{"id":"62e0d662-9ce6-49fe-ad50-7b318abcc025","arxiv_id":"2607.19920","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"For a specific colloidal-nematic system, the error between full and reduced Landau–de Gennes models scales roughly as (neglected constant / GOS threshold)^0.8–0.9, providing a threshold criterion for one-constant approximations.","lead":"This paper derives quantitative thresholds for when the Landau–de Gennes liquid-crystal elastic energy can be reduced from three elastic constants to two or one, and tests these thresholds in a single colloidal simulation. The thresholds could give computational physicists a principled rule for choosing simplified models, but the numerical validation covers only one geometry and the error scaling is fitted rather than predicted.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"GOS threshold for L3 lies outside the Ericksen-admissible range for the chosen parameters, so the numerical validation never tests the regime where Eq. (5) would fail.","rationale":"The central claim has two parts: (i) GOS yields thresholds Eq. 5/7, and (ii) openQmin confirms that below these thresholds the model error is O(λ_p). The reader's concern focuses on numerical resolution (Remark 9) and narrow parameter coverage. I share that concern, but the more decisive issue is that the chosen test geometry makes Eq. 5 non-falsifiable: for the Section IV parameters, the GOS threshold L3≈9.276 pN exceeds the Ericksen upper bound 2L1≈7.826 pN. Thus every admissible L3 is already below threshold; the boundary where the reduced model should start to fail lies outside the model's stability domain. The numerical sweep nevertheless includes λ3=1 (Appendix C, Table IV), a state violating the paper's own constraint K24<2K2. Under these conditions, the log-log fit of µ0 vs λ3 cannot be read as confirmation that Eq. 5 is the correct validity boundary—it only shows that small L3 gives small deviations. A re-run with parameters chosen so that L3 lies inside the Ericksen interval and covers both sides of the threshold would settle the matter. This does not change the overall verdict: the paper is promising but conditional on corrective validation; hence verdict unchanged.","tokens_in":18504,"tokens_out":14857,"duration_ms":156822,"concrete_test":"Choose a geometry/anchoring such that the GOS threshold lies inside the Ericksen-stability region (e.g., reduce L, R, and/or W so that L3<2L1), then rerun the λ3 sweep over both sides of the threshold with all states Ericksen-admissible. If the measured error does not change behavior near λ3=1—or if no admissible state can be found with L3 comparable to L3—then Eq. (5) is not a falsifiable validity threshold in the physical domain.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV sets L1≈3.913 pN, L2≈1.565 pN and reports the GOS threshold L3≈9.276 pN (Eq. 5). The Ericksen inequalities in Eq. (4) require L3<2L1≈7.826 pN for this material. Hence every physically admissible L3 automatically satisfies Eq. (5); the threshold cannot be approached from above within the model's stability domain, and the numerical sweep includes λ3=1 (Appendix C, Table IV: rescaled L3≈5.99 vs 2L1≈5.06), i.e., a non-physical point violating the paper's own constraint K24<2K2. As a result, the openQmin data confirm only that small L3 gives small deviations; they cannot confirm that Eq. (5) marks the boundary of validity, because that boundary is never crossed by admissible states. The claimed quantitative criterion is therefore untested for the regime where it would make a difference.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper applies the Generalized Optimal Scaling (GOS) procedure, introduced in a previous paper by two of the authors, to derive explicit power-law thresholds for reducing the Landau–de Gennes elastic energy from (L1,L2,L3) to (L1,L2,0) and from (L1,L2,0) to (L1,0,0). The thresholds are Eqs. (5) and (7). The authors then use openQmin simulations of a spherical colloid in a periodic nematic cell to measure the relative L2 deviation between full and reduced minimizers, and report log-log scaling laws (6) and (8) that they interpret as confirming the GOS predictions. The paper concludes that the GOS thresholds provide a quantitative validity criterion for the one-constant approximation.","tokens_in":18746,"tokens_out":12522,"duration_ms":124312,"significance":"If the quantitative claim were fully established, the paper would give practitioners a much-needed, non-ad-hoc rule for when the one-constant approximation is safe in Landau–de Gennes simulations. The GOS algebra in Appendix B is explicit and reproducible, the input tables (Appendix C) are detailed, and the raw simulation data are deposited on Zenodo. The paper also connects a formal dimensional-analysis framework to concrete parameter regimes, which is potentially useful. However, as discussed below, the numerical validation does not currently support the central claim that the GOS threshold marks the boundary of validity, and there are internal inconsistencies in the stated physical parameters.","major_comments":[{"comment":"The GOS threshold for the (L1,L2,L3)->(L1,L2,0) reduction lies outside the Ericksen-admissible domain. With L1≈3.913 pN, Eq. (4) requires L3<2L1≈7.826 pN, but Eq. (5) gives L3≈9.276 pN. Thus no Ericksen-valid L3 can equal or exceed the threshold; the numerical sweep includes λ3=1 (Table IV, last row), whose rescaled L3≈5.993 exceeds 2L1≈5.056 and violates the paper's own stability constraint (K24<2K2). The simulations therefore probe only the regime where L3 is small relative to the threshold and never test whether the reduction fails as L3 approaches the predicted boundary. The data show that small L3 gives small deviations, but they cannot distinguish the GOS boundary from the Ericksen stability boundary. To validate Eq. (5), the authors need a parameter set for which the threshold lies inside the admissible interval, or a direct test of the error across the boundary.","section":"§IV, Eq. (5), Appendix C, Table IV"},{"comment":"The error scalings (6) and (8) are fitted to the same openQmin data that they are then used to confirm. The exponents 0.820 and 0.928 and the prefactors 0.174 and 0.925 are free parameters of a log-log least-squares regression, not outputs of the GOS derivation. GOS provides thresholds, not error-scaling exponents; Remark 1's claim that the error has 'the same order of magnitude as λ_p' is not a quantitative prediction of a specific exponent. Consequently, the numerical agreement is a consistency check with a power-law ansatz, not an independent validation of a derived scaling law. Moreover, the fits include λ3=1, which is nonphysical as noted above, so the reported regression may be biased.","section":"§III, Eqs. (6) and (8), Remark 1"},{"comment":"The second reduction case contains an internal inconsistency in the physical parameters. The text states that for the (L1,L2,0)->(L1,0,0) case, K1=K3=8 pN, K2=5 pN, and K24=5 pN. Using Eq. (11), this gives L2=α(K1-K24)=1.565×(8-5)=4.695 pN, not the 1.565 pN reported in Eq. (14). A consistent set yielding L2=1.565 pN and L3=0 would require K1=K3=6 pN. This propagates to Remark 6, where λ2≈0.060 is used, whereas the stated K values would give λ2≈4.695/25.897≈0.181. The input data in Table V are consistent with L2≈4.695 pN at the initial state (dimensionless L2≈3.033), not with Eq. (14). The authors should specify the exact OF constants used for the openQmin runs and reconcile them with Eq. (14), the threshold value L2≈25.897 pN, and the numerical examples.","section":"§IV, Eqs. (13)-(14), Remark 6, Appendix C.2"},{"comment":"The validation is sensitive to the numerical minimization tolerance. The authors state that using a FIRE force cutoff of 10^-10 instead of 10^-12 inflates the measured µ0 by one to two orders of magnitude and changes the scaling coefficients. The smallest λ values correspond to µ0 ~10^-7, close to typical convergence noise, so the reported power laws at the low-λ end may reflect numerical artifacts rather than physical model error. A convergence study over force tolerance, lattice size, and several independent random initial conditions is needed to establish that the observed scaling is robust. As written, the evidence rests on a single random configuration and a single geometry.","section":"§V, Remark 9"}],"minor_comments":[{"comment":"The definition of α is ambiguous: 'α=4/9S^2' should read α=4/(9S^2) to yield the stated numerical value ≈1.565.","section":"§IV, Eq. (10)"},{"comment":"In the row for target L3, the right-hand side contains L15_3 instead of L15_1; this is inconsistent with Eq. (5) and with the dimensional balance of the inequality.","section":"Appendix B, Table II"},{"comment":"The symbol L3 is used both for the elastic constant and for the threshold in Eqs. (5), (7), and Appendix C. This overloaded notation is confusing; consider using L_3^c or an overbar.","section":"Throughout"},{"comment":"The figures show no error bars or confidence bands for the fits. Given the tolerance sensitivity reported in Remark 9, the authors should indicate whether the data points are deterministic for the fixed seed and provide residual or confidence information for the regression.","section":"Figures 1 and 2"},{"comment":"The heading 'The (L1,L2,L3)->(L1,L2,0) case' appears twice; the second section should be titled 'The (L1,L2,0)->(L1,0,0) case'.","section":"Appendix C, headings"},{"comment":"'Semipositive definite' should be 'positive semidefinite'.","section":"Appendix B, Eqs. (B4)-(B5)"}],"recommendation":"major_revision","confidential_remarks":"The paper's central idea is sensible and the GOS derivation appears algebraically consistent, but the validation section does not yet support the stated quantitative claim. The most serious issue is that the L3 threshold cannot be approached within the Ericksen-stable parameter range, making the headline validation untested in the regime of interest. The internal parameter inconsistency for the L2 case (Section IV vs Appendix C) must also be resolved before the results are reproducible. Given that these are fixable with additional simulations or a more careful parameter choice, major revision is appropriate rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper's product — the explicit GOS thresholds in Eqs. (5) and (7) telling you when the three-constant LdG elastic energy can be reduced to two- or one-constant form — is new, dimensionally consistent, and worth having. Appendix B works through the GOS machinery carefully; Appendix D's invariance check is a nice catch; and they post the raw openQmin outputs on Zenodo, so the numerics are checkable. That part deserves credit.\n\nThe validation is where it goes soft, and the stress-test note holds up when I put the numbers in. For their parameters, L1 ≈ 3.91 pN, so the Ericksen condition caps L3 below 2L1 ≈ 7.83 pN. Their GOS threshold is L3* ≈ 9.28 pN — above the cap. So every stable material automatically satisfies Eq. (5), and the sweep never crosses the boundary it is meant to test. The λ3 = 1 row in Table IV is rescaled L3 ≈ 5.99 versus 2L1 ≈ 5.06, an inadmissible configuration by the paper's own K24 < 2K2 constraint. The data therefore show the trivial direction — smaller L3 gives smaller deviations — and the honest phrasing is that the numerics are consistent with the threshold, not that they confirm it. The L2 case is not subject to the same objection: Eq. (4) puts no upper bound on L2, so the criterion could bite, though the λ2 = 1 end needs K1 ≈ 21.5 pN, formally admissible but a strange material.\n\nThe reader's other complaint is fair: Eqs. (6) and (8) are power-law fits to the same data they are then said to confirm. GOS predicts the threshold; it does not predict the fitted exponents. The exponents are close to but not at 1, and Remark 9 concedes that at the smallest λ the measured µ0 is inflated by one to two orders of magnitude under a more realistic force tolerance — which is exactly the end of the curve that anchors the fits. Add in a single geometry, a single material, one random initial configuration (their Remark 8), and a GOS benchmark that lives in the authors' own 2025 preprint, and the phrase 'direct connection' in the abstract is more than the evidence supports.\n\nNone of this kills the paper. It is an openly written derivation with reproducible data and a practically useful rule: compute P/P*, and if it is well below 1, drop the constant. The right audience is anyone running LdG simulations who currently picks the one-constant model by habit.\n\nMy recommendation: send it to peer review. The thresholds deserve referee time. A competent referee will press on the Ericksen issue, on labeling (6) and (8) as fits rather than predictions, and on removing or flagging the non-physical λ3 = 1 point. Reframed or re-validated, this is a solid methods paper.","headline":"Useful new thresholds for reducing the LdG elastic energy, but the numerics never test the L3 boundary: the GOS value sits above the Ericksen cap, so the validation shows only that smaller is smaller.","tokens_in":19271,"tokens_out":12597,"would_cite":true,"duration_ms":122542,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["61.30.-v"],"model":"deepseek-v4-flash","headline":"This paper derives explicit thresholds on the Landau–de Gennes elastic constants below which dropping L3 or L2 leaves minimizers essentially unchanged, with error proportional to the dropped ratio.","keywords":["Landau-de Gennes","elastic constants","one-constant approximation","generalized optimal scaling","liquid crystals","Q-tensor","scale separation","numerical validation"],"falsifier":"Repeat the same comparison with the FIRE force cutoff set to 10^-10; if the measured relative deviation μ0 no longer tracks λ^0.82 or λ^0.93 at small λ but instead saturates at a noise floor, the numerical confirmation of the GOS scaling fails. Also, run the (L1,L2,0)→(L1,0,0) comparison at a different domain size (e.g., L=1 μm instead of 300 nm) or a different anchoring strength; the GOS threshold changes with these parameters, so the predicted μ0 should change accordingly.","tokens_in":18373,"feed_emoji":"🧪","tokens_out":9513,"duration_ms":82393,"temperature":0.7,"pith_summary":"This paper tries to turn a commonly used but ad-hoc simplification in liquid-crystal modeling — replacing the three-elastic-constant Landau–de Gennes energy (L1,L2,L3) with (L1,L2,0) or even the one-constant (L1,0,0) — into a quantitatively justified decision. Using the Generalized Optimal Scaling (GOS) procedure, the authors derive explicit thresholds: L3 must be much smaller than a 41st-root power-law combination of the other elastic constants, bulk coefficients, anchoring strength, and geometric lengths, and L2 must be much smaller than a 26th-root combination. Their numerical simulations of a spherical colloid in a nematic confirm that below these thresholds the relative deviation between full and reduced minimizers stays at the order of the ratio of the neglected constant to its threshold, with fitted power laws μ0 ≈ 0.174 λ3^0.820 and μ0 ≈ 0.925 λ2^0.928. If correct, this gives computational physicists a ready-to-use diagnostic for when simplified models are trustworthy, and it anchors the ubiquitous one-constant approximation in a parameter-dependent criterion rather than convention.","feed_headline":"Explicit cutoff says when dropping L3 or L2 is safe","feed_subtitle":"Below the computed threshold, simplified nematic minimizers deviate only by the order of the dropped ratio — confirmed in runs.","key_machinery":"The central object is the Generalized Optimal Scaling (GOS) procedure, an equation-free method that builds a complete set of dimensionless governing parameters from the physical parameters' units, then minimizes a cost function in log-space so that all dimensionless coefficients except the targeted one are driven to order unity. Its output is a power-law monomial threshold for the targeted parameter — for L3 a 41st root, for L2 a 26th root — and the paper's claim is that the ratio λ = P/P_crit controls the error: when λ ≪ 1, the minimizers of the full and reduced energies deviate only at order λ. The paper also proves (Appendix D) that these thresholds are invariant under the spatial and ene","core_discovery":"The claim, stated on the paper's own terms, is that both model reductions admit explicit boundaries: L3 << (L1^15 L2^15 a b c W^8 L^7 R^7)^(1/41) for (L1,L2,L3) → (L1,L2,0), and L2 << (L1^15 a b c L^7 R^7 W^8)^(1/26) for (L1,L2,0) → (L1,0,0). These thresholds are computed by GOS, which rescales space and energy so that every dimensionless coefficient except the targeted one is pushed toward unity, forcing the target constant to scale away. The numerical experiments bear out the scaling: the relative L2 deviation between minimizers is of the same order as λ = (neglected constant)/(critical threshold), decreasing to ~10^-7 when λ=10^-7, and the fitted exponents (0.82 and 0.93) are both near 1.","pith_inferences":["The two fitted exponents (0.82 and 0.93) sit close to, but not exactly at, 1; whether the deviation from unity is a finite-size effect, a consequence of the single geometry, or an intrinsic feature of GOS is not resolved by the paper, and testing a second geometry would distinguish these.","Because the thresholds depend on W, L, and R as powers, the criterion can be rephrased as a dimensionless comparison — for instance, against the anchoring extrapolation length K/W — which could connect GOS to familiar regimes in colloid science without additional computation.","The strong sensitivity of μ0 to the FIRE force tolerance (Remark 9) suggests that any practical use of this diagnostic should include a tolerance check; otherwise, numerical noise at looser tolerances will be mistaken for the physical scaling.","A natural next test is a multi-colloid or patterned-wall system, where the elastic energy landscape is qualitatively different; if the same thresholds hold there, the criterion would become a general tool for complex nematic geometries."],"forward_implications":["Computational physicists can pre-check whether a planned reduction is safe by evaluating one power-law monomial, and quantify the expected error before running expensive simulations.","The two reductions can be chained (L1,L2,L3)→(L1,L2,0)→(L1,0,0), with the total error bounded by roughly the sum of the two ratios; in the paper's example this justifies the one-constant approximation within 14%.","The same GOS machinery gives similar thresholds for other parameters (W, L, R, a, b, c) in Tables II and III, so the framework extends beyond elastic constants.","The fitted power laws (exponents near 1) mean that the deviation shrinks essentially linearly with the ratio λ, so a factor-of-ten smaller constant gives a factor-of-ten smaller error."],"fun_headline_variants":["Explicit cutoffs for dropping L2 or L3 in nematic energy","Numerically validated thresholds for simplified LdG models","When is it safe to reduce Landau-de Gennes? Now we know","Critical criteria for one- or two-constant nematic elasticity","Scaling-validated limits for complexity reduction in LdG"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The numerical verification assumes that the minimizer at force tolerance 10^-12 resolves Q-tensor differences as small as those produced by λ=10^-7, and that one spherical-colloid geometry with one initial random configuration represents the general parameter dependence of the GOS threshold — an assumption the paper itself flags in Remark 9, noting that looser tolerances inflate the measured error by one to two orders of magnitude.","fun_headline_variants_meta":{"raw":{"variants":["Explicit cutoffs for dropping L2 or L3 in nematic energy","Numerically validated thresholds for simplified LdG models","When is it safe to reduce Landau-de Gennes? Now we know","Critical criteria for one- or two-constant nematic elasticity","Scaling-validated limits for complexity reduction in LdG"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000196,"raw_usage":{"total_tokens":1225,"prompt_tokens":801,"completion_tokens":424,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":545,"completion_tokens_details":{"reasoning_tokens":333}},"tokens_in":545,"tokens_out":424,"duration_ms":5605,"temperature":1.0,"reasoning_tokens":333,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T11:17:49.725660+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the same comparison with the FIRE force cutoff set to 10^-10; if the measured relative deviation μ0 no longer tracks λ^0.82 or λ^0.93 at small λ but instead saturates at a noise floor, the numerical confirmation of the GOS scaling fails. Also, run the (L1,L2,0)→(L1,0,0) comparison at a different domain size (e.g., L=1 μm instead of 300 nm) or a different anchoring strength; the GOS threshold changes with these parameters, so the predicted μ0 should change accordingly.","supporting_citations":[],"review_version":1}