{"id":"bbdb4787-1507-4611-8570-45b9a6158ccf","arxiv_id":"2607.04047","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"In the triple-scaling limit the top singular value of sum-products of Gaussian matrices coincides with the REM log-partition function at inverse temperature β=√[2(N-1)/(n log m)].","lead":"The top singular value of a sum of products of Gaussian random matrices equals the free energy of a random energy model in the triple limit m,n,N to infinity. This identifies a high/low-temperature phase transition for the singular value controlled by an inverse-temperature parameter built from N, n and m.","discovery_kind":"unification","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The reader correctly isolates the Cramér-range restriction as the weakest (and self-acknowledged) technical point while correctly judging that it does not undermine the theorems as stated. The proofs are complete, the phase diagram is cleanly derived from the REM free-energy formula, and the non-asymptotic error bounds are explicit. No stronger load-bearing concern surfaces upon re-examination of the reduction steps, the Gaussian approximation, or the Laplace-method concentration arguments. Consequently the ACCEPT / high-confidence verdict stands without adjustment.","tokens_in":30756,"tokens_out":448,"duration_ms":4344,"concrete_test":"Independently re-derive the moderate-deviation statement of Proposition 10 for the centered log-chi-squared variables under the weaker hypothesis x = o(N^{1/4}) (or better) using modern Berry–Esseen or Edgeworth expansions; if the same tail asymptotics hold, the global hypothesis log m = o(N^{1/3}) can be relaxed exactly as the authors conjecture, confirming that the restriction is technical rather than essential.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Theorems 1–2) is a non-asymptotic approximation of log s1(X) and log ||X\theta|| by the REM free energy Z at the stated eta, under the explicit growth conditions of the paper. The reduction (Lemmas 4–5 + Corollary 6) to a non-Gaussian REM, the Cramér moderate-deviation approximation of the energies (Proposition 10), and the subsequent concentration of the log-partition function (Propositions 7–8) are all carried through with explicit error terms. The only technical restriction that is not fully optimal is log m = o(N^{1/3}) (and N = o(n^3)), which the authors themselves flag in Remark 1 as coming from the classical Cramér range and as likely improvable; it does not create an internal inconsistency or invalidate the stated theorems. No circularity, hidden free parameters, or gaps in the argument appear.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper studies the top singular value s1(X) (equivalently the top Lyapunov exponent) of the sum-product matrix X = m^{-1/2} sum_{i=1}^m X_i, where each X_i is a product of N i.i.d. n\times n Gaussian matrices with N(0,n^{-1}) entries. In the triple-scaling regime m,n,N\to∞ the authors show that log s1(X) and log∥Xθ∥ (for fixed unit θ) are approximated by the log-partition function of a random energy model (REM) whose inverse temperature is β=√[2(N-1)/(n log m)] and whose energies are non-Gaussian but approximately Gaussian with law depending on N/n. Theorems 1 and 2 give explicit non-asymptotic error bounds that exhibit a high-temperature/low-temperature phase transition at β=√2, recovering the classical Gaussian REM free energy after recentering. The argument proceeds by reducing the matrix problem to a non-Gaussian REM (Lemmas 4–5, Corollary 6), approximating the energies via a Cramér moderate-deviation theorem (Proposition 10), and controlling the REM partition function by Markov and Laplace-method arguments (Propositions 7–8).","tokens_in":30977,"tokens_out":898,"duration_ms":8255,"significance":"The work supplies the first non-asymptotic description of the top singular value of a sum of many random matrix products in the simultaneous limit m,n,N\to∞. The identification with a REM free energy at a temperature that depends on all three parameters is conceptually clean and yields a transparent phase diagram (Figures 2–3) that interpolates between free-probability, ergodic-theory and double-scaling regimes previously studied only for m=1. The proofs are fully quantitative, track error terms explicitly, and rest only on classical concentration and moderate-deviation tools; the authors themselves flag the technical restrictions log m=o(N^{1/3}) and N=o(n^3) as improvable. The results therefore constitute a solid, self-contained contribution to non-asymptotic random matrix theory and to the interface with disordered systems.","major_comments":[],"minor_comments":[{"comment":"Remark 1 correctly notes that the hypothesis log m=o(N^{1/3}) originates from the classical Cramér range x=o(N^{1/6}). A short additional sentence indicating whether sharper moderate-deviation results (e.g., under sub-exponential tails of the log-chi-squared variables) could remove the restriction would help the reader assess optimality.","section":"Remark 1"},{"comment":"In the statement of Theorem 2 the additive error (1+2α)log n/(2β^{2} log m) appears only after the sphere-supremum removal (Lemma 4). It would be clearer to display this term already in the high-temperature regime of Theorem 1, or to note explicitly that it is an artifact of the net argument rather than of the REM approximation.","section":"Theorem 2"},{"comment":"Figures 1–3 are informative but the captions are dense. A one-line summary of the phase transition (β=√2) placed above each figure would improve readability.","section":"Figures 1–3"},{"comment":"The comparison with the classical REM literature (Section 3.1) is accurate, yet a brief remark on the difference between almost-sure convergence along the subsequence m=2^M and the in-probability statements of Propositions 7–8 would prevent possible confusion for readers coming from statistical physics.","section":"Section 3.1"}],"recommendation":"accept","confidential_remarks":"The manuscript is technically solid and fits well within the scope of a probability journal that publishes non-asymptotic random-matrix results. The only potential novelty concern is the extent to which the REM reduction itself is new versus a clever packaging of existing concentration tools; I view the packaging and the quantitative phase diagram as sufficient for acceptance."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The new thing here is the identification of the top singular value of an m-sum of N-fold Gaussian products with a REM free energy at inverse temperature beta = sqrt(2(N-1)/(n log m)), together with explicit non-asymptotic error bounds that produce a phase diagram controlled by beta rather than by the usual ratio N/n. That regime was open; previous work (Hanin–Paouris, free-probability literature, Liu et al.) treated only m=1 or n to infinity with N fixed.\n\nWhat they do well is the reduction itself. Lemma 4 removes the sphere supremum by the same right-invariance argument that works for a single product. Lemma 5 and Corollary 6 concentrate the last Gaussian layer so that log ||X theta|| is close to the log-partition function of a non-Gaussian REM whose energies are sums of log-chi-squared variables. Proposition 10 then supplies a location-dependent Cramér moderate-deviation approximation of those energies, and Propositions 7–8 give quantitative concentration of the resulting partition function in the ultra-high and moderate-to-low temperature regimes. Error terms are tracked carefully; the comparison with the classical Gaussian REM free energy is only for orientation, not an input that forces the answer. The proofs are complete and self-contained.\n\nThe soft spot is exactly the one the authors flag in Remark 1: the Cramér range forces log m = o(N^{1/3}) (and N = o(n^3)). That is a genuine technical limitation, not an inconsistency, and they conjecture it is improvable. Everything else—circularity, free parameters, citation pattern—looks clean. The paper is pure analytic probability aimed at people who already care about Lyapunov exponents of matrix products or REM free energies. It deserves a serious referee; I would bring it to reading group and expect to cite the phase diagram when the triple-scaling regime comes up.","headline":"Clean non-asymptotic REM reduction for the top singular value of sum-products in the open triple-scaling regime; the only soft spot is a flagged technical range restriction that does not break the theorems.","tokens_in":31565,"tokens_out":502,"would_cite":true,"duration_ms":4959,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60B20","15B52","82B44"],"pacs":[],"model":"grok-4.5","headline":"The top singular value of a sum of random matrix products equals a random-energy-model free energy at temperature set by N, n and m.","keywords":["random matrix products","top singular value","random energy model","Lyapunov exponent","triple scaling","moderate deviations","phase transition"],"falsifier":"Compute the top singular value of the sum-product matrix for a sequence of triples (m,n,N) that violate log m = o(N^{1/3}) and check whether the observed value still tracks the REM free-energy formula within the claimed error; a systematic deviation would falsify the approximation.","tokens_in":31672,"feed_emoji":"🎲","tokens_out":634,"duration_ms":5488,"temperature":0.7,"pith_summary":"This paper studies the largest singular value of a sum of m independent products of N Gaussian matrices of size n. When all three dimensions grow, that singular value is shown to coincide with the free energy of a classical random-energy model whose inverse temperature is fixed by the combination sqrt of 2(N-1)/(n log m). The energies themselves are non-Gaussian but become Gaussian in the scaling limit, so the well-known REM phase transition at critical temperature sqrt(2) appears as a sharp change in the singular-value asymptotics. High-temperature regimes are dominated by many typical matrix products; low-temperature regimes are dominated by a few extreme products. Explicit non-asymptotic error bounds make the correspondence quantitative under mild growth conditions, giving a single phase diagram that interpolates between free-probability, ergodic-theory and moderate-deviation regimes previously treated separately.","feed_headline":"Top singular value equals REM free energy","feed_subtitle":"A single temperature parameter built from N, n and m organises the whole phase diagram","key_machinery":"Reduction of the top Lyapunov exponent to the log-partition function of a non-Gaussian REM, followed by a Cramer moderate-deviation approximation of the energies by Gaussians and Laplace-method concentration of that partition function.","core_discovery":"When m, n and N tend to infinity simultaneously, the top singular value of the sum-product matrix X is asymptotically identical to the log-partition function of a random energy model whose inverse temperature is beta = sqrt(2(N-1)/(n log m)) and whose energies depend on the ratio N/n. The identification is made precise by two non-asymptotic theorems that control the approximation error both above and below the critical temperature sqrt(2).","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Top singular value coincides with REM free energy","Sum-product matrix top singular value equals REM log-partition","Matrix sum-products top value matches REM free energy","Top singular value of sum-products is REM partition function","Sum of matrix products top singular value equals REM free energy"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"The Gaussian approximation of the individual energies is valid only when log m grows slower than N to the power one-third; if that growth condition fails the error bounds no longer hold.","fun_headline_variants_meta":{"raw":{"variants":["Top singular value coincides with REM free energy","Sum-product matrix top singular value equals REM log-partition","Matrix sum-products top value matches REM free energy","Top singular value of sum-products is REM partition function","Sum of matrix products top singular value equals REM free energy"]},"model":"grok-4.5","effort":"low","cost_usd":0.003296,"raw_usage":{"total_tokens":1048,"prompt_tokens":652,"num_sources_used":0,"completion_tokens":80,"cost_in_usd_ticks":32960000,"prompt_tokens_details":{"text_tokens":652,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":316,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":652,"tokens_out":80,"duration_ms":3301,"temperature":1.0,"reasoning_tokens":316,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-11T22:02:15.379068+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Compute the top singular value of the sum-product matrix for a sequence of triples (m,n,N) that violate log m = o(N^{1/3}) and check whether the observed value still tracks the REM free-energy formula within the claimed error; a systematic deviation would falsify the approximation.","supporting_citations":[],"review_version":1}