{"id":"434c9925-212e-4c34-9d67-200c0cb042a0","arxiv_id":"2507.17082","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"New covariant density functionals include electron binding energies and infinite basis corrections in the fitting protocol, shrinking the global calculation error from about 0.8 MeV to around 25 keV.","lead":"This paper improves nuclear binding energy fits by correcting two effects that earlier covariant density functional fits ignored: the electron contribution to atomic masses, and the error from truncating the single-particle basis. These corrections shift results by up to several MeV and reduce a newly defined global calculation error from roughly 0.8 MeV to tens of keV.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 0.8 MeV error estimate is an uncontrolled Y-vs-Z functional difference: changes in AME version, anchor data, and charge-radius expression are not ablated, so the claimed attribution to basis/Bel neglect is not established.","rationale":"The reader's verdict is CONDITIONAL, and the reader's weakest_assumption already names the core issue: the 0.8 MeV estimate requires that the Y-Z difference is entirely due to the new protocol ingredients and that the extrapolation is reliable. My stress-test sharpens this into a concrete attribution problem: the Y and Z functionals differ in at least four dimensions (AME version, number of nuclei, charge-radius expression, and the claimed corrections), and no ablation isolates the basis/Bel corrections. This matters because δB_negl is not an experimental residual; it is an rms difference between two fitted models. If data-set or charge-radius changes contribute meaningfully to this difference, the central claim that 'neglect leads to errors of order 0.8 MeV' is weaker than stated. The paper's own convergence benchmarks for DD-MEZ and NL5(Z) are genuine supporting evidence, and the iterative procedure reaching δ(ΔB_F∞) ≈ 8 keV is a real strength. The PC branch, however, is explicitly biased and unquantified, so the abstract's 'three major classes' formulation oversells the PC support. None of this changes the verdict from CONDITIONAL: the new functionals, convergence tests, and iterative protocol are credible contributions, but the 0.8 MeV headline should be read as a plausible, benchmark-dependent model-sensitivity estimate pending a controlled decomposition.","tokens_in":25765,"tokens_out":8059,"duration_ms":97138,"concrete_test":"Run a controlled DDME fit ladder with all other inputs fixed: (i) refit DD-MEY with the Z protocol data (AME2020, same 12 anchors, full Eq. (1) charge radii) but keeping NF=20/NB=20 and no Bel subtraction; (ii) repeat with Bel subtraction only; (iii) repeat with infinite-basis corrections only; compare each stage's rms binding-energy difference from the original DD-MEY and from DD-MEZ. If stage (i) alone shifts binding energies by more than 0.2 MeV rms, the 0.8 MeV headline is not attributable to the neglected physical corrections; if stage (i) is below 50 keV and stages (ii)/(iii) reproduce Fig. 9(a), the attribution is validated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline error estimate is defined in Sec. VI E as δB_negl^rms = rms[B(Y) − B(Z)] (Eq. 10, Fig. 9). This is a model-sensitivity measure between two independently optimized functionals, not a direct measurement of the error caused by the three neglected ingredients. The Y and Z fits differ in several controlled variables beyond basis truncation and Bel: the AME evaluation (2016 vs 2020, 853 vs 882 even-even nuclei), the anchor set, and the charge-radius functional used in the fit (full Eq. (1) including spin-orbit terms for Z, first terms only for Y). The paper asserts the AME update 'does not play a principal role' but provides no ablation or sensitivity analysis. Parameter changes between Y and Z (Tables V, VII, IX) are large for density-dependent couplings, so the 0.8 MeV rms could partly reflect shallow-minimum/parameter-shift effects or data-set changes rather than the physics corrections. The PC case is explicitly weaker: Sec. V and VI C state PC-Z corrections cannot be defined for actinides/superheavy nuclei, the fit is biased, and Sec. VI E says δB_negl cannot be accurately quantified for PC functionals; the abstract's '0.8 MeV or higher for the three major classes' therefore overstates support for the PC class. Thus the central claim is plausible but not pinned down.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents three modifications to the fitting protocol of covariant energy density functionals (CEDFs): (i) inclusion of infinite-basis corrections to binding energies in both fermionic and bosonic sectors, (ii) subtraction of total atomic electron binding energies when converting AME atomic binding energies into nuclear ones, and (iii) use of the full charge-radius formula with spin-orbit terms for anchor nuclei. New functionals DD-MEZ, NL5(Z), and PC-Z are fitted within the anchor-based optimization approach and compared with previous DD-MEY, NL5(Y), and PC-Y fits. The authors report that neglect of the basis corrections and electron binding energies in earlier fits produces a global calculation error δB_negl of about 0.8 MeV for the DDME and NLME classes, and larger than 1.0 MeV for the PC class, while the new Z functionals have δB_negl of about 25–30 keV. The new DD-MEZ functional yields ΔB_rms = 1.601 MeV over 882 even–even nuclei, improving to 1.557 MeV when a Wigner term is added.","tokens_in":26133,"tokens_out":8951,"duration_ms":82402,"significance":"If the central claim is correct, it identifies a hidden systematic error of ~0.8 MeV in previous CDFT mass tables, comparable to the reported rms deviations and thus potentially affecting conclusions drawn from those tables. The paper is also significant for introducing the first chart-wide calculations of infinite-basis corrections in both sectors, for analyzing the neutron-excess dependence of electron binding energies, and for providing new global fits that reach the best mean-field accuracy among CEDFs. The convergence benchmarks (NF = 40/60, NB = 40/120) and the iterative ABOA procedure for defining basis corrections are well documented and could be reused by other groups. The main weakness is that the 0.8 MeV error estimate is derived from a Y-vs-Z functional comparison that does not isolate the three protocol changes, so the causal attribution is not fully established.","major_comments":[{"comment":"The quantity δB_negl^rms defined in Eq. (10) and displayed in Fig. 9 is computed as rms[B(Y) − B(Z)], i.e., as a difference between two independently optimized functionals. However, the Y and Z fits differ not only in the treatment of basis truncation and electron binding energies but also in the AME evaluation (2016 vs 2020; 853 vs 882 even–even nuclei), in the anchor data values taken from those evaluations, and in the charge-radius expression (full Eq. (1) including spin-orbit terms for Z, first three terms only for Y). The statement in Sec. VI E that the AME update “does not play a principal role” is an assertion with no supporting ablation or sensitivity analysis. Large parameter changes for density-dependent couplings (e.g., c_ω ratio 1.439 in Table V) suggest that the observed 0.8 MeV differences may partly reflect shallow-minimum parameter shifts or data-set changes rather than the physics corrections. To support the central claim, the authors should provide controlled refits that vary one ingredient at a time, or demonstrate that the AME version and charge-radius expression have negligible effect on δB_negl^rms.","section":"Sec. VI E, Eq. (10), Fig. 9"},{"comment":"The abstract states that the neglect of the new ingredients leads to errors “of the order of 0.8 MeV or higher for the three major classes” of CEDFs. For the DDME and NLME classes, the paper gives δB_negl^rms = 0.77 and 0.84 MeV (Fig. 9(a,b)), though with the attribution caveat above. For the PC class, however, Sec. VI E explicitly states that “at present it is impossible to accurately quantify the δB_negl^rms values,” and Sec. VI C explains that the PC-Z fit is biased by the exclusion of actinides and superheavy nuclei from the correction function because PC functionals do not converge at NF = 40 (Fig. 2). The PC estimate “larger than 1.0 MeV” from Fig. 9(c) is therefore not an accurate quantification and should not be used to support the abstract’s “three major classes” claim. The wording of the abstract and conclusions should be revised to restrict the quantitative 0.8 MeV statement to the DDME and NLME classes, or the authors should provide a reliable PC error estimate that accounts for the biased fitting region.","section":"Abstract, Secs. V, VI C, VI E"},{"comment":"The infinite-basis corrections ΔBF∞(Z,N) are defined through the functional being fitted (Eq. (8)) and are updated iteratively in the ABOA rounds. The validation of the extrapolation procedure is performed against NF = 40 results for a “testing set of nuclei scattered more or less equally across experimentally known nuclear landscape,” with claimed global accuracy better than 15 and 10 keV for DD-MEZ and NL5(Z), respectively. These numbers, however, apply only to the selected test nuclei; the procedure is then applied to all 882 nuclei without an explicit chart-wide uncertainty estimate. Because the central error budget for the Y functionals (δB_negl^rms ≈ 0.8 MeV) is much larger than this validation accuracy, this is not the main limitation, but the paper should report how the extrapolation uncertainty propagates into the 28 keV (DD-MEZ) and 23 keV (NL5(Z)) estimates for the Z functionals, and should state more carefully that the 15/10 keV values are validation-set accuracies rather than global uncertainties.","section":"Sec. V, Supplemental Sec. III"},{"comment":"The same methodological issue that affects δB_negl^rms for binding energies also affects the reported global calculation errors for charge radii and separation energies. The values δ(rch)_negl^rms = 0.010 fm (DD-MEY) and δ(S2n)_negl^rms = 0.387 MeV (DD-MEY) are obtained from the Y–Z differences in the same uncontrolled comparison. Unless the authors provide an ablation separating the contributions of the three protocol changes, these quantities should be described as model-sensitivity measures rather than as well-defined “global calculation errors,” or the analysis should be extended to control the other variables.","section":"Sec. VI E"}],"minor_comments":[{"comment":"The manuscript text contains “d ensity” in the title and running header; this should be corrected to “density.”","section":"Title, Sec. I"},{"comment":"The use of “±” to combine ΔB_rms and δB_negl^rms is not standard; please specify whether the stated uncertainty is one sigma, a quadrature sum, or the rms of the difference itself.","section":"Sec. VI E, Eq. (10)"},{"comment":"The table classifies existing functionals as atomic or nuclear, but the new Z functionals are not listed; considering they are the main subject of the paper, they should be included or referenced.","section":"Table III"},{"comment":"The conclusion that the isotopic dependence of Bel is negligible is based on only three chains (Pb, Fm, Og); a sentence on why these are representative would be useful.","section":"Sec. IV"},{"comment":"The color scales for the two panels are different (max 0.30 vs 1.0 MeV), which makes visual comparison of the magnitude of bosonic corrections between DD-MEZ and NL5(Z) difficult; a common scale would improve clarity.","section":"Fig. 5"},{"comment":"This footnote contains a substantial quantitative result (the additional 0.199 MeV contribution for PC-PK1) that is relevant to the main text; consider moving it into the main body.","section":"Sec. VI E, footnote 4"},{"comment":"“Eliminate numerical uncertainties” is too strong; the text in Sec. VI E states residual errors of ~25–30 keV for the Z functionals. Rephrase to “substantially reduce.”","section":"Sec. VII, first bullet"}],"recommendation":"major_revision","confidential_remarks":"The paper is from the group that developed ABOA and the electron binding energy calculations in Ref. [67], so the use of their own atomic tables and iterative corrections in the fitting protocol is understandable but should be disclosed more prominently. The main issue is not novelty but the lack of an ablation in the Y–Z comparison; this is fixable. If the authors can add a controlled sensitivity study, the paper would be a strong candidate for acceptance. I also note that the claim of being “first” for several items may be challenged by the non-relativistic DFT literature, where similar corrections have been used; the comparison in Sec. III is helpful in this regard."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuine technical advance—first CEDF fits that include infinite basis corrections in both fermionic and bosonic sectors and subtract total electron binding energies from AME data—but the paper's headline number, 0.8 MeV global error, is an indirect measure. It comes from comparing two independently optimized functionals (Y vs Z), not from an ablation of the three new ingredients. The referee should push for an isolation of those effects.\n\nWhat is new and good: the iterative ABOA procedure for defining basis corrections is clearly described, with convergence checks at NF=40/60 and NB=40/120 that are meaningful. The PC-Z limitation is admitted up front: the correction function cannot be defined for actinides and superheavy nuclei, and the fit is biased accordingly. The atomic calculations behind the electron binding energies are state of the art, and the demonstration that isotope shifts in Bel are <1 keV justifies using Z-only values. The new functionals themselves are new, with tables of parameters and nuclear matter properties. That is all solid.\n\nThe soft spot: Eq. (10) and Fig. 9 define δB_negl as the rms difference between Y and Z binding energies. The Y and Z fits differ in more than the new protocol: AME2016 vs AME2020 (853 vs 882 nuclei), the charge-radius expression (full Eq. (1) for Z, truncated for Y), and possibly the anchor set. The paper says the AME update 'does not play a principal role' but gives no sensitivity test. Parameter shifts are large for density-dependent couplings, so part of the 0.8 MeV may be normal functional re-fitting, not the specific physics. The abstract's '0.8 MeV or higher for three major classes' is also ahead of the evidence: Section VI E explicitly says PC errors cannot be accurately quantified, and the PC-Z fit is biased. The PC comparison gives >1 MeV but the authors themselves call it a conservative estimate.\n\nThere is also a self-referential aspect: the corrections come from the same group's atomic tables and are defined through the functionals being fitted. The iterative scheme converges (89 keV -> 16 keV -> 8 keV rms), so this is not fatal, but it is worth noting.\n\nThe citation pattern is dominated by the group's own ABOA papers and atomic calculations, but that is appropriate here since the work builds directly on them.\n\nWho this is for: anyone fitting CEDFs or using CDFT mass tables. It deserves peer review, but the referee should require an ablation or at least a sensitivity analysis that separates the effects of AME update, charge-radius expression, and the basis/Bel corrections. The abstract should be softened for the PC class.\n\nRecommendation: send for peer review with revision.","headline":"A real methodological step in CEDF fitting, but the headline 0.8 MeV error is an indirect model-difference measure, not a clean attribution.","tokens_in":26691,"tokens_out":3475,"would_cite":true,"duration_ms":36523,"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":"The paper claims that earlier covariant nuclear mass fits carried a hidden systematic error of about 0.8 MeV because they ignored infinite-basis corrections and total electron binding energies, and shows that a corrected fitting protocol…","keywords":["covariant density functional theory","nuclear binding energies","nuclear mass tables","infinite basis corrections","electron binding energies","anchor-based optimization","relativistic Hartree-Bogoliubov","binding energy rms deviations"],"falsifier":"Run the new functionals on deformed actinide and superheavy nuclei such as 240Pu and 290Lv in fermionic bases of 56 or 60 shells without pairing and with a converged bosonic basis, then compare the resulting binding energies with the extrapolated pseudodata; if the differences cluster systematically above about 100 keV, the extrapolation and the 0.8 MeV error estimate would need revision.","tokens_in":25573,"feed_emoji":"⚛️","tokens_out":12693,"duration_ms":120799,"temperature":0.7,"pith_summary":"Accurate nuclear masses are the main calibration target for covariant density functional theory, but the paper says that previous global fits of those functionals were aiming at a slightly wrong quantity: the experimental table gives atomic binding energies, not nuclear ones, and the numerical equations were solved in a truncated basis. The paper shows that these two omissions are not small, producing a hidden global error of about 0.8 MeV or more in binding energies across the nuclear chart. It removes the omissions by subtracting total electron binding energies and by adding extrapolated infinite-basis corrections to the fit data, producing three new functionals DD-MEZ, NL5(Z) and PC-Z. The best of these, DD-MEZ, reaches 1.557 MeV rms deviation over 882 even-even nuclei when a Wigner term is included, with the remaining numerical error near 0.025 MeV. If the paper is right, no high-precision covariant mass fit can afford to leave these corrections out.","feed_headline":"Hidden 0.8 MeV error sits in earlier nuclear mass fits","feed_subtitle":"New fits that subtract electron binding and basis limits cut that error to under 0.03 MeV.","key_machinery":"The load-bearing object is the correction function $\\Delta B_{\\rm cor}(Z,N) = B_{\\rm el}(Z) + \\Delta B^F_\\infty(Z,N)$ used to generate pseudodata for the fit. $B_{\\rm el}(Z)$ is the total energy needed to strip all electrons from an atom with $Z$ protons, converting the experimental table's atomic binding energies into nuclear ones. $\\Delta B^F_\\infty$ is the difference between the binding energy computed in an infinite fermionic basis and in the truncated $N_F=20$ basis used in practice; since infinite bases are numerically impossible, the paper extrapolates from runs with up to $N_F=40$ (and $N_F=60$ without pairing), benchmarks the extrapolation on test nuclei to better than 15 keV for DD-MEZ and 10 keV for NL5(Z), and iterates the correction inside the anchor-based optimization until it stabilizes below 8 keV. The bosonic sector is handled more directly by using $N_B=40$ full shells instead of the standard $N_B=20$, which brings the bosonic basis error below 10 keV everywhere. This combination turns the fitted parameters into quantities defined against infinite-basis, purely nuclear binding energies.","core_discovery":"The central claim is that the accuracy budget of covariant energy density functionals is controlled, at the level of around 0.8 MeV, by three protocol details that are not part of the nuclear interaction: the truncation of the harmonic-oscillator basis in the fermionic and bosonic sectors, and the use of atomic rather than nuclear binding energies as fit targets. To remove them, the paper fits to pseudodata $B_{\\rm pseudo}(Z,N) = B_{\\rm AME}(Z,N) - [B_{\\rm el}(Z) + \\Delta B^F_\\infty(Z,N)]$, where $B_{\\rm el}$ is the total electron binding energy and $\\Delta B^F_\\infty$ is an extrapolated correction from the truncated fermionic basis to the infinite limit. The correction is redetermined iteratively inside the anchor-based optimization, with convergence judged by the global rms change between rounds falling below 8 keV. On this basis the paper constructs new functionals in all three classes, DD-MEZ, NL5(Z) and PC-Z, and reports the first evaluation of the global calculation error of earlier fits: about 0.77 MeV for DD-MEY, 0.84 MeV for NL5(Y), and more than 1 MeV for the point-coupling class. DD-MEZ improves the rms deviation from 1.802 MeV to 1.601 MeV (1.557 MeV with Wigner energy) while shrinking its own numerical error to roughly 0.025 MeV. The conclusion is that the older fits were not just less convenient but systematically biased, and that convergence to the infinite-basis limit plus the electron correction is a prerequisite for sub-MeV covariant mass tables.","pith_inferences":["If the paper is right, a similar audit would be worth doing for non-relativistic mass fits that also rely on truncated harmonic-oscillator bases: part of their reported agreement with experiment may likewise come from absorbing basis and electron-conversion errors into fitted parameters rather than removing them.","The Y-versus-Z parameter shifts in the paper suggest a testable prediction: the difference between old and new functionals should be concentrated in shell-correction regions, especially around 208Pb and the actinides, where single-particle level ordering most strongly changes binding energies.","The iterative correction procedure implies that infinite-basis corrections cannot be tabulated once and reused for a different functional; reusing another functional's correction map can introduce errors of a few hundred keV in heavy nuclei."],"forward_implications":["Previous covariant mass tables based on meson-exchange functionals should be treated as carrying a hidden global error of about 0.8 MeV on top of their reported rms deviations, which changes how their agreement with experiment is interpreted.","The corrected DD-MEZ functional becomes the best covariant mean-field functional for global binding energies, reaching an rms deviation of 1.557 MeV over 882 even-even nuclei when a Wigner term is included, with a residual numerical error near 0.025 MeV.","The iterative procedure shows that infinite-basis corrections must be recomputed after each refit: a one-shot correction map can be off by almost 90 keV globally and by up to about 0.4 MeV in heavy nuclei, so static correction tables are not reliable.","Within the experimentally known region, total electron binding energies can be treated as a function of proton number alone, because their isotopic variation stays below about 1 keV even in superheavy chains.","For point-coupling functionals, the harmonic-oscillator basis converges too slowly to define infinite-basis corrections in actinides and superheavy nuclei; the new PC-Z fit therefore covers only part of the chart and still carries a global error above 1 MeV."],"supporting_citations":[{"why":"Supplies the extrapolation method for infinite fermionic basis corrections and the survey of basis truncation errors that the new fitting protocol builds on.","marker":"[20]"},{"why":"Introduces the anchor-based optimization approach and defines the Y-type functionals that serve as the comparison baseline for the new Z-type fits.","marker":"[22]"},{"why":"Provides the total electron binding energies for atoms with Z=2-120 that the paper subtracts when converting atomic binding energies into nuclear ones.","marker":"[67]"},{"why":"Provides the AME2020 atomic mass evaluation, the experimental source for the 882 even-even nuclei and their binding energies used in the pseudodata.","marker":"[28]"},{"why":"Defines the PC-PK1 point-coupling functional fitted in a finite basis; the paper uses it to assess the slow basis convergence of the point-coupling class.","marker":"[45]"},{"why":"Documents the roughly 0.5 MeV global accuracy of a modern non-relativistic mass fit, the level of precision that makes a 0.8 MeV hidden error unacceptable.","marker":"[13]"},{"why":"Provides the 0.8 MeV-level non-relativistic fit accuracy used by the paper as the benchmark for what covariant fits should reach.","marker":"[91]"}],"fun_headline_variants":["Nuclear mass fits corrected: electron and basis errors slashed","0.8 MeV bias fixed in covariant density functionals","New functional fits shave mass error tenfold","Infinite-basis and electron corrections overhaul nuclear fits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline error estimate depends on the extrapolation that turns finite-basis calculations into infinite-basis binding energies being accurate to tens of keV across the whole chart, including the heavy, deformed, and superheavy nuclei where the paper's own basis runs do not fully converge.","fun_headline_variants_meta":{"raw":{"variants":["Nuclear mass fits corrected: electron and basis errors slashed","0.8 MeV bias fixed in covariant density functionals","New functional fits shave mass error tenfold","Infinite-basis and electron corrections overhaul nuclear fits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000229,"raw_usage":{"total_tokens":1527,"prompt_tokens":1040,"completion_tokens":487,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":656,"completion_tokens_details":{"reasoning_tokens":422}},"tokens_in":656,"tokens_out":487,"duration_ms":5499,"temperature":1.0,"reasoning_tokens":422,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:57:17.190472+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the new functionals on deformed actinide and superheavy nuclei such as 240Pu and 290Lv in fermionic bases of 56 or 60 shells without pairing and with a converged bosonic basis, then compare the resulting binding energies with the extrapolated pseudodata; if the differences cluster systematically above about 100 keV, the extrapolation and the 0.8 MeV error estimate would need revision.","supporting_citations":[{"cited_title":"Note that to our knowl- edge this is the first time when all terms of Eq","cited_arxiv_id":null,"evidence_quote":"Supplies the extrapolation method for infinite fermionic basis corrections and the survey of basis truncation errors that the new fitting protocol builds on."},{"cited_title":"Taninah and A","cited_arxiv_id":null,"evidence_quote":"Introduces the anchor-based optimization approach and defines the Y-type functionals that serve as the comparison baseline for the new Z-type fits."},{"cited_title":"Total electron binding energies are different for differ- ent isotopes due to the difference in nuclear charge ra- dius (field shift) and nuclear mass (mass shift)","cited_arxiv_id":null,"evidence_quote":"Provides the total electron binding energies for atoms with Z=2-120 that the paper subtracts when converting atomic binding energies into nuclear ones."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the AME2020 atomic mass evaluation, the experimental source for the 882 even-even nuclei and their binding energies used in the pseudodata."},{"cited_title":"Gonzalez-Boquera, M","cited_arxiv_id":null,"evidence_quote":"Provides the 0.8 MeV-level non-relativistic fit accuracy used by the paper as the benchmark for what covariant fits should reach."}],"review_version":1}