'high-precision' claim rests on the assumption that a matrix element's uncertainty is honestly given by the difference between two theory methods plus a hand-set per-element floor; if that difference hides a common bias, every error bar on the portal is too small.","fun_headline_variants_meta":{"raw":{"variants":["Atomic data portal: 28 systems, every value with uncertainty","Automated precision for 28 atoms: coupled-cluster plus NIST checks","New atomic data hub: 28 species, auto-validated against NIST","Coupled-cluster data for 28 ions, each with estimated error","Industrial-scale atomic data: 28 systems, uncertainties built-in"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000904,"raw_usage":{"total_tokens":3851,"prompt_tokens":870,"completion_tokens":2981,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":486,"completion_tokens_details":{"reasoning_tokens":2886}},"tokens_in":486,"tokens_out":2981,"duration_ms":21779,"temperature":1.0,"reasoning_tokens":2886,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:16:39.444200+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Download the portal's E1 matrix elements or lifetimes for a specific element such as Sr, and compare them transition by transition with the most precise experimental measurements available for those same transitions; if the measured values fall outside the quoted one-sigma uncertainties for a substantial fraction of transitions, the uncertainty model fails. A cleaner version is to recompute one portal system with an independent method family, for example a B-spline or multiconfiguration Dirac-Fock code, and check whether the two calculations agree within the portal's quoted error bars on every published value.","supporting_citations":[{"cited_title":"Cheung, M","cited_arxiv_id":null,"evidence_quote":"pCI, the parallel configuration-interaction software package that the whole computation and automation pipeline runs on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the CI+all-order method used for multivalent systems on the portal."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Source of the uncertainty procedure that differences CI+all-order and CI+MBPT results and applies a minimum floor, as developed for Sr clock shifts."},{"cited_title":"Safronova, W","cited_arxiv_id":null,"evidence_quote":"The all-order (coupled-cluster) methods review that supplies the monovalent computation framework."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Sternheimer's inhomogeneous-equation method used for on-the-fly valence polarizability calculations."},{"cited_title":"Dalgarno, J","cited_arxiv_id":null,"evidence_quote":"Dalgarno-Lewis perturbation approach, the companion method for the polarizability solver."}],"review_version":1}