{"id":"45541911-2cfc-4485-adbc-4476646cec18","arxiv_id":"2505.05116","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"For linear elasticity in any dimension d ≥ 2, the paper proves Lipschitz stability of the density from Neumann-to-Dirichlet data and, under a monotonicity condition, of the triple (λ, μ, ρ).","lead":"This mathematics paper derives Lipschitz stability bounds for recovering the density and Lamé parameters of an elastic material from boundary displacement measurements. The bounds give worst-case error guarantees for elastography-style inverse problems when parameters lie in a known finite-dimensional family.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2's proof assumes every nonzero element of the finite-dimensional subspace has an open sign-definite region; this is false for general L∞ subspaces, so the main stability theorem is unproven as stated.","rationale":"The reader's weakest assumption identifies exactly the step I consider most load-bearing: the proof of Theorem 2 requires every normalized parameter difference to have an open set with a fixed sign and a positive lower bound. This is not a consequence of finite-dimensionality in L∞(Ω); dense sets with empty interior provide counterexamples. Since the localized-potentials argument cannot even be started without such an open set, the lower bound in Lemma 4 is unsupported, and the general Lipschitz stability theorem is unproven as stated. This affects the paper's broadest claim, not just a technical lemma, so it is the single most important concern. I also note, as an additional independent weakness, that Lemma 5(ii) applies the monotonicity inequality (21) to signed differences even though (21) is only stated and valid for nonnegative δ; on S\\S_j the difference is negative and the inequality direction reverses. The proof also introduces an undefined u_η. This second issue threatens Theorem 3, the constructive piecewise-constant result, but the sign-definite-region gap is more central because it undermines the general finite-dimensional theorem and the framework behind Theorem 5. The reader's CONDITIONAL verdict is appropriate: the results are plausible but the stated assumptions are insufficient for the proofs, and the paper needs corrected hypotheses and constants before the main claims can be accepted.","tokens_in":19263,"tokens_out":19891,"duration_ms":203646,"concrete_test":"Take Ω=(0,1)^d and let C⊂Ω be a fat Cantor set (positive measure, empty interior); set A=(Q∩Ω)∪C. Then A and Ω\\A are both dense, have empty interior, and have positive measure. Let E=span{1, χ_A−χ_{Ω\\A}} and choose ρ1=(a+b)/2+ε(χ_A−χ_{Ω\\A}), ρ2=(a+b)/2−ε(χ_A−χ_{Ω\\A}) with ε small enough that both lie in [a,b]. The normalized difference is proportional to χ_A−χ_{Ω\\A}, which has no open subset where it is bounded away from zero with a fixed sign. Verify that the proof of (18) in Theorem 2 cannot produce the open set D1 required by the localized-potentials bound; if no alternative argument establishes a positive lower bound for Lemma 4 on this subspace, Theorem 2 requires an additional sign-definite-region assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 2 in §3.1.3 asserts that for every ζ∈K (the unit sphere of the finite-dimensional subspace E⊂L∞(Ω)) there exists a non-empty open set D1 and β>0 with ζ|D1≥β or −ζ|D1≥β; this is the sole basis for (18) and hence for the positive lower bound in Lemma 4. The assertion is false for arbitrary L∞ subspaces. For example, let A⊂Ω be a measurable set such that both A and Ω\\A have positive measure and empty interior, and take E=span{1, χ_A−χ_{Ω\\A}} (the constant is included to allow admissible densities). The normalized difference of two admissible parameter values can be proportional to f=χ_A−χ_{Ω\\A}; f has no non-empty open subset on which it has a fixed sign and a positive lower bound. Consequently, the localized-potentials construction in Theorem 1 cannot be applied to the required D1, and the claimed uniform positivity of the infimum in Lemma 4 is not established. Thus Theorem 2 as stated, with only E⊂L∞(Ω), is unproven; a corrected statement needs an explicit sign-definite-open-region condition, or a restriction to continuous or piecewise-constant functions on a fixed open partition. The same gap appears in the proof of Theorem 5 for E+.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the stable recovery of the Lamé parameters (λ, μ) and density ρ in the isotropic linear elasticity system (1) from the Neumann-to-Dirichlet map. After proving a monotonicity relation between the material parameters and the boundary map, the authors invoke localized-potential arguments to derive: (i) a Lipschitz stability estimate for ρ when λ, μ are known and ρ lies in a finite-dimensional subspace of L∞(Ω) with two-sided bounds (Theorem 2); (ii) a constructive version for piecewise-constant densities supported in a fixed partition (Theorem 3); and (iii) a simultaneous Lipschitz stability estimate for (λ, μ, ρ) under a definiteness/monotonicity condition and a finite-dimensional parametrization (Theorem 5). The proofs are built from variational identities and unique-continuation-based localized potentials, in the spirit of recent monotonicity methods for inverse coefficient problems.","tokens_in":19575,"tokens_out":10964,"duration_ms":97480,"significance":"The paper targets a genuinely ill-posed multiparameter inverse problem, and a quantitative Lipschitz estimate of the form ‖ρ1−ρ2‖∞ ≤ C‖Λ(ρ1)−Λ(ρ2)‖∗, or the simultaneous version (39), would be a useful addition to the elasticity imaging literature. The monotonicity identities (5) and (27) are derived carefully from the variational form, and the constructive character of Theorem 3, with an explicit stability constant in (24), is attractive for numerical applications. However, the main theorems as stated are not established: the positivity of the infimum — the load-bearing step in both stability proofs — rests on a hidden sign-definiteness-open-region property that is false for general finite-dimensional L∞ subspaces, and the localized-potential lemmas carry unstated connectivity and regularity assumptions. These are correctness issues rather than presentation issues, and they affect the central claims. The paper is therefore not acceptable in its current form, but the gaps appear repairable by adding explicit structural hypotheses.","major_comments":[{"comment":"The proof of Theorem 2 asserts that for every ζ∈K there exists a non-empty open set D1⊂Ω and 0<β<1 such that either ζ|D1≥β or −ζ|D1≥β, and this assertion is the only mechanism producing the uniform lower bound in Lemma 4. The assertion is not implied by the standing assumption that E is a finite-dimensional subspace of L∞(Ω). For instance, if A⊂Ω is measurable with A and Ω\\A both of positive measure and empty interior and E=span{1, 1_A−1_{Ω\\A}}, then the normalized element f=1_A−1_{Ω\\A} has no non-empty open set on which it is sign-definite with a positive lower bound. Theorem 2 is therefore unproved as stated; it needs an explicit assumption such as piecewise constancy on a fixed open partition, continuity, or an abstract open-sign-definiteness condition on every element of the unit sphere of E.","section":"Section 3.1.3, proof of (18)"},{"comment":"Both localized-potential results rely on hidden geometric hypotheses. In Lemma 2, D is only assumed to be a subset of Ω of positive measure, yet the proof uses traces on ∂D and concludes v|Ω\\D=0 from unique continuation 'since Ω\\D is connected'; neither the connectivity nor the regularity of D that makes the trace statement meaningful is stated. In Lemma 8, the assertions that R(Tj) and R(Zj) are dense and intersect only trivially are justified in one sentence by unique continuation, with no stated condition on the geometry of D1, D2 or Ω\\(D1∪D2). Because Theorems 1 and 4 are direct consequences of these lemmas, the existence of the localized-potentials sequences used throughout the paper is not fully justified as written.","section":"Lemmas 2 and 8 (Sections 3.1.2 and 4.1)"},{"comment":"The proof of Theorem 5 repeats the same hidden sign-definiteness assumption in the multiparameter setting. For (ζ1,ζ2,ζ3)∈E+ it asserts the existence of a non-empty open D1 and δ>0 such that one component is bounded below by δ on D1 while the other two are nonnegative everywhere. This does not follow from ζi∈span(P), ζi≥0, and ‖(ζ1,ζ2,ζ3)‖Δ=1: a nonnegative L∞ function of unit norm may be supported only on a dense set with empty interior. Consequently the positivity of the infimum in Lemma 9, and with it the stability estimate (39), is not established without adding a structural hypothesis to P.","section":"Section 4.2.1, proof of (44)"},{"comment":"The constructive result also has missing justifications. Lemma 5(i) states that the boundary data g(j,k) exist by Theorem 1, but Theorem 1 requires D1,D2 open, disjoint, with Ω\\(D1∪D2) connected and meeting Γ_N; for D1=Sj and D2=S\\Sj these conditions are not verified from the stated assumptions on S and Sj. In part (ii), inequality (21) is stated only for δ∈L∞_+(S), but it is applied to a sign-changing δ (positive on Sj, negative on S\\Sj); the proof chain also refers to the undefined solution u_{η(j,k)}. In part (iii), the displayed convergence uses (5b/(3a)−3/2) while the target (20) has (5b/(2a)−3/2). These points affect the validity of the explicit stability constant α in (24).","section":"Section 3.1.4, Lemma 5 and Theorem 3"}],"minor_comments":[{"comment":"There are numerous typos: 'constrcutive' and 'simultameousely' in the abstract, 'dispslacement' and 'symetric' in Section 2, and inconsistent hyphenation throughout; a careful proofreading pass is needed.","section":"Abstract and Section 2"},{"comment":"Inequality (21) omits the squares inside the integrals: it should read ∫_S δ|u_ρ|² dx ≥ ∫_S δ|u_{ρ+δ}|² dx, and the restriction δ∈L∞_+(S) is unnecessarily strong; Lemma 1 gives the inequality for any δ with ρ+δ∈L∞_+(S).","section":"Lemma 5, inequality (21)"},{"comment":"In the sentence following (17), 'The second argument of the function h' refers to a function h that has not been defined; the intended object is J.","section":"Lemma 3"},{"comment":"In the line after (43), Φ is called with arguments (g, Θ1, Θ2, Θ3, (λ1,μ1,ρ1), (λ2,μ2,ρ2)), which does not match the definition of Φ in (40).","section":"Lemma 9 and proof of Theorem 5"},{"comment":"The sentence introducing the proof of Theorem 5 says 'monotonicity relations in Lemma 1' but the relevant statement is Lemma 6.","section":"Section 4.2"},{"comment":"References [33] and [37] are the same work (Eberle, Harrach, Meftahi, Rezgui, Lipschitz stability estimate and reconstruction of Lamé parameters in linear elasticity), listed with different publication years and page ranges; one duplicate should be removed.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the manuscript contains promising ideas and the monotonicity framework is standard, but the theorems as stated are not supported by the proofs. The missing sign-definiteness and connectivity hypotheses are likely fixable at the cost of narrowing the admissible classes; if the authors add those assumptions and repair Lemma 5, a resubmission would be worth considering. I do not see evidence of circularity or misconduct; the issue is incomplete mathematical support."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper extends the monotonicity/localized-potentials toolbox to linear elasticity and proves Lipschitz stability for the density (Theorems 2, 3) and for simultaneous (λ, μ, ρ) under a monotonicity condition (Theorem 5). The density result and the two-control localized-potentials construction (Theorem 4) are genuinely new relative to the earlier work by the same group. The writing is clear and the monotonicity lemmas check out. The citation pattern is fine; the relevant prior work is cited.\n\nThe soft spot sits in Theorem 2. The proof needs every normalized difference ζ in the finite-dimensional subspace E to have an open set where it is bounded away from zero with a fixed sign. That is not implied by E ⊂ L∞(Ω). Take Ω = (0,1), A ⊂ Ω with positive measure and empty interior, and E = span{1, χ_A − χ_{Ω\\A}}; the normalized difference proportional to χ_A − χ_{Ω\\A} has no open sign-definite region, so the localized-potentials argument cannot be started. The statement can be rescued by assuming continuous or piecewise-constant functions on a fixed partition, but as written Theorem 2 is unproven. The same issue appears in Theorem 5's positivity step, though there the piecewise-continuous assumption may be enough if the partition is fixed.\n\nTwo smaller issues. In Theorem 3, the displayed inequalities do not justify the stated constant α; the lower bound from Lemma 5 is I(j,k), not 1, so α should incorporate that. And the abstract overstates Theorem 5 by omitting the comparability condition (λ1 ≤ λ2 etc.). There are also missing hypotheses in Lemmas 2 and 8: connectivity of the complement of the localization set is used but not stated.\n\nBottom line: the central approach is sound, the gaps are specific and addressable, and the paper deserves a serious referee. With corrected assumptions and constants it would be a solid contribution. I'd send it to peer review.","headline":"New Lipschitz stability results for elasticity via monotonicity and localized potentials, but Theorem 2's proof assumes a sign-definiteness property that its L∞ assumption does not deliver.","tokens_in":20061,"tokens_out":6751,"would_cite":false,"duration_ms":63110,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35R30","74B05","35J47"],"pacs":[],"model":"deepseek-v4-flash","headline":"For parameters in a finite-dimensional class, the boundary force-to-displacement map determines the density—and, under an ordering assumption, the Lamé parameters—so that parameter error is bounded by a constant times the boundary-map…","keywords":["linear elasticity","inverse coefficients problem","Neumann-to-Dirichlet map","Lipschitz stability","monotonicity","localized potentials","Lamé parameters","density recovery"],"falsifier":"Let $\\Omega$ be the unit square and let $E$ be the one-dimensional subspace spanned by a bounded function $\\varphi$ that takes the value $+1$ on a dense set and $-1$ on its dense complement. For $\\zeta=\\varphi$, the proof of inequality (18) requires an open set $D_1$ with $|\\zeta|\\ge\\beta>0$, which does not exist; computing the infimum in Lemma 4 for this $E$—analytically or by high-resolution approximation—decides whether the positivity of the lower bound survives without the sign-definite-region condition.","tokens_in":19049,"feed_emoji":"🩺","tokens_out":10323,"duration_ms":95574,"temperature":0.7,"pith_summary":"This paper studies the inverse problem of recovering the density $\\rho$, or simultaneously the Lamé parameters $\\lambda,\\mu$ and density $\\rho$, of an elastic body from its Neumann-to-Dirichlet map—the boundary operator mapping an applied surface force to the resulting boundary displacement. The central claim is that, inside an a-priori known finite-dimensional class of parameters with fixed upper and lower bounds, the boundary map determines the parameters with Lipschitz stability: the parameter error is bounded by a constant times the operator-norm error of the measured boundary map. For the density-only problem with piecewise constant parameters, the authors give a constructive version whose stability constant $\\alpha$ is explicit and whose boundary loads are produced by finitely many well-posed forward solves. If the estimates are correct, small measurement noise in elastography and non-destructive testing leads to errors in recovered tissue stiffness and density that grow only linearly, which is the property needed for stable numerical reconstruction.","feed_headline":"Boundary measurements recover tissue density and stiffness linearly","feed_subtitle":"Lipschitz stability links small measurement noise to small parameter error for finite-dimensional elastic models.","key_machinery":"The load-bearing object is the Neumann-to-Dirichlet operator $\\Lambda_{\\lambda,\\mu,\\rho}$, which maps a boundary load $g$ to the boundary displacement $u|_{\\Gamma_N}$. The argument turns on two tools. First, the monotonicity relations of Lemma 1 and Lemma 6 bracket the quadratic-form difference of two boundary maps between integrals of the coefficient differences against the energies of the two corresponding solutions. Second, the localized-potentials theorems (Theorem 1 and Theorem 4) supply boundary loads whose solutions concentrate energy in a chosen open set $D_1$ while making it vanish on a disjoint set $D_2$. Combining the monotonicity bracket with localized potentials, and using compactness of the finite-dimensional parameter set, produces a uniform positive lower bound for the ratio of boundary-map difference to parameter difference—which is precisely Lipschitz stability.","core_discovery":"On the paper's own terms, the discovery is that the severe ill-posedness of this inverse elasticity problem can be tamed by imposing finite-dimensional structure. Theorem 2 states a nonconstructive Lipschitz stability estimate $\\|\\rho_1-\\rho_2\\|_{L^\\infty(\\Omega)} \\le C\\|\\Lambda(\\rho_1)-\\Lambda(\\rho_2)\\|_\\ast$ for densities in a finite-dimensional subspace with uniform positive bounds. Theorem 3 makes this constructive for piecewise constant densities: $\\|\\rho_1-\\rho_2\\|_\\infty \\le \\alpha\\|\\Lambda(\\rho_1)-\\Lambda(\\rho_2)\\|_\\ast$ with an explicit $\\alpha$ computed from finitely many boundary loads. Theorem 5 extends the Lipschitz bound to simultaneous recovery, $\\|(\\lambda_1-\\lambda_2,\\mu_1-\\mu_2,\\rho_1-\\rho_2)\\|_\\Delta \\le C\\|\\Lambda_{\\lambda_1,\\mu_1,\\rho_1}-\\Lambda_{\\lambda_2,\\mu_2,\\rho_2}\\|_\\ast$, provided the two parameter triples are ordered by coordinatewise inequalities. If true, these results convert the qualitative statement 'the problem is ill-posed' into a quantitative one: within the admissible class, boundary-measurement error controls parameter error linearly.","pith_inferences":["The proof of Theorem 2 requires, although the statement does not say so, that every normalized parameter difference be bounded away from zero with a fixed sign on some open subset of the domain; this suggests the finite-dimensional theorem is really guaranteed for piecewise-constant or otherwise locally sign-definite parameter families. This is our inference from the proof strategy, not an explici","A natural testable strengthening would be to quantify the size of the sign-definite regions—their measure or diameter—and to track how the stability constant depends on it, potentially giving dimension-dependent constants for adaptive partitions.","The monotonicity-plus-localized-potentials template is not specific to isotropic linear elasticity; it should produce analogous Lipschitz stability estimates for other elliptic systems with monotone coefficient dependence, such as anisotropic elasticity or poroelastic models, whenever a Runge approximation and unique continuation are available.","The coordinatewise ordering condition in Theorem 5 means that simultaneous recovery is proved stable only along monotone parameter paths; outside those paths, instabilities of the kind already known for general Schrödinger-type inverse problems may persist."],"forward_implications":["Equal Neumann-to-Dirichlet maps force equal parameters inside the admissible class: uniqueness follows directly from the Lipschitz estimates, as the paper notes in Remark 1 and as the ordered-pair formulation of Theorem 5 implies.","Stability estimates of this form transfer to numerical inversion: the error in a recovered density or Lamé pair is controlled by the error in the measured boundary map, so iterative reconstruction algorithms inherit convergence rates from the stability constant.","The constructive estimate for piecewise constant densities supplies an explicit constant and a recipe for boundary loads from finitely many well-posed PDE solves, making the stability bound usable rather than purely existential.","Because the bound is Lipschitz, finite-dimensional approximations of the Neumann-to-Dirichlet operator can be plugged into the estimate, so errors made in approximating the operator by a finite amount of data propagate linearly into parameter errors."],"supporting_citations":[{"why":"supplies the finite-dimensional monotonicity and localized-potentials template for Lipschitz stability that the density and multiparameter proofs adapt to elasticity.","marker":"[8]"},{"why":"gives the monopolar localized-potentials construction and global Lipschitz stability framework for a Robin transmission problem that the paper extends.","marker":"[9]"},{"why":"establishes uniqueness, Lipschitz stability, and reconstruction for the closely related scalar inverse optical tomography problem, the nearest scalar analogue.","marker":"[30]"},{"why":"documents exponential instability for general coefficients, which motivates the finite-dimensional and monotonicity restrictions needed for the stability estimates.","marker":"[31]"},{"why":"provides the quantitative strong unique continuation result for the Lamé system with low-regularity coefficients used to justify the Runge approximation and localized potentials.","marker":"[42]"},{"why":"provides the localized-potentials result and the operator-range corollary used in Theorem 4 to produce energy-concentrating boundary data.","marker":"[43]"}],"fun_headline_variants":["Lipschitz stability tames ill-posed inverse elasticity from boundary data","Finite-dimensional elastic inversion gains Lipschitz stability","Constructive Lipschitz bounds for density and Lamé inversion from Neumann data","Inverse elasticity gains Lipschitz stability under finite-dimensional assumptions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs every admissible normalized parameter difference to be bounded away from zero with a fixed sign on some open subset of the domain, and it needs the complement of the localization sets to be connected—conditions that are not implied by the stated finite-dimensional-subspace hypothesis.","fun_headline_variants_meta":{"raw":{"variants":["Lipschitz stability tames ill-posed inverse elasticity from boundary data","Finite-dimensional elastic inversion gains Lipschitz stability","Constructive Lipschitz bounds for density and Lamé inversion from Neumann data","Inverse elasticity gains Lipschitz stability under finite-dimensional assumptions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000876,"raw_usage":{"total_tokens":3824,"prompt_tokens":1014,"completion_tokens":2810,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":630,"completion_tokens_details":{"reasoning_tokens":2735}},"tokens_in":630,"tokens_out":2810,"duration_ms":21034,"temperature":1.0,"reasoning_tokens":2735,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:13:55.758587+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Let $\\Omega$ be the unit square and let $E$ be the one-dimensional subspace spanned by a bounded function $\\varphi$ that takes the value $+1$ on a dense set and $-1$ on its dense complement. For $\\zeta=\\varphi$, the proof of inequality (18) requires an open set $D_1$ with $|\\zeta|\\ge\\beta>0$, which does not exist; computing the infimum in Lemma 4 for this $E$—analytically or by high-resolution approximation—decides whether the positivity of the lower bound survives without the sign-definite-region condition.","supporting_citations":[{"cited_title":"Uniqueness and Lipschitz stability in electrical impedan ce tomography with finitely many electrodes","cited_arxiv_id":null,"evidence_quote":"supplies the finite-dimensional monotonicity and localized-potentials template for Lipschitz stability that the density and multiparameter proofs adapt to elasticity."},{"cited_title":"Global uniqueness and Lipschitz stability f or the inverse Robin transmission problem","cited_arxiv_id":null,"evidence_quote":"gives the monopolar localized-potentials construction and global Lipschitz stability framework for a Robin transmission problem that the paper extends."},{"cited_title":"Uniqueness, Lipschitz Stability, and Reconstructio n for the Inverse Optical Tomography Problem","cited_arxiv_id":null,"evidence_quote":"establishes uniqueness, Lipschitz stability, and reconstruction for the closely related scalar inverse optical tomography problem, the nearest scalar analogue."},{"cited_title":"Exponential instability in an inverse problem for the Schr¨ odinger equation","cited_arxiv_id":null,"evidence_quote":"documents exponential instability for general coefficients, which motivates the finite-dimensional and monotonicity restrictions needed for the stability estimates."},{"cited_title":"Quantitative strong un ique continuation for the Lam´ e system with less regular coefficients","cited_arxiv_id":null,"evidence_quote":"provides the quantitative strong unique continuation result for the Lamé system with low-regularity coefficients used to justify the Runge approximation and localized potentials."},{"cited_title":"Localized potentials in electrical impedance tomogra phy","cited_arxiv_id":null,"evidence_quote":"provides the localized-potentials result and the operator-range corollary used in Theorem 4 to produce energy-concentrating boundary data."}],"review_version":1}