{"id":"55780cfa-7842-4eab-a247-6e2d58dd6fdc","arxiv_id":"2607.22120","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In dimensions d≥2, every genuine spectral point of a dissipative Schrödinger operator is detected by its bounded-domain truncations, so the barrier method cannot hide true spectrum.","lead":"This paper proves that for Schrödinger operators with absorbing (dissipative) barriers, truncating the domain to a finite box never loses genuine spectral points in dimensions two and higher. It settles a decade-old 'graveyard problem' in computational spectral theory, giving a rigorous guarantee for an approximation tool used to suppress spurious spectra.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Essential-spectrum no-invisibility rests entirely on Gil's 2024 reversed Hansmann inequality (Lemma 4.1), cited but not reproduced; it must be independently verified before the central theorem can be accepted.","rationale":"The paper's central theorem is proved by reducing spectral invisibility to eigenvalue counting. Lemma 3.1 provides the uniform Schatten control on H_R^{-1} − A_R^{-1}; Lemma 3.2 provides infinitely many self-adjoint eigenvalues near any essential-spectral point; Proposition 5.2 then derives a contradiction from the reversed Hansmann inequality. Thus every step except Lemma 4.1 is either standard or transparently proved in the paper. Lemma 4.1 is the linchpin: it converts uniform Schatten smallness into a lower bound on how many dissipative eigenvalues can disappear from a spectral window. The paper cites Gil [25] for this inequality, but does not state or prove the theorem; the proof of Lemma 4.1 simply appeals to (4.3) and a classical Weyl inequality. If Gil's theorem is false, has hidden hypotheses, or does not apply to compact B with algebraic-multiplicity enumeration, the essential-spectrum conclusion collapses. I do not see a competing internal flaw of comparable weight: the uniform estimates and the generalized strong resolvent convergence appear sound, and the numerical example is illustrative rather than load-bearing. The abstract overclaim about 'every spectral point' without the unique-continuation qualification is a genuine but secondary issue, already noted by the reader and partly mitigated by the body's conditional statement. The reader's weakest assumption identifies exactly the same point, so I agree with the assessment.","tokens_in":18530,"tokens_out":20123,"duration_ms":216944,"concrete_test":"Independently re-derive Lemma 4.1 from Gil [25, Thm 1.1], checking the exact hypotheses needed when B is compact self-adjoint and C − B ∈ S^r: confirm that the constant b_r depends only on r and that the eigenvalue enumeration is by algebraic multiplicity. Then numerically test the counting consequence (4.2) on finite-rank truncations of the actual operators: for a box or disk exhaustion, compute B_n = A_{R_n}^{-1}, C_n = H_{R_n}^{-1}, choose an interval J separated from σ(C_n) by η, and compare N_n η^r with (2 + 2b_r)^r ||B_n − C_n||_{S^r}^r for a few R and r just above p. A violation would locate the failure precisely.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The essential-spectrum proof in Proposition 5.2 reduces no-invisibility to an eigenvalue-counting contradiction: N_n η^r ≤ C ||B_n − C_n||_{S^r}^r, where B_n = A_{R_n}^{-1}, C_n = H_{R_n}^{-1}, and N_n = rank 1_J(B_n). Lemma 3.1 gives the uniform Schatten bound, Lemma 3.2 gives N_n → ∞, and the only mechanism that forces the eigenvalues of C_n to remain near those of B_n is Lemma 4.1, inherited verbatim from Gil [25, Thm 1.1] as a 'reverse Hansmann' inequality. The paper does not state Gil's theorem fully or reproduce its proof; the cited paper is recent, and every step of Proposition 5.2 after Lemma 3.1 depends on Lemma 4.1. If Gil's theorem has hidden hypotheses — for example, if B must be bounded rather than compact, if the eigenvalue enumeration uses geometric rather than algebraic multiplicity, or if the constant b_r is not independent of B — then the contradiction fails and the no-invisibility claim for the essential spectrum is unsupported. This is not an accusation against Gil's result; it is a missing verification at a load-bearing point. The rest of the argument is coherent, domain-uniform, and does not appear to have an internal gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that for Dirichlet Schrödinger operators H = -Δ + V + iS on connected open sets Ω ⊂ R^d (d ≥ 2), with S ≥ 0 in L^p at the Sobolev critical scale, every nested bounded Dirichlet exhaustion Ω_R ↗ Ω has no spectral invisibility: σ(H) ⊆ liminf_{R→∞} σ(H_R). The proof combines uniform-in-R Schatten bounds for H_R^{-1} - A_R^{-1}, generalized strong resolvent convergence of A_R to A, and a counting contradiction based on a reversed Hansmann–Weyl inequality. The discrete-spectrum case is handled via the Bögli–Marletta–Tretter framework under a unique continuation condition. A two-dimensional finite-element experiment illustrates recovery of the periodic band and lifting of a defect eigenvalue.","tokens_in":18782,"tokens_out":25106,"duration_ms":253026,"significance":"If correct, this settles the graveyard problem in all dimensions and gives the first general no-invisibility theorem for dissipative barrier truncations at the critical Sobolev scale. The proof is structural: it avoids diagonal resolvent-kernel estimates and depends only on Schatten-class and spectral-variation tools. The paper is refreshingly free of fitted parameters or ad hoc hypotheses, and the numerical example is reproducible with detailed convergence data. The main caveat is that the central inequality for the essential-spectrum case is imported from a recent paper of Gil' and is not verified in the manuscript, so the significance is conditional on that theorem.","major_comments":[{"comment":"The central reversed Hansmann inequality is entirely delegated to [25, Thm 1.1], which is neither stated nor proved in the manuscript. Proposition 5.2 reduces no-invisibility for the essential spectrum to exactly this inequality, so the reader cannot verify that the theorem applies to B_n = A_{R_n}^{-1} and C_n = H_{R_n}^{-1} (compact operators with infinitely many zero eigenvalues), or that the constant b_r is independent of B, C. Please state Gil's theorem with full hypotheses (multiplicity conventions, boundedness, permutation) and either include a proof or a detailed verification of the hypotheses. As written, the main essential-spectrum claim is conditional on this external result.","section":"§4, Lemma 4.1, Eq. (4.3)"},{"comment":"The proof that eigenvalues of H are non-real asserts that ∫S|u|^2 = 0 implies u = 0 by the unique continuation principle. The standard UCP [31,46] requires u to vanish on a nonempty open set. Under the assumptions S ≥ 0, S ∈ L^p, the set {S > 0} need not contain an open set (e.g., a fat Cantor set of positive measure). Thus the discrete-spectrum part of Theorem 1.1 is not proven for all S allowed by (1.4). Either add a hypothesis such as S > 0 a.e. or supp S containing an open set, or give a valid UCP argument from vanishing on a positive-measure set.","section":"§5, Proposition 5.1, step (1)"},{"comment":"This proof invokes [10, Thm 5.4], which requires generalized strong resolvent convergence of both H_R → H and H_R^* → H^*. Lemma 3.2 only establishes H_R → H. The proof of Lemma 3.2 extends verbatim with S replaced by -S, but the adjoint convergence must be stated explicitly because it determines the limiting essential set in the hypothesis of [10, Thm 5.4].","section":"§5, Proposition 5.1"}],"minor_comments":[{"comment":"The text refers to 'Theorem 2.3' and 'Theorem 2.4'; these should be Lemmas 2.3 and 2.4.","section":"§2, proof of Lemma 2.4"},{"comment":"The inequality for imaginary parts is not the classical Weyl eigenvalue inequality; please state the precise theorem from [26] being used.","section":"§4, Eq. (4.4)"},{"comment":"The bound on the eigenvalue counting function for -Δ_{Ω_R} is cited as Weyl's law; for arbitrary bounded open sets the classical asymptotic may require boundary regularity. The needed upper bound is due to Berezin/Lieb/Rozenblum and should be cited.","section":"§6.2"},{"comment":"The narrative about ChatGPT-based agents is out of place in a mathematical paper; I suggest moving it to the acknowledgments or removing it.","section":"§1.3"}],"recommendation":"major_revision","confidential_remarks":"The paper is strong and likely correct if Gil's theorem is valid. The main risk is the unverified external inequality; I recommend asking the authors to include a full statement and proof or a precise verification. The unique continuation issue is local and fixable. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe one thing to know: this is the first proof of the d ≥ 2 no-invisibility theorem for dissipative barrier truncations — the higher-dimensional graveyard problem. For H = −Δ + V + iS with S ≥ 0 at the natural Sobolev scale (p ≥ d/2 for d ≥ 3, p > 1 for d = 2), every nested Dirichlet exhaustion satisfies the lower spectral inclusion for the essential spectrum, with no regularity on ∂Ω. That part is unconditional. The discrete-eigenvalue statement carries a unique-continuation qualifier.\n\nI read the proof carefully and it holds together. The mechanism is genuinely new: uniform-in-R Cwikel-type Schatten bounds on H_R^{-1} − A_R^{-1}, generalized strong resolvent convergence, then an eigenvalue-counting contradiction — the self-adjoint sections A_R have unboundedly many eigenvalues near each point of σ_ess(A), the dissipative perturbation is uniformly small in S^r, and a reverse Hansmann inequality forces enough eigenvalues to survive. I checked the domain-monotonicity arguments that make the constants R-independent; they're fine. The numerics are an honest illustration, not evidence.\n\nSoft spots, in proportion. The load-bearing one: Lemma 4.1, and through it the entire essential-spectrum argument, rests on two external inequalities — Gil's 2024 reverse Hansmann theorem [25] and a Weyl-type bound on imaginary parts of eigenvalues from Gohberg–Krein. Both are cited, neither reproduced. The stress-test is correct: if Gil's theorem has hidden hypotheses (the constant's dependence, the multiplicity convention, applicability to compact operators), the contradiction in Proposition 5.2 fails. This is a verification gap, not evidence of a wrong proof — the paper states the lemma precisely enough that a referee can check it against [25, 24] directly — but \"can check\" is not \"has checked.\" Acceptance should be conditional on that check. A smaller issue: the abstract and the conclusion claim every spectral point without qualification, while Theorem 1.1 itself correctly requires unique continuation for isolated eigenvalues. Fix the abstract, not the theorem. The AI-search disclosure in §1.3 is honest and, if anything, makes the dependence on [25, 24] more explicit; it changes nothing mathematically.\n\nWho this is for: anyone working on spectral approximation of non-self-adjoint operators, PML-type methods, or spectral pollution. The result settles a decade-old open case and gives users of dissipative barriers a precise guarantee. Send it to a careful referee. If Gil's theorem checks out, accept.","headline":"First proof of the d≥2 no-invisibility theorem for dissipative barrier truncations; the argument is coherent and new, but the essential-spectrum case rides on one cited-and-unreproduced inequality that a referee must verify.","tokens_in":19315,"tokens_out":20599,"would_cite":true,"duration_ms":197943,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35J10","47A10","47B44","46N40","47-08","65J10","65N25"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the dissipative barrier method cannot cause spectral invisibility for Schrödinger operators in any dimension.","keywords":["spectral invisibility","graveyard problem","dissipative barrier","domain truncation","Schatten class","spectral pollution","Schrödinger operator","essential spectrum"],"falsifier":"Exhibit a compact self-adjoint operator B and a compact operator C with C-B in S^r for some r > 1, and a Borel set J separated from σ(C) ∪ {0}, such that rank 1_J(B) exceeds (2+2b_r)^r η^{-r} ||C-B||_{S_r}^r. That would disprove the key eigenvalue-counting lemma and, with it, the proof of Theorem 1.1.","tokens_in":18349,"feed_emoji":"🎯","tokens_out":6436,"duration_ms":59080,"temperature":0.7,"pith_summary":"The paper settles the 'graveyard problem' in computational spectral theory: the dissipative barrier method, which adds an imaginary absorbing potential to suppress spurious eigenvalues, cannot hide genuine spectral points when the computational domain is truncated. For any Dirichlet Schrödinger operator H = -Δ + V + iS in dimension d ≥ 2, with S at the critical Sobolev scale, every nested family of bounded domains gives spectra that accumulate on all of σ(H). This means every true spectral point — essential or discrete — is detected by the finite-domain approximations, so the cure for spectral pollution does not create the complementary disease of spectral invisibility. The proof is an eigenvalue-counting argument using Schatten-class bounds and a reversed spectral-variation inequality.","feed_headline":"No spectral invisibility in any dimension","feed_subtitle":"Domain truncation with an absorbing potential still reveals the full spectrum of the non-self-adjoint operator.","key_machinery":"The load-bearing identity is the resolvent formula H_R^{-1} - A_R^{-1} = A_R^{-1/2}(I + i W_R)^{-1}(-i W_R) A_R^{-1/2}, where W_R is the non-negative compact Birman–Schwinger operator A_R^{-1/2} S A_R^{-1/2}. Its Schatten norm is uniformly bounded in R by Cwikel-type estimates, so the perturbation is uniformly trace-class-like. The other essential ingredient is the reversed spectral-variation inequality of Hansmann–Weyl type: for compact self-adjoint B and compact C with C-B in S^r, the eigenvalues of C cannot all be driven away from those of B; the counting consequence is that if J is a Borel set separated from σ(C) ∪ {0}, then rank 1_J(B) ≤ (const / η^r) ||C-B||_{S_r}^r. This turns the Sch","core_discovery":"The central claim is Theorem 1.1: for every λ in the essential spectrum of H and every open neighborhood U of λ, the spectrum of the truncated operator H_R meets U for all sufficiently large R, and the same holds for eigenvalues when H has the unique continuation property. Equivalently, σ(H) ⊆ liminf σ(H_R). This is proved by showing that near each point of σ_ess(A) the self-adjoint truncations A_R have arbitrarily many eigenvalues, while the resolvent difference H_R^{-1} - A_R^{-1} has Schatten norm uniformly bounded in R; a reversed Hansmann-type inequality then forces any uniformly small non-self-adjoint perturbation to leave some eigenvalues in every neighborhood. The result holds withou","pith_inferences":["If the reversed spectral-variation inequality holds with explicit constants, the proof could yield quantitative rates at which spectral points are detected, e.g., how large R must be for a given accuracy.","The result suggests that spectral invisibility is not a generic failure of domain-truncation schemes for non-self-adjoint operators; it may be avoidable whenever the perturbation is compact in a suitable Schatten sense and the self-adjoint truncations have growing eigenvalue multiplicities.","The method could be adapted to other non-self-adjoint perturbations beyond the dissipative Schrödinger case, such as complex potentials that are not purely imaginary, as long as the resolvent difference falls in a Schatten class.","Experimental tests on the two-dimensional example show convergence of the discrete spectra; a natural extension would be to verify the theorem numerically for a case with essential spectrum that has a gap, checking that the band edges are detected."],"forward_implications":["Every genuine spectral point of H is detected by the finite-domain approximations, so numerical methods that use dissipative barriers and then discretize cannot fail solely because of the truncation step.","The theorem holds at the critical Sobolev scale S ∈ L^{d/2} for d ≥ 3 and any S ∈ L^p, p > 1, for d = 2, matching the natural compactness threshold.","No boundary regularity of Ω is needed, so the result covers irregular computational domains.","Together with the earlier one-dimensional result, the no-invisibility question is now settled in all dimensions.","The same mechanism applies to higher-order elliptic operators under analogous integrability conditions (Remark 2.6)."],"fun_headline_variants":["Dissipative barriers can't hide spectrum in any dimension","Graveyard problem solved: no spectral invisibility","No spectral hiding by dissipative truncations","Spectral invisibility ruled out in all dimensions","Dissipative barriers reveal full spectrum in any dimension"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof relies on the reversed spectral-variation inequality of Hansmann–Weyl type: that a compact non-self-adjoint perturbation with finite Schatten norm cannot move all eigenvalues of a compact self-adjoint operator away from any Borel set. If that inequality is false or its hypotheses do not apply to the resolvent pairs in this setting, the central no-invisibility claim for the essential spectrum would fail.","fun_headline_variants_meta":{"raw":{"variants":["Dissipative barriers can't hide spectrum in any dimension","Graveyard problem solved: no spectral invisibility","No spectral hiding by dissipative truncations","Spectral invisibility ruled out in all dimensions","Dissipative barriers reveal full spectrum in any dimension"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000779,"raw_usage":{"total_tokens":3375,"prompt_tokens":937,"completion_tokens":2438,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":681,"completion_tokens_details":{"reasoning_tokens":2365}},"tokens_in":681,"tokens_out":2438,"duration_ms":15207,"temperature":1.0,"reasoning_tokens":2365,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T05:44:21.813636+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a compact self-adjoint operator B and a compact operator C with C-B in S^r for some r > 1, and a Borel set J separated from σ(C) ∪ {0}, such that rank 1_J(B) exceeds (2+2b_r)^r η^{-r} ||C-B||_{S_r}^r. That would disprove the key eigenvalue-counting lemma and, with it, the proof of Theorem 1.1.","supporting_citations":[],"review_version":1}