{"id":"640d06b3-f3d0-4631-8cff-5fe20fffaf00","arxiv_id":"2412.20124","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"An explicit kernel K_H built from the Hamiltonian controls the Stieltjes transform of the eigenvalue counting function up to constants, yielding trace-class and sparse-spectrum criteria.","lead":"This paper proves that for a broad class of two-dimensional canonical systems, the density of eigenvalues is determined, up to universal constants, by an explicit integral kernel built from the Hamiltonian. It settles the long-open question of which such systems have trace-class resolvents and gives a way to construct systems with prescribed sparse spectra.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Dependence on black-box Theorem 2.15 is the only load-bearing concern; no internal flaw found.","rationale":"The reader identified Theorem 2.15 as the weakest assumption, and I agree: it is the single load-bearing external result. I scrutinized the internal proof chain for gaps—the reflected-Hamiltonian construction in Lemma 3.7, the zero-counting identity in Lemma 3.8, the limit-point approximation via truncations, and the trace-class criterion in Theorem 4.12—and found no internal inconsistency. The paper is a rigorous conditional result: given Theorem 2.15, the proofs of Theorems 3.2, 3.4, and 4.12 are complete. Since Theorem 2.15 is published, the default academic standard is to accept the citation; the appropriate action is therefore to keep the reader's ACCEPT verdict unchanged while noting that an independent audit of [23, Thm 1.1] would settle the residual risk. I did not identify a more concrete internal flaw, so recommending a verdict change would be unjustified.","tokens_in":56581,"tokens_out":24725,"duration_ms":249110,"concrete_test":"Re-derive [23, Theorem 1.1] from its proof, tracking every constant, and verify that for the family H^(t) from (3.6) the implicit constants are independent of H^(t), t, and the chosen compatible function. Concretely, take a simple limit-point Hamiltonian such as H(t)=diag(1,t^{-2}) on (0,1), form the reflected H^(t), and check algebraically that Im q_{H^(t)}(ir) is bounded above and below by positive multiples of 1/(r omega_{H^(t),2}(a, hat t^{(t)}(r))) with the same constants for all t in (0,1) and all large r. If such a family exhibits t-dependent ratios, Theorem 2.15's uniformity fails and Theorem 3.2 collapses; if the ratios are bounded, the load-bearing assumption is confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The main estimates (3.3) and (3.4) hinge on Theorem 2.15, imported verbatim from [23]. Lemma 3.7 applies it to the family of reflected Hamiltonians H^(t) defined in (3.6); t-uniformity of the implicit constants is essential because the available t-range is unbounded. If the constants in [23, Thm 1.1] depended on H (for example on det Omega_H(a,b) or on the particular compatible function), the pointwise relation (3.8) would fail and Theorems 3.4 and 3.2 would not follow. The paper does not reprove, quantify, or otherwise audit this theorem. I found no internal contradiction in the surrounding argument: Lemma 3.5 is an exact derivative identity, Lemma 3.8 is a standard zero-counting formula, the type-0 indivisible case in Lemma 3.7 is handled correctly, and the limit-point approximation in Section 3.3 is sound. Thus the central claim is correct exactly to the extent that Theorem 2.15 is correct.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies two-dimensional canonical systems y' = z J H(t) y with definite Hamiltonian H and discrete spectrum. Its first main result, Theorem 3.2, bounds the Stieltjes transform ∫_0^∞ (t+r^2)^{-1} n_H(√t)/t dt above and below by r^{-2} ∫_a^b K_H(t;r) dt with universal constants, where K_H is defined in (3.1) via a compatible pair. From this the paper derives criteria for convergence of ∑ 1/|λ|^p for 0<p<2 (Theorem 4.8), in particular a trace-class criterion (Theorem 4.12) together with the trace formula tr(A_H^{-1}) = -lim_{t→b} ∫_a^t h_3(s) ds. For limit-circle Hamiltonians, Theorem 5.3 gives an algorithmic estimate of ∫ K_H in terms of a partition function κ_H, up to a logarithmic error. Applications include monotonicity under deletions (Theorem 5.10), pointwise growth bounds for Hölder and bounded-variation angles (Proposition 5.13), exact growth for chirp and Weierstraß examples (Theorems 6.9 and 7.4), sharpness of the logarithmic upper bound (Proposition 7.5), and an inverse construction prescribing regularly varying growth of the monodromy (Theorem 6.13).","tokens_in":101,"tokens_out":14403,"duration_ms":273238,"significance":"If the results are correct, the paper settles a long-standing problem: for p ≤ 1, the earlier criteria from [32] fail and no general characterisation of ∑ |λ|^{-p} for canonical systems was known; Theorem 4.8 and its trace-class consequence fill this gap. The proof strategy is new: it combines the Weyl-coefficient estimates of [23] with a reflection construction of auxiliary Hamiltonians H^(t), exact identities for log |w_22|, and zero-counting via Hadamard products. A notable strength is that all asymptotic equivalences carry universal constants depending only on the compatible-pair constants c±, and the paper explicitly tracks this uniformity. The paper is also honest about limitations: it states in the introduction that actual eigenvalue asymptotics cannot be recovered because the constants C± cannot be made close, and Theorem 4.8 for ind g ∈ {0,2} is conditional on the existence of an auxiliary function g* satisfying (4.5)/(4.6), with examples in §4.2 showing large classes where such g* exist. The sharpness examples and the inverse theorem substantially increase the value of the paper.","major_comments":[],"minor_comments":[{"comment":"The passage from Theorem 2.15 to the estimate (3.8) uses the fact that the constants in Theorem 2.15 are independent of the Hamiltonian, and this uniformity is exactly what makes the reflection construction work. Since this is the only external input in the central proof, it would help the reader if §3.1 added a sentence explicitly recording that Theorem 2.15 is applied to the family H^(t) and that the definiteness and limit-point hypotheses are verified by the formulas (3.9)–(3.11).","section":"Section 3.1, Lemma 3.7"},{"comment":"The abstract and introduction describe the Schatten-class criterion as explicit, but for ind g = 0 and ind g = 2 the theorem is conditional on the existence of a function g* satisfying (4.5) or (4.6), a fact acknowledged in the text but not in the abstract. Consider adding a one-sentence caveat in the introduction or abstract to avoid overstatement.","section":"Theorem 4.8"},{"comment":"The trace formula gives a signed trace, which may be non-positive; for example, for a diagonal Hamiltonian the spectrum is symmetric and the trace is 0. A brief remark explaining the sign convention for tr(A_H^{-1}) would help readers, since the word 'trace' in the theorem statement might suggest a positive quantity.","section":"Theorem 4.12, Eq. (4.16)"}],"recommendation":"accept","confidential_remarks":"I agree with the reader's assessment. The reliance on [23, Theorem 1.1] is real but not problematic: it is a published, peer-reviewed result, and the uniformity required in Lemma 3.7 is part of its stated conclusion. The paper is within the scope of the journal and the presentation is careful and detailed. I recommend acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a serious paper and the right way to close the trace-class problem for canonical systems. The main theorem (3.2) gives a two-sided bound for the Stieltjes transform of the eigenvalue counting function in terms of a kernel K_H built from compatible pairs, and Theorem 4.12 turns that into the first p≤1 Schatten-class criterion for non-diagonal Hamiltonians. The reflected-Hamiltonian construction in Lemma 3.6 is original, and the way it converts the Weyl-coefficient estimate into a bound on log|w_22| is clean. The inverse theorem (6.13) is a real step beyond [22], and the examples—chirp, Weierstraß, the log-squared sharpness example—are worked out carefully and serve their purpose.\n\nThe one thing I want to flag is the dependence on Theorem 2.15, imported from [23]. The stress-test worry is that the t-uniformity of the constants might fail for the reflected Hamiltonians H^(t). I read the statement of 2.15: the constants depend on c± but not on H, r0, or t-hat. So the universality transfers. Importing a published, peer-reviewed theorem is normal practice; a short appendix restating the exact constant dependence would have been courteous, but this is not a load-bearing flaw.\n\nThe softer spot is Theorem 4.8 at the boundary indices ind g = 0 and 2: the characterization is conditional on the existence of an auxiliary function g* satisfying (4.5)/(4.6), and the authors themselves note it is open whether these always exist. They supply examples where g* can be chosen. That limits the \"explicit criterion\" claim at those boundary indices, but the trace-class case (g(r)=r, ind g=1) is unconditional.\n\nI did not find an internal contradiction. The limit-point approximation in Section 3.3 is sound, and the monotone-convergence passage is legitimate. The paper is long, but the structure pays off.\n\nThis is for people working on canonical systems, Krein strings, or Jacobi matrices who need resolvent ideal criteria for slow eigenvalue growth. It deserves a serious referee; the referee should double-check the applications of Theorem 2.15 in Lemma 3.7 and the Karamata passages, but the main argument is solid.\n\nRecommendation: send to peer review. Accept if the referee confirms the constants in [23] are as stated.","headline":"Solves the p≤1 resolvent ideal problem for general canonical systems with a genuinely new kernel method; the imported black-box theorem looks safe and the main argument holds up.","tokens_in":57328,"tokens_out":2300,"would_cite":true,"duration_ms":24550,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34L15","37J99","30D15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper establishes that the Stieltjes transform of the spectrum of a definite canonical system is fixed, up to universal constants, by an explicit integral of the Hamiltonian.","keywords":["canonical system","discrete spectrum","eigenvalue density","trace class","Schatten-von Neumann class","Stieltjes transform","regularly varying function","inverse spectral theorem"],"falsifier":"Take an explicit Hamiltonian satisfying the paper's assumptions, for instance the alternating rank-one-block Hamiltonian of Proposition 7.5, and numerically compute both sides of (3.3) along a sequence $r_n\\to\\infty$. If the ratio tends to $0$ or $\\infty$, the claimed universal comparison fails; equivalently, check whether finiteness of $\\int_1^\\infty r^{-2}\\int_a^b K_H(t;r)\\,dt\\,dr$ coincides with convergence of $\\sum |\\lambda|^{-1}$.","tokens_in":56382,"feed_emoji":"📐","tokens_out":14231,"duration_ms":133393,"temperature":0.7,"pith_summary":"The paper aims for a direct, quantitative bridge from the coefficient matrix $H$ of a two-dimensional canonical system to the clustering of its eigenvalues. It establishes that the Stieltjes transform of the eigenvalue counting function is bounded above and below, with constants depending only on universal parameters, by an explicit integral of $H$, whenever the spectrum is discrete and the $(1,1)$-entry of $H$ is integrable. From this comparison it derives a criterion for the resolvent of the model operator to lie in the Schatten--von Neumann class $S_p$ for every $p \\in (0,2)$, settling in particular the trace-class case $p=1$. In the limit-circle case it produces an algorithm whose output estimates the growth of the monodromy matrix up to a logarithmic factor, and it constructs explicit Hamiltonians whose monodromy grows like a prescribed regularly varying function of index in $(1/2,1)$. A reader should care because questions that were previously tractable only for diagonal systems, or only for $p>1$, now have explicit Hamiltonian-based answers for general systems, including sparse logarithmic spectra.","feed_headline":"One formula decides trace-class resolvents of canonical systems","feed_subtitle":"An explicit integral in the Hamiltonian controls eigenvalue density and settles the p=1 trace-class case.","key_machinery":"The load-bearing object is the kernel $K_H(t;r)$ of Definition 3.1, $K_H(t;r)=\\mathbf{1}_{[a,\\hat t(r))}(t)\\,\\frac{\\omega_2(a,t)h_1(t)}{c_+/r^2+\\omega_3(a,t)^2}+\\mathbf{1}_{[\\hat t(r),b)}(t)\\,\\frac{h_1(t)}{\\omega_1(\\hat s(t;r),t)}$, built from a compatible pair $(\\hat t,\\hat s)$: $\\hat t(r)$ is a point with $\\det\\Omega(a,\\hat t(r))$ comparable to $c/r^2$, and $\\hat s(t;r)$ is a point behind $t$ with the same property for $\\det\\Omega(\\hat s,t)$, where $\\Omega(s,t)=\\int_s^t H$. Theorem 3.4 is the mechanism: $\\log|w_{22}(x;ir)| \\asymp \\int_a^x K_H(t;r)\\,dt$. It is proved by writing the $t$-derivative of $\\log|w_{22}|$ as $r\\,\\mathrm{Im}(-w_{21}/w_{22})h_1$, recognising that ratio as the Weyl coefficient of a reflected Hamiltonian, and estimating it with the imported quantitative estimate of [23]. Lemma 3.8 converts the growth of $w_{22}$ into the Stieltjes transform of the counting function through a symmetrisation product, and a monotone-convergence limit passes to the limit-point case; all later statements are applications of this comparison.","core_discovery":"On the paper's own terms the central discovery is Theorem 3.2. For a definite Hamiltonian $H$ with discrete spectrum and $\\int_a^b h_1(t)\\,dt<\\infty$, choose a compatible pair $(\\hat t,\\hat s)$ marking where $\\det \\int H$ reaches $c/r^2$, and form the kernel $K_H(t;r)$ of Definition 3.1. Then $\\int_0^\\infty \\frac{1}{t+r^2}\\,\\frac{n_H(\\sqrt t)}{t}\\,dt \\asymp r^{-2}\\int_a^b K_H(t;r)\\,dt$ for $r>r_0$, with constants depending only on the chosen $c_\\pm$, and one side is finite if and only if the other is. The same comparison, after Tauberian steps, characterises convergence of $\\sum_{\\lambda\\in\\sigma_H\\setminus\\{0\\}} |\\lambda|^{-p}$ for every $p\\in(0,2)$; for $p=1$ it gives an explicit trace-class criterion and the trace formula $\\mathrm{tr}(A_H^{-1})=-\\lim_{t\\to b}\\int_a^t h_3(s)\\,ds$. In the limit-circle case, the algorithm of Definition 5.1 produces $\\kappa_H(r)$ with $\\int_a^b K_H(t;r)\\,dt$ between $\\kappa_H(r)\\log 2-O(\\log r)$ and $2e\\,\\kappa_H(r)(\\log r+O(1))$; examples show both bounds can be attained. The inverse construction supplies explicit Hamiltonians with prescribed regularly varying monodromy growth of index in $(1/2,1)$, and for some functions of index $1/2$.","pith_inferences":["If the comparison were proved with explicit constants, Theorem 4.8 would become a numerical spectral estimator for sparse spectra; the paper does not attempt to optimise constants, so this is an extrapolation.","A natural test of the logarithmic error in Theorem 5.3 is to run the partition algorithm on random Hamiltonians; the paper exhibits only extremal examples where the error is attained.","Because the trace formula is a signed limit of off-diagonal integrals, oscillating $h_3$ could make a trace-class resolvent have unexpectedly small or negative trace; this is a checkable consequence the paper leaves implicit."],"forward_implications":["For every $p\\in(0,2)$, convergence of $\\sum_{\\lambda\\neq0}|\\lambda|^{-p}$ is equivalent to $\\int_1^\\infty r^{-(p+1)}\\int_a^b K_H(t;r)\\,dt\\,dr<\\infty$; for $p>1$ this extends the endpoint criterion of [32], and for $p\\le1$ it covers general Hamiltonians with off-diagonal entries.","The resolvent $A_H^{-1}$ is trace class if and only if $\\int_1^\\infty r^{-2}\\int_a^b K_H(t;r)\\,dt\\,dr<\\infty$, and then its trace is $-\\lim_{t\\to b}\\int_a^t h_3(s)\\,ds$.","In the limit-circle case, the order of the monodromy matrix is $\\limsup_{n\\to\\infty}\\log\\kappa_H(r_n)/\\log r_n$ along any sequence $r_n\\to\\infty$ with bounded successive ratios.","Cutting a measurable piece out of the Hamiltonian cannot increase $\\kappa_H$, and the integral of $K_H$ grows by at most a logarithmic factor, so removing part of the domain never increases the order of the spectrum.","The estimates are sharp: a nowhere-differentiable H\\\"older rotation attains order $1/(1+\\nu)$ for every H\\\"older exponent $\\nu\\in(0,1)$, and an alternating rank-one Hamiltonian attains the upper $(\\log r)^2$ growth."],"supporting_citations":[{"why":"Supplies the quantitative Weyl-coefficient estimate imported as Theorem 2.15, the black box on which Theorem 3.4 and the main comparisons rest.","marker":"[23]"},{"why":"Establishes the discreteness criterion and the Schatten-class criterion for p>1 that this paper extends to p in (0,2).","marker":"[32]"},{"why":"Gives the diagonal-system criterion for p<=1 that is generalised to arbitrary Hamiltonians with off-diagonal entries.","marker":"[12]"},{"why":"Provides the earlier growth estimate and cutting-out comparison that the algorithmic theorems improve.","marker":"[22]"},{"why":"Supplies the Weyl-coefficient estimates and determinant identities used for compatible pairs and Lemma 3.7.","marker":"[18]"},{"why":"Supplies the Tauberian theorem for Stieltjes transforms used in Theorem 4.8.","marker":"[19]"},{"why":"Is the methodological source for the partition algorithm of Theorem 5.3.","marker":"[31]"},{"why":"Is the standard regular-variation reference behind the Tauberian and Abelian transitions.","marker":"[5]"}],"fun_headline_variants":["Trace-class resolvents: one integral decides","Sparse spectrum and trace class: new formula","Integral criterion for trace-class resolvents","Canonical systems: trace class solved","Eigenvalue density formula settles trace class"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole chain rests on an imported estimate, Theorem 2.15 of [23], asserting that the imaginary part of a certain solution ratio grows with constants depending only on universal parameters; if that estimate fails for any Hamiltonian satisfying the paper's stated assumptions, the comparison formulas and the trace-class criterion would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Trace-class resolvents: one integral decides","Sparse spectrum and trace class: new formula","Integral criterion for trace-class resolvents","Canonical systems: trace class solved","Eigenvalue density formula settles trace class"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000182,"raw_usage":{"total_tokens":1347,"prompt_tokens":1015,"completion_tokens":332,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":631,"completion_tokens_details":{"reasoning_tokens":265}},"tokens_in":631,"tokens_out":332,"duration_ms":4031,"temperature":1.0,"reasoning_tokens":265,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T23:31:37.120269+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take an explicit Hamiltonian satisfying the paper's assumptions, for instance the alternating rank-one-block Hamiltonian of Proposition 7.5, and numerically compute both sides of (3.3) along a sequence $r_n\\to\\infty$. If the ratio tends to $0$ or $\\infty$, the claimed universal comparison fails; equivalently, check whether finiteness of $\\int_1^\\infty r^{-2}\\int_a^b K_H(t;r)\\,dt\\,dr$ coincides with convergence of $\\sum |\\lambda|^{-1}$.","supporting_citations":[{"cited_title":"Romanov, Order problem for canonical systems and a conje cture of Valent, Trans","cited_arxiv_id":null,"evidence_quote":"Is the methodological source for the partition algorithm of Theorem 5.3."},{"cited_title":"Bingham, C.M","cited_arxiv_id":null,"evidence_quote":"Is the standard regular-variation reference behind the Tauberian and Abelian transitions."},{"cited_title":"Reiffenstein, A quantitative formula for the imaginary part of a Weyl coefficient, J","cited_arxiv_id":null,"evidence_quote":"Supplies the quantitative Weyl-coefficient estimate imported as Theorem 2.15, the black box on which Theorem 3.4 and the main comparisons rest."},{"cited_title":"Romanov and H","cited_arxiv_id":null,"evidence_quote":"Establishes the discreteness criterion and the Schatten-class criterion for p>1 that this paper extends to p in (0,2)."},{"cited_title":"Kac, Integral estimates for the distribution of the spect rum of a string [Russian], Sibirsk","cited_arxiv_id":null,"evidence_quote":"Gives the diagonal-system criterion for p<=1 that is generalised to arbitrary Hamiltonians with off-diagonal entries."},{"cited_title":"Pruckner and H","cited_arxiv_id":null,"evidence_quote":"Provides the earlier growth estimate and cutting-out comparison that the algorithmic theorems improve."},{"cited_title":"Langer, R","cited_arxiv_id":null,"evidence_quote":"Supplies the Weyl-coefficient estimates and determinant identities used for compatible pairs and Lemma 3.7."},{"cited_title":"Langer and H","cited_arxiv_id":null,"evidence_quote":"Supplies the Tauberian theorem for Stieltjes transforms used in Theorem 4.8."}],"review_version":1}