{"id":"f1343fc2-af1f-41d5-b665-22c3bbee5284","arxiv_id":"2411.15830","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Microscopic deformations of biorthogonal ensembles converge to the marking-and-conditioning deformation of the universal limit process, yielding new probabilistic interpretations of Painlevé kernels.","lead":"This paper proves a transfer theorem for random particle systems with algebraic structure: if an unmodified system converges to a universal limit, then any microscopic modification converges to the correspondingly modified limit. The result gives probabilistic interpretations of several Painlevé-type kernels previously known only through complicated formulas.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The transfer theorem hinges on the factorized L2 domination bound in Assumption 2.2(3)/2.6(3), which is verified only through imported full-line kernel estimates; if that bound fails, the central convergence claim is not established.","rationale":"The reader's verdict CONDITIONAL is appropriate. The factorized L2 domination bound in Assumption 2.2(3)/2.6(3) is indeed the most load-bearing hypothesis: it is precisely the hypothesis used to dominate the Fredholm determinant series, and its verification in all applications rests on imported full-line kernel bounds that are not re-proven here. I found no internal inconsistency in the proof given the assumption; the proof for h∈[0,1) is coherent and the dominated-convergence argument is valid. The overclaim in the theorem statements for h≥0 is a real but secondary issue, since weak convergence only requires h<1. The proposed test of the full-line diagonal bound would settle whether the assumption is actually available for the stated class of weights. Hence the correct disposition is to keep CONDITIONAL, pending either a proof or a citation for the full-line bounds and a correction of the h-range in Theorems 2.3 and 2.7.","tokens_in":32623,"tokens_out":25337,"duration_ms":204400,"concrete_test":"Independently verify the full-line uniform diagonal bound underlying Assumption 2.2(3): for a weight V satisfying (3.1), choose a bulk point x* with κ_V(x*)>0 and prove or numerically compute sup_{u∈R, n≥n0} (1/(κ_V(x*)n)) k_n(x*+u/(κ_V(x*)n), x*+u/(κ_V(x*)n)). If this supremum is infinite for some V in the stated class, then the factorized domination bound in §3.1 fails for deformations σ_n with unbounded support, and Corollary 3.1 is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central proof (Section 4, Lemmas 4.4–4.5; Section 5, Lemmas 5.2–5.4) uses dominated convergence on the Fredholm determinant series, and the domination bound is exactly Assumption 2.2(3): for each bounded F, sqrt(1_F+σ_n)|K_n|sqrt(1_F+σ_n) ≤ Φ(u)Ψ(v) with Φ,Ψ∈L2(µ). This factorized L2 form is what makes the Hadamard bound yield |S_{n,k}| ≤ k^{k/2}(∫ΦΨ)^k, giving uniform-in-n control for the series. In the applications, this assumption is imported from known full-line one-point bounds: (3.5) extended from compacts via restricted range inequalities [76], (3.14) from Riemann-Hilbert asymptotics, and [4, Lemma 7.13] for the discrete case. These estimates are not re-proven here, and the factors 1_F+σ_n require control of the kernel on the whole of K, not just on compacts, because σ_n and the determinant variables range over all of K. If the full-line bound fails while a mere compact version holds, Assumption 2.2(3) is not satisfied and Lemmas 4.4–4.5 do not apply; the conclusion of Theorem 2.3 is then unproven. A secondary statement/proof mismatch (theorems state h≥0; proofs only handle 0≤h<1) does not affect the weak-convergence conclusion but should be corrected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a transfer theorem for deformed biorthogonal ensembles. Under pointwise kernel convergence plus a factorized L² domination bound (Assumption 2.2), Theorem 2.3 shows that the probability generating functionals of the deformed ensembles converge to those of the σ-deformation of the limiting point process; Theorem 2.7 gives a varying-measure analogue (Assumption 2.6). The proof expands the relevant Fredholm determinants and applies dominated convergence, after establishing a finite-rank identity for the deformed kernel in Lemma 4.1. Applications are given to bulk and soft-edge deformations of orthogonal polynomial ensembles and to bulk deformations of discrete Coulomb gases, yielding deformed sine, Airy, and discrete sine processes.","tokens_in":32959,"tokens_out":32329,"duration_ms":305306,"significance":"If the results are correct, the paper offers a conceptually clean and quite general mechanism for universality transfer: deformations by marking and conditioning commute with scaling limits under mild assumptions, without a separate asymptotic analysis of the deformed models. The use of probability generating functionals rather than correlation kernels is a genuine methodological novelty, and the resulting probabilistic interpretation of several Painlevé-type kernels is valuable. The finite-n deformed kernel identity and the dominated-convergence argument for the continuous cases are coherent and carefully written. The discrete application, however, contains a normalization issue that needs correction before the corollary as stated can be trusted.","major_comments":[{"comment":"The normalization in the discrete Coulomb gas application appears inconsistent. Since ∫ k_n(x,x)dν_N(x)=n and dν_N is the counting measure on N sites, the per-site one-point function cannot converge to κV(x) if κV is the Lebesgue density of a probability measure on [0,1] as in (3.24); with n=βN the correct per-site limit should involve a factor β/ρ, namely βκV(x)/ρ(x). Moreover, with the scaling (3.26), the lattice spacing of the limiting set Ω is δ=κV(x*)/ρ(x*), so a discrete sine process with intensity β per unit length has kernel δ sin(πβ(u−v))/(π(u−v)) in continuous coordinates, equivalently site-index kernel sin(πβδ(i−j))/(π(i−j)), not (βκV(x*)/ρ(x*)) sin(π(u−v))/(π(u−v)) as printed. As written, Kdsin in Corollary 3.5 has the wrong prefactor and oscillation frequency unless β=1. The authors should correct (3.25) and the Kdsin formula, and reconcile them with the quoted [4, Lemma 7.13]; otherwise the discrete corollary is not established.","section":"Section 3.3, Eqs. (3.25)–(3.26) and Corollary 3.5"}],"minor_comments":[{"comment":"The theorem statements allow continuous h: K → [0,+∞) with bounded support, but the proofs in Sections 4 and 5 explicitly restrict to 0 ≤ h < 1 with sup h < 1. Since weak convergence is characterized by the class h ∈ [0,1], the statements should either be restricted to this class or an analytic-continuation argument in h should be supplied.","section":"Theorems 2.3 and 2.7, Sections 4 and 5"},{"comment":"The verification of Assumption 2.6(3) sets Φ=Ψ=2C1_{[-L,L]}, which is not continuous on all of R; the argument should explicitly invoke the relaxed continuity condition stated after Theorem 2.7, namely continuity on a neighborhood of supp μ.","section":"Section 3.3, verification of Assumption 2.6(3)"},{"comment":"The proof uses without comment that the measure of ∂G under ⊗(ΦΨ dμ) is zero; this follows because G is defined by ΦΨ≠0 and the measure has density ΦΨ, but a one-sentence justification would avoid a gap for the reader.","section":"Lemma 5.2"},{"comment":"The weak convergence of the product measures ⊗(ΦΨ dμ_n) to ⊗(ΦΨ dμ) is used without proof; it follows from the uniform boundedness of the total masses guaranteed by Assumption 2.6(3), but this step should be stated explicitly.","section":"Lemmas 5.2 and 5.4"},{"comment":"Corollaries 3.1, 3.2, and 3.5 inherit imported full-line or compact one-point bounds, namely (3.5), (3.14), and [4, Lemma 7.13]; the paper should state explicitly that these estimates are not re-proved here and that the conclusions depend on them.","section":"Section 3, applications"}],"recommendation":"major_revision","confidential_remarks":"The central transfer theorem appears sound and well-presented, and the continuous applications are convincing. My main concern is the discrete application: the normalization of the one-point function and of the discrete sine kernel in Section 3.3 seems incorrect as written, and the authors should be asked to reconcile (3.25), the scaling in (3.26), and the quoted [4, Lemma 7.13]. In addition, the paper depends on [24, Theorem 2.4(2)] for the identification of the limiting deformed process; if [24] is not yet accepted, this dependence should be made explicit for the editor and the reader."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here is my read. This is a solid transfer paper. It proves that if an undeformed biorthogonal ensemble converges weakly to a limit point process, then the weight-deformed version converges to the corresponding deformed limit, under Assumptions 2.2/2.6. The PGF-based approach is genuinely different from the usual kernel-scaling analysis, and it pays off: the hypotheses are mild and the applications are real, including the probabilistic interpretation of several Painlevé kernels as conditional/thinned sine or Airy processes.\n\nThe proof is careful. Lemmas 4.1–4.2 re-derive the finite-n deformed kernel and PGF ratio from scratch; the dominated-convergence lemmas are clean. I found no gap in the central weak-convergence claim.\n\nSoft spots, in proportion. (1) The theorem statements say h ≥ 0, but the proofs (start of Section 4, and Section 5) restrict to continuous h with 0 ≤ h < 1 and sup h < 1. That is the standard class for PGF convergence, so the conclusion is fine, but the statement should be corrected or the extension supplied. (2) The factorized L2 domination bound, Assumption 2.2(3)/2.6(3), is the main input, and in the applications it is imported from known full-line one-point estimates: (3.5) via restricted range inequalities, (3.14) from RH asymptotics, and [4, Lemma 7.13] for the discrete gas. These are not re-proven. I would want a referee to verify those citations support exactly the full-line form used. The stress-test worry about full-line vs compact does not hit the discrete application, since σ is compactly supported and the factor is localized; for the bulk and edge cases the full-line bounds are explicitly stated and look standard. (3) The identity G^σ[h] = det(1 − √h K^σ √h) at the end of Theorem 2.3's proof, and the analogous one in Theorem 2.7, are imported from [24, Theorem 2.4(2)]. Not circular, and [24] is the right prior reference, but it is a load-bearing import.\n\nThe audience is the random-matrix/integrable-probability community; anyone working on conditional or thinned DPPs will get direct value. This paper deserves a serious referee. I would send it to a good probability or math-physics journal, and in revision ask for the h-range mismatch to be fixed and for the imported bounds to be stated precisely with the exact citations checked.","headline":"A solid, genuinely useful transfer principle for deformed biorthogonal ensembles; the main fix before publication is the h-range statement, and the imported full-line bounds should be verified.","tokens_in":33472,"tokens_out":5639,"would_cite":true,"duration_ms":50757,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60G55","60B20","42C05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that when a biorthogonal ensemble converges to a universal limit, any microscopic deformation of it converges to the corresponding deformation of that same limit, with the deformed limit built by marking and conditioning.","keywords":["biorthogonal ensembles","determinantal point processes","probability generating functionals","universality","marking and conditioning","Painlevé kernels","discrete Coulomb gases","sine kernel and Airy kernel"],"falsifier":"Search for a sequence of biorthogonal ensembles whose kernels converge pointwise to a limit but whose diagonal grows too fast for the factorized $L^{2}$ bound to hold, for instance an orthogonal polynomial ensemble at a point where the equilibrium density vanishes. For a fixed deformation $\\sigma$, compute the deformed probability generating functional numerically for increasing n; if the limit differs from the Fredholm determinant of the deformed kernel K^$\\sigma$, the transfer principle fails exactly where the domination assumption is essential.","tokens_in":32417,"feed_emoji":"🎲","tokens_out":9593,"duration_ms":85009,"temperature":0.7,"pith_summary":"This paper establishes a transfer principle for universality in a large family of random particle systems known as biorthogonal ensembles. It proves that if a biorthogonal ensemble converges, as the number of particles grows, to a limiting point process such as the sine or Airy process, then a microscopic deformation of the ensemble converges to the corresponding deformation of that same limiting process. The deformed limit is built by marking each particle with an independent Bernoulli label and conditioning on the event that no label is observed. The proof works through probability generating functionals rather than correlation kernels, so it needs only mild regularity and a factorized square-integrable bound on the scaled kernels. As a consequence, several Painlevé-type kernels studied in the literature are identified as conditional and thinned versions of sine and Airy point processes.","feed_headline":"Marking and conditioning carries a limit law to deformed ensembles","feed_subtitle":"Random particle ensembles that converge keep converging after microscopic weight changes, only the universal limit law changes.","key_machinery":"The machinery is the probability generating functional (PGF) of a determinantal point process, G[h] = det(1 - $\\sqrt$(h) K $\\sqrt$(h)). For a deformed ensemble the PGF has the ratio form $G_n^{{sigma_n}}$[h] = det(1 - (sigma_n + h - sigma_n h)K_n) / det(1 - sigma_n K_n), and the proof shows that both determinants in the ratio converge by dominated convergence of their Fredholm series. The domination comes from Hadamard's inequality plus a factorized $L^{2}$ bound on the kernels, the product Phi(u)Psi(v) that is Assumption 2.2(3). The deformed limiting kernel K^$\\sigma$ = $\\sqrt$(1-$\\sigma$) K (1-$\\sigma$ K)^{-1} $\\sqrt$(1-$\\sigma$), supplied by the marking-and-conditioning construction, is the object that carries the limit through.","core_discovery":"The central claim is that the deformation operation commutes with the large-n limit. Given a sequence of biorthogonal ensembles whose scaled kernels converge pointwise to a kernel K, and deformations sigma_n that converge to a function $\\sigma$, the deformed ensembles converge weakly to the deformed limit point process X^$\\sigma$, the process obtained by taking the ground process X, assigning each point an independent Bernoulli mark with probability $\\sigma$(u), and conditioning on all marks being zero. X^$\\sigma$ is again determinantal, with kernel K^$\\sigma$ = $\\sqrt$(1-$\\sigma$) K (1-$\\sigma$ K)^{-1} $\\sqrt$(1-$\\sigma$). Theorems 2.3 and 2.7 state this for fixed and varying reference measures, respectively, and the applications show that bulk deformations of orthogonal polynomial ensembles yield deformed sine processes, edge deformations yield deformed Airy processes, and deformed discrete Coulomb gases yield deformed discrete sine processes.","pith_inferences":["The formula K^sigma = sqrt(1-sigma) K (1-sigma K)^{-1} sqrt(1-sigma) is an operator-valued transformation on the limiting kernel; viewing the limit as a functional calculus could allow classifying which sigma produce genuinely new universal processes.","The sub-microscopic case (t > 0) shows a sharp scale threshold, and one could test whether the deformed limit varies continuously as the deformation scale crosses from microscopic to sub-microscopic.","The same PGF-based transfer could apply to other biorthogonal limits, such as Bessel or Pearcey kernels, whenever analogous one-point bounds are available; the paper's assumptions are formulated to make such checks routine.","For the discrete sine case, the paper's condition uses only local bounds on the kernel from the existing literature, suggesting the method extends to other tiling or exclusion models where such bounds are known."],"forward_implications":["Bulk deformations of orthogonal polynomial ensembles with weight e^{-nV} converge to sigma-deformations of the sine point process; if the deformation is on sub-microscopic scales, the limit is the plain sine process (Corollary 3.1).","Edge deformations of orthogonal polynomial ensembles converge to sigma-deformations of the Airy point process, covering hard-edge-to-soft-edge transitions and kernels tied to Painlevé II (Corollary 3.2).","Deformed discrete Coulomb gases converge to deformed discrete sine point processes on the limiting lattice (Corollary 3.5).","Several Painlevé-type kernels in the literature are revealed to be the correlation kernels of conditioned and thinned sine or Airy processes, giving them a ready probabilistic interpretation (Remark 3.3).","The framework turns universality into a transferable property: any class of deformations sigma generates new explicit limit kernels from an existing limit kernel."],"supporting_citations":[{"why":"Claeys and Glesner's marking-and-conditioning construction, which supplies the identity K^sigma = sqrt(1-sigma) K (1-sigma K)^{-1} sqrt(1-sigma) and the deformed PGF formula used as the starting point.","marker":"[24]"},{"why":"Borodin's biorthogonal ensembles paper, establishing that ensembles of the form (2.1) are determinantal point processes with kernel (2.6).","marker":"[15]"},{"why":"Daley and Vere-Jones' point process text, providing the equivalence between weak convergence of point processes and pointwise convergence of probability generating functionals.","marker":"[28]"},{"why":"Levin and Lubinsky's bulk universality result for varying measures, used in Section 3.1 for the sine-kernel limit and the one-point bound (3.5).","marker":"[65]"},{"why":"Deift, Kriecherbauer, McLaughlin, Venakides, and Zhou's uniform asymptotics for orthogonal polynomials with varying exponential weights, used for the Airy-kernel edge limit and the estimate (3.14).","marker":"[34]"},{"why":"Baik, Kriecherbauer, McLaughlin, and Miller's monograph on discrete orthogonal polynomials, supplying the discrete sine kernel convergence and local bounds that verify Assumption 2.6 for discrete Coulomb gases.","marker":"[4]"}],"fun_headline_variants":["Deformed ensembles keep their limit law via marking","Marking and conditioning preserve universality","Universality survives deformations in particle ensembles","Limit and deformation commute for biorthogonal ensembles"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs a uniform domination bound: for every bounded region, the matrix entries of the scaled kernel, with the deformation factor inserted, must be bounded by a product Phi(u)Psi(v) with Phi and Psi square-integrable; if no such bound holds, the dominated-convergence argument in the paper does not go through.","fun_headline_variants_meta":{"raw":{"variants":["Deformed ensembles keep their limit law via marking","Marking and conditioning preserve universality","Universality survives deformations in particle ensembles","Limit and deformation commute for biorthogonal ensembles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00016,"raw_usage":{"total_tokens":1201,"prompt_tokens":881,"completion_tokens":320,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":497,"completion_tokens_details":{"reasoning_tokens":263}},"tokens_in":497,"tokens_out":320,"duration_ms":3408,"temperature":1.0,"reasoning_tokens":263,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:51:41.844400+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search for a sequence of biorthogonal ensembles whose kernels converge pointwise to a limit but whose diagonal grows too fast for the factorized $L^{2}$ bound to hold, for instance an orthogonal polynomial ensemble at a point where the equilibrium density vanishes. For a fixed deformation $\\sigma$, compute the deformed probability generating functional numerically for increasing n; if the limit differs from the Fredholm determinant of the deformed kernel K^$\\sigma$, the transfer principle fails exactly where the domination assumption is essential.","supporting_citations":[{"cited_title":"Determinantal point processes conditioned on randomly incomplete configurations","cited_arxiv_id":"2112.10642","evidence_quote":"Claeys and Glesner's marking-and-conditioning construction, which supplies the identity K^sigma = sqrt(1-sigma) K (1-sigma K)^{-1} sqrt(1-sigma) and the deformed PGF formula used as the starting point."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Borodin's biorthogonal ensembles paper, establishing that ensembles of the form (2.1) are determinantal point processes with kernel (2.6)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Daley and Vere-Jones' point process text, providing the equivalence between weak convergence of point processes and pointwise convergence of probability generating functionals."},{"cited_title":"Levin and D","cited_arxiv_id":null,"evidence_quote":"Levin and Lubinsky's bulk universality result for varying measures, used in Section 3.1 for the sine-kernel limit and the one-point bound (3.5)."},{"cited_title":"Deift, T","cited_arxiv_id":null,"evidence_quote":"Deift, Kriecherbauer, McLaughlin, Venakides, and Zhou's uniform asymptotics for orthogonal polynomials with varying exponential weights, used for the Airy-kernel edge limit and the estimate (3.14)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Baik, Kriecherbauer, McLaughlin, and Miller's monograph on discrete orthogonal polynomials, supplying the discrete sine kernel convergence and local bounds that verify Assumption 2.6 for discrete Coulomb gases."}],"review_version":1}