{"id":"b2c1af95-0962-4157-a268-3598969d6275","arxiv_id":"2608.00788","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Positive bidiagonal factorizations turn banded Markov kernels into ordered urn experiments and yield Karlin-McGregor type spectral formulas without reversibility.","lead":"This paper develops a spectral and probabilistic theory for Markov chains with wide jumps by factoring each transition matrix into simple one-step birth-or-stay and death-or-stay moves. It connects these factorizations to orthogonal polynomials, urn experiments, and return-time formulas, giving explicit models beyond the classic birth-death case.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The Karlin–McGregor representation (9) is the load-bearing pillar, and its proof in Cor. 3.2 is entirely an import from [10] and [11, Thm 2.19]; the paper does not state or verify their hypotheses, and the companion continued-fraction evaluation [33] is explicitly not reproved.","rationale":"The paper's strongest claim—that an ordered PBF determines the mixed-type spectral representation, the elementary Markov transitions, and the return law—depends on two imported results: the Favard-type theorem of [10]/[11, Thm 2.19] for formula (9), and the continued-fraction boundary evaluation of [33] for formulas (7)--(8) and (39)--(40). The manuscript explicitly flags the second as not reproved, and the first is used in Corollary 3.2 with no statement of hypotheses. This is a genuine correctness risk: if either import is wrong or has narrower scope, the general Karlin--McGregor formulas lose support. The paper's self-contained portions, especially the Piñeiro positivity classification and the explicit factor formulas, are detailed and contain no evident internal contradiction, so I do not propose rejection. Rather, the conditional status is appropriate until the imported hypotheses are stated and checked, or until a nontrivial PBF example is used to validate the representation independently. My proposed numerical/analytical check on the explicit rational (3,2) Piñeiro model would settle whether the imported representation is at least consistent on a concrete instance inside the paper's claimed domain.","tokens_in":66985,"tokens_out":11210,"duration_ms":140665,"concrete_test":"For the rational (p,q)=(3,2) Piñeiro example of §8.2, compute both sides of formula (101) for all k=0,...,10 and N,M≤5: the left side by direct multiplication of the five explicit stochastic factors (or matrix powers of P^P), and the right side by quadrature of Σ_{a,b} ∫ \\tilde B_N^{(b)}(x)\\tilde A_M^{(a)}(x) x^{k+α_a+β_b} dx using the explicit polynomials from (60)/(62). Any mismatch beyond quadrature error would show that the imported Favard theorem does not apply in its present form to a valid PBF; exact agreement would support the import and make the paper's central example self-checking.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Corollary 3.2 proves the central discrete spectral representation by asserting that an ordered product of positive stochastic bidiagonal factors satisfies the hypotheses of [10] and [11, Theorem 2.19], without stating those hypotheses or verifying them for the infinite banded stochastic matrix. Every later Karlin--McGregor formula—(16), (17), (27), (29), and the endpoint criteria of Theorems 4.7--4.9—inherits this representation, so a failure or narrowing of the imported Favard theorem would remove support from the paper's central claim, not just from a peripheral example. The same pattern appears for the first-return law: §1.1 and §3.2 explicitly say the boundary evaluation of the matrix continued fraction from [33] is not reproved, and (7) is then combined with the renewal identity to obtain (8), and again in §6 for the QBD return matrices (39)--(40). No internal inconsistency was found in the self-contained Piñeiro and Jacobi-like algebra, but that algebra does not independently establish the general representation. The load-bearing condition is therefore: the imported theorems must apply to every ordered positive stochastic bidiagonal product, and their hypotheses must be made explicit.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a spectral and probabilistic theory for banded Markov transition matrices that admit an ordered positive bidiagonal factorization (PBF). It claims that the individual bidiagonal factors determine two families of mixed-type multiple orthogonal polynomials, an entrywise positive q×p matrix of measures, and a sequence of elementary death-or-stay and birth-or-stay transitions, yielding Karlin–McGregor formulas for transition probabilities, Green kernels, resolvents, potentials, and first-passage transforms without reversibility or block symmetrizability. The paper also gives finite-urn equivalences, Darboux intertwinings by cyclic reordering, a factor-resolved continued fraction for first-return laws, QBD reformulations, bounded and unbounded continuous-time generators, and explicit classifications for the mixed Piñeiro and Jacobi-like systems, including a rational (3,2) example.","tokens_in":67264,"tokens_out":3263,"duration_ms":44067,"significance":"If the main claims hold, the paper gives a genuinely unified framework: the same bidiagonal factorization simultaneously certifies total nonnegativity, supplies spectral measures and orthogonal polynomial families, and provides elementary Markov transitions and continued-fraction return laws. The explicit Piñeiro treatment is a substantial concrete contribution: the Cauchy construction in §7, the exact PBF region in Theorem 7.8, the structural-band positivity regions in Theorem 7.12, and the worked (3,2) example in §8 are all given by finite, explicit formulas and sign counts. The block-symmetrizability obstruction in Corollary 6.4 is a useful and clearly stated result. The paper is weaker where it depends on external results: the key spectral representation and the continued-fraction boundary evaluation are imported from the author's earlier work and are not reproved or even stated with their hypotheses.","major_comments":[{"comment":"The central Karlin–McGregor representation is proved by the sentence: 'The bounded spectral Favard theorem [10] and its Markov specialization [11, Theorem 2.19] give the two mixed-type polynomial families, the entrywise positive matrix of measures, and formula (9).' The hypotheses of those theorems are not stated, and the paper does not verify that an arbitrary infinite ordered product of positive stochastic bidiagonal factors satisfies them. Since Eqs. (16), (17), (27), (29), and the endpoint criteria of Theorems 4.7–4.9 all inherit this representation, this is a load-bearing gap. The authors should either state the precise hypotheses and prove that every ordered positive stochastic PBF product satisfies them, or give a self-contained proof of (9).","section":"§3.3, Corollary 3.2 and Eq. (9)"},{"comment":"The first-return law (8) and the QBD return matrices (39)–(40) depend on the boundary evaluation of the matrix continued fraction from the companion paper [33]. The paper explicitly says this construction is 'not reproved here' and is 'combined with the renewal identity.' This is acceptable only if the companion result is stated with enough precision to allow independent verification. As written, the reader cannot check the central probabilistic application from the present manuscript. Please state the exact theorem imported from [33], including its hypotheses and convergence conditions, or provide the proof.","section":"§3.2 and §6, Eqs. (7)–(8), (39)–(40)"},{"comment":"The stochastic normalization factor by factor uses [11, Proposition 2.3] to convert the positive PBF of T^P into a positive stochastic PBF of P^P. The conversion requires the intermediate vectors h^[r] to be strictly positive. In the ordered cyclic PBF region this follows from positivity of the bidiagonal factors, and the harmonic vector is explicit via (63). However, the paper does not state the domain of validity of [11, Proposition 2.3] for infinite matrices. Since the same proposition is used in Theorem 3.1 for the rational urn equivalence, the infinite-matrix hypothesis should be made explicit.","section":"§7.4, Eqs. (97)–(99)"},{"comment":"The proof of the Jacobi-like PBF region invokes Lemma 9.2 and a Gasca–Peña criterion for strict total positivity, and it introduces a 'large-minor obstruction' in Proposition 9.5 whose proof is only sketched ('the term omitting the largest s labels is the unique dominant term'). This asymptotic dominance argument is load-bearing for the claimed maximality in Proposition 9.5 and Corollaries 9.6–9.8. Please provide the full error estimates or a complete proof of the dominance claim.","section":"§9, Theorem 9.4"}],"minor_comments":[{"comment":"The phrase 'withinfn vn >0' is a typo; it should read 'with inf_n v_n >0'.","section":"Definition 5.3"},{"comment":"The notation L^{(1)} ... L^{(p)} with upper indices is clear, but in §3.1 and later the same factors are written M_1,...,M_d with a type word ε. A short table linking the two notations would improve readability.","section":"§2.1.2, Definition 2.2"},{"comment":"In Proposition 8.5, the claim that (P^P)^k_{0,0} = ϑ/(k+ϑ) is a striking closed form. It would help to note explicitly that this uses the normalization choices A_0^{(1)} = ϑ and B_0^{(1)} = 1 under the Piñeiro spectral measure, so the reader does not confuse this with the generic normalization in Corollary 3.2.","section":"§8.3, Eq. (117)"},{"comment":"The Abel limit lim_{r↑1} is introduced but the spectral measure ψ^{(λ)} is not defined in the text. Please define it precisely, including the normalization of A^{(a,λ)} and B^{(b,λ)}.","section":"§5.3, Eq. (27)"}],"recommendation":"major_revision","confidential_remarks":"This manuscript builds a very large structure on a small number of imported results from the author's own previous papers ([10], [11], [33]). For a journal publication, the editor should require that the hypotheses of the imported Favard theorem and continued-fraction boundary evaluation be stated and checked, since they are the load-bearing pillars of the claimed Karlin–McGregor representation and return laws. The self-contained Piñeiro and Jacobi-like computations appear careful and are likely publishable even if the general framework must be trimmed or qualified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this is a serious and mostly explicit paper. The genuinely new pieces — Theorem 3.1 (rational positive stochastic bidiagonal factorizations exactly equivalent to ordered finite-urn experiments), Proposition 5.1 (the sharp obstruction forcing unbounded conservative generators under scalar shifts to be tridiagonal), and the exact PBF region for mixed Piñeiro systems (Theorem 7.8) — are self-contained, clearly argued, and I found no internal inconsistency in the parts that are proved here.\n\nThe positivity classification in Sections 7–9 is the real workhorse and it is well done: explicit Pochhammer-product formulas, exact ordered cyclic region, larger structural-band positivity regions, Gamma-cancellation strata for the Jacobi-like family, and a fully worked rational (3,2) example with every factor, urn probability, first-return law, and QBD block spelled out. That is reproducible and worth having.\n\nThe soft spot is the one the stress-test flags, and I think it is a genuine load-bearing issue. The central spectral representation — formulas (9), (16), (17), (27), (29) — is not proved in this paper. Corollary 3.2 imports it from the author's earlier bounded Favard theorem [10] and [11, Thm 2.19], and the first-return law (8) rests on the boundary evaluation of a matrix continued fraction from the companion paper [33]. The author is transparent about this (\"The continued-fraction construction of [33] is not reproved here\"), but the hypotheses of the imported theorems are not stated in the present text. If those theorems apply to every ordered positive stochastic bidiagonal product as claimed, everything holds together; if their hypotheses are narrower, the spectral pillar of the paper loses support. This is not a minor expositional issue, because the paper's abstract explicitly claims the factors determine the spectral data. A referee cannot verify that claim without going to three prior papers.\n\nThis also means the citation pattern deserves scrutiny: the paper leans heavily on the author's own prior work for the load-bearing results. Self-citation is not a flaw when the cited results are solid, but here the most important ones are not reproduced or even stated.\n\nWho should read this: researchers in total positivity, multiple orthogonal polynomials, and spectral theory of Markov processes. It deserves a serious referee rather than a desk rejection. The revision should state the hypotheses of the imported Favard theorems and verify that the PBF products in Theorem 3.1 satisfy them, or prove the needed special case directly. My recommendation: send to peer review, with that condition attached.","headline":"A substantial, mostly self-contained extension of Mañas's PBF spectral program, with the central Karlin–McGregor representation inherited from two prior papers whose hypotheses are never stated.","tokens_in":67762,"tokens_out":3071,"would_cite":true,"duration_ms":32879,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["15B48","42C05","60J10","60J27","47B36"],"pacs":[],"model":"deepseek-v4-flash","headline":"An ordered positive bidiagonal factorization is a full spectral and probabilistic description of a banded Markov chain, not just a certificate of total nonnegativity.","keywords":["positive bidiagonal factorization","total positivity","banded Markov processes","mixed-type multiple orthogonality","Karlin–McGregor representation","matrix continued fractions","Piñeiro polynomials","Jacobi-like weights"],"falsifier":"Take the rational (3,2) Piñeiro chain with α=(−1/3,0,1/3), β=(0,1/3), enumerate the return probabilities to state 0 from the explicit transition matrix (111), and compare the Taylor coefficients of f0(z) with f0(z)=1−1/2F1(1,2/3;5/3;z); a single mismatch at any order would refute the continued-fraction/renewal identity.","tokens_in":66849,"feed_emoji":"🎲","tokens_out":9279,"duration_ms":102215,"temperature":0.7,"pith_summary":"This paper tries to establish that an ordered positive bidiagonal factorization (PBF) of a banded Markov transition matrix carries the entire spectral and probabilistic structure of the chain. The same factors that certify total nonnegativity are shown to determine two families of mixed-type multiple orthogonal polynomials, an entrywise positive matrix of measures, and an ordered sequence of elementary death-or-stay and birth-or-stay Markov transitions. From these, the paper derives Karlin–McGregor formulas for transition probabilities, Green kernels, resolvents, potentials, and first-passage transforms without requiring reversibility or block symmetrization. It also characterizes rational stochastic PBFs by ordered finite-urn experiments, links cyclic reorderings to Darboux transformations, and obtains exact positivity classifications for the mixed Piñeiro and Jacobi-like systems. A sympathetic reader would care because this extends the classical birth–death spectral theory to arbitrary bandwidth under one explicit structural hypothesis.","feed_headline":"One factorization carries the whole spectrum of a banded Markov chain","feed_subtitle":"The same factors give mixed orthogonal polynomials, finite urns, and return laws without reversibility.","key_machinery":"The central object is the ordered positive stochastic bidiagonal factorization T=M1⋯Md, where every Mi is bidiagonal and positive: lower factors allow only n→n−1 or stay, upper factors only n→n+1 or stay. This one structure simultaneously encodes the elementary Markov transitions, the mixed-type multiple orthogonal polynomials and positive matrix of measures through the Favard theorem, the finite-urn realization through reduced rational probabilities, and the matrix continued fraction for return laws. A Doob-type normalization by the positive harmonic vector h_N=B_N(1) turns the recurrence matrix into the row-stochastic transition kernel without changing the spectral measure.","core_discovery":"The central claim is that a banded transition matrix factored as T=L1⋯LpUq⋯U1, with positive lower and upper bidiagonal factors, is more than a totally nonnegative matrix: the ordered factors themselves are the spectral and probabilistic data. Each lower factor is a death-or-stay binary experiment and each upper factor is a birth-or-stay experiment; Theorem 3.1 proves that for rational entries this is exactly an ordered sequence of finite two-color urn experiments. Through the mixed-type Favard theory, the same factors generate two left–right polynomial families and a positive q×p matrix of measures on [0,1], giving explicit Karlin–McGregor formulas for all powers of the kernel, the Green ke","pith_inferences":["If Theorem 3.1 is read as a simulation recipe, any rational banded chain with a PBF can be sampled exactly by fixed-order draws from finite urns, which suggests a practical exact Monte Carlo scheme even for large bandwidths and non-reversible wide jumps.","The Piñeiro classification that strict cyclic ordering is the only open PBF chamber for q∈{2,3,4} invites the conjecture that cyclic order may be the generic open chamber for every q, with lower-dimensional cancellation strata described by combinatorial conditions; the paper does not assert this for all q.","Corollary 6.4 implies that when p≠q one should not search for a block-diagonal symmetrizer to apply classical matrix-valued orthogonal polynomial methods; the two-sided mixed spectral representation is the natural object, potentially guiding numerical algorithms for phase-resolved return probabilities.","The state-dependent construction Q=V(T−I) suggests that the embedded chain, not the holding-time clock, controls qualitative long-time behavior; one testable extension is whether recurrence and eventual hitting probabilities remain invariant under arbitrary positive state-dependent time changes of the displayed type."],"forward_implications":["For any banded stochastic matrix with a positive stochastic bidiagonal factorization, every transition probability admits the explicit mixed Karlin–McGregor form (9); in bounded continuous time, uniformization merely replaces x^k by e^{-νt(1−x)} and leaves the polynomial families and measure matrix unchanged.","Rational stochastic PBFs are exactly ordered finite-urn experiments: each factor is realized by the smallest finite two-color urn determined by its reduced probabilities, and one full cycle of draws realizes one transition of the chain.","The complete first-return law from state 0 is determined by the boundary matrix continued fraction of the individual factors via the renewal identity f0(z)=1−1/g0(z); for the rational (3,2) Piñeiro chain this becomes the explicit formula f0(z)=1−1/2F1(1,2/3;5/3;z).","Bounded generators built by Poissonization inherit the discrete spectral data: recurrence is decided by divergence of ∫ dψ_{1,1}(x)/(1−x), ergodicity by a mass point at x=1, and endpoint exponents give power-law decay in k or t.","For unbounded exit rates, a conservative generator whose shifted leading truncations all admit scalar PBFs must be tridiagonal, but the construction Q=V(T−I) preserves arbitrary fixed bandwidth, is non-explosive when holding rates are linear, and its recurrence and hitting probabilities are those of the embedded chain T."],"supporting_citations":[{"why":"Supplies the bounded spectral Favard theorem that produces the two mixed-type multiple orthogonal polynomial families and the positive matrix of measures used in Corollary 3.2 and Theorem 4.3.","marker":"[10]"},{"why":"Provides the stochastic specialization (Theorem 2.19) and Proposition 2.3, converting positive bidiagonal factorizations into stochastic factorizations and yielding the Karlin–McGregor representation.","marker":"[11]"},{"why":"Supplies the factor-resolved matrix continued fraction whose boundary evaluation gives the first-return law (7)–(8); the paper uses it without reproving the construction.","marker":"[33]"},{"why":"Supplies the general Christoffel–Gauss–Borel bidiagonal construction from which the explicit Piñeiro and Jacobi-like factor coefficients are derived.","marker":"[7]"},{"why":"Provides the Jacobi-like mixed system and the hypergeometric/Kampé de Fériet formulas that Section 9 generalizes and whose Gamma cancellations are classified.","marker":"[9]"},{"why":"Gives the initial-minor criterion for strict total positivity used to prove the Jacobi-like PBF region and the cancellation statements.","marker":"[22]"},{"why":"Supplies the classical uniformization identity used to pass from discrete spectral formulas to continuous-time semigroup and resolvent formulas.","marker":"[13]"},{"why":"Provides the shifted-truncation Favard theorem for unbounded banded matrices against which the scalar-shift obstruction and the state-dependent construction are framed.","marker":"[12]"},{"why":"Supplies the Doob h-transform used to normalize the positive harmonic vector into a row-stochastic banded Markov kernel.","marker":"[16,15]"}],"fun_headline_variants":["Bidiagonal factorization: full spectral data for Markov chains","No reversibility needed: one PBF gives the whole spectrum","From factors to urn experiments and return laws in one go","Banded Markov chain? Factor it and read off every law","Ordered bidiagonal factors replace reversibility assumptions"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that the earlier Favard theorems ([10] and [11, Theorem 2.19]) and the companion paper's boundary evaluation of the factor-resolved continued fraction [33] apply exactly as stated to every positive stochastic bidiagonal factorization; the paper explicitly says the continued-fraction construction is not reproved here, so if those imported results fail or need narrower hypotheses, the spectral formulas and the first-return law lose support.","fun_headline_variants_meta":{"raw":{"variants":["Bidiagonal factorization: full spectral data for Markov chains","No reversibility needed: one PBF gives the whole spectrum","From factors to urn experiments and return laws in one go","Banded Markov chain? Factor it and read off every law","Ordered bidiagonal factors replace reversibility assumptions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000405,"raw_usage":{"total_tokens":1993,"prompt_tokens":844,"completion_tokens":1149,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":588,"completion_tokens_details":{"reasoning_tokens":1067}},"tokens_in":588,"tokens_out":1149,"duration_ms":13083,"temperature":1.0,"reasoning_tokens":1067,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T00:11:16.338055+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the rational (3,2) Piñeiro chain with α=(−1/3,0,1/3), β=(0,1/3), enumerate the return probabilities to state 0 from the explicit transition matrix (111), and compare the Taylor coefficients of f0(z) with f0(z)=1−1/2F1(1,2/3;5/3;z); a single mismatch at any order would refute the continued-fraction/renewal identity.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the bounded spectral Favard theorem that produces the two mixed-type multiple orthogonal polynomial families and the positive matrix of measures used in Corollary 3.2 and Theorem 4.3."},{"cited_title":"arXiv:2601.10890 [math.PR]","cited_arxiv_id":null,"evidence_quote":"Provides the stochastic specialization (Theorem 2.19) and Proposition 2.3, converting positive bidiagonal factorizations into stochastic factorizations and yielding the Karlin–McGregor representation."},{"cited_title":"Matrix continued fractions from bidiagonal factorizations","cited_arxiv_id":null,"evidence_quote":"Supplies the factor-resolved matrix continued fraction whose boundary evaluation gives the first-return law (7)–(8); the paper uses it without reproving the construction."},{"cited_title":"Bidiagonal factorization of banded recursion matrices for mixed-type multiple orthogonal polynomials.Linear Algebra and its Applications(2026)","cited_arxiv_id":null,"evidence_quote":"Supplies the general Christoffel–Gauss–Borel bidiagonal construction from which the explicit Piñeiro and Jacobi-like factor coefficients are derived."},{"cited_title":"Jacobi-like and Laguerre-like mixed-type multiple orthogonal polynomials","cited_arxiv_id":null,"evidence_quote":"Provides the Jacobi-like mixed system and the hypergeometric/Kampé de Fériet formulas that Section 9 generalizes and whose Gamma cancellations are classified."},{"cited_title":"Total positivity, QR factorization, and Neville elimination.SIAM Journal on Matrix Analysis and Applications 14(1993) 1132–1140.doi:10.1137/0614077","cited_arxiv_id":null,"evidence_quote":"Gives the initial-minor criterion for strict total positivity used to prove the Jacobi-like PBF region and the cancellation statements."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the classical uniformization identity used to pass from discrete spectral formulas to continuous-time semigroup and resolvent formulas."},{"cited_title":"arXiv:2601.12453 [math.CA]","cited_arxiv_id":null,"evidence_quote":"Provides the shifted-truncation Favard theorem for unbounded banded matrices against which the scalar-shift obstruction and the state-dependent construction are framed."}],"review_version":1}