{"id":"5323d382-6f5a-4aa7-be60-99deec26f227","arxiv_id":"1908.10353","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The logarithmic derivative of the finite-dimensional distributions of the KPZ fixed point solves the matrix and scalar KP equation.","lead":"The paper shows that the probability distributions describing one-dimensional random interface growth satisfy the Kadomtsev-Petviashvili (KP) equation, a classical integrable PDE. This links the KPZ universality class to the theory of integrable systems and recovers the Tracy-Widom distributions as special solutions.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1's derivation appears sound; the load-bearing gap is that the title/abstract claim KP governs the whole one-dimensional KPZ class rests on unproven universality, not on the TASEP-based proof.","rationale":"The reader's weakest assumption is the right one: the paper's mathematical theorem is about the TASEP-derived KPZ fixed point, and the universal interpretation in the title and abstract is only a widely believed conjecture. My read of the proof of Theorem 1.1 found no internal flaw: the kernel differential relations (3.7)/(3.9) are consistent with the definitions in (3.5)-(3.6), the subsequent algebra respects the identities KR=RK=R-I, and the analytic continuation from large r is legitimate provided the asserted real analyticity holds. The conditional verdict is therefore appropriate: accept the theorem conditional on the imported [MQR17] results, but do not take the abstract's universal phrasing as established. The acknowledgement about the flat-initial-data KdV error reinforces the need for this caution. I recommend no change to the reader's conditional verdict.","tokens_in":24705,"tokens_out":30396,"duration_ms":303635,"concrete_test":"Take a second model in the one-dimensional KPZ class whose 1:2:3 limit is not yet proven to be the TASEP fixed point but for which exact finite-time or conjectural limit formulas are available (for example, the KPZ equation with flat initial data, using the formulas discussed in [LDC12] and the correction in the acknowledgements). Compute φ=∂_r^2 log F and evaluate the residual of the scalar KP-II equation (1.7). A nonzero residual would falsify the abstract's universal reading; a zero residual would support it, though one model alone would not prove universality.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The theorem itself is internally well supported: the proof reduces to the differential relations (3.7)/(3.9), a long but checkable operator calculation, and an analytic-continuation argument from large r, and I do not see an internal inconsistency. The load-bearing gap is in moving from this theorem to the abstract and title. Theorem 1.1 is proved for the KPZ fixed point obtained as the 1:2:3 limit of TASEP. The only support for the broader statement is the sentence after (1.2): 'It is widely believed that this KPZ fixed point governs the limiting fluctuation for all models in the class.' That is a conjecture, not a theorem, and the paper adds no new universality input. If universality failed, 'KP governs random growth off a one dimensional substrate' would not follow for non-TASEP models, even though Theorem 1.1 would remain true for the TASEP-derived fixed point. The authors' own correction in the acknowledgements (a flat-initial-data KPZ generating function claimed in v1 to solve KdV was inconsistent with [LDC12]) underscores that the algebraic mechanism is delicate and is not automatically present in every finite-time KPZ-family formula. A smaller presentation issue: the scalar result is for φ=∂_r^2 log F, not for the logarithmic derivative ∂_r log F itself; the abstract's opening phrase is one derivative loose, although the body is precise.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that the logarithmic derivatives of the finite-dimensional distributions of the KPZ fixed point satisfy the matrix Kadomtsev-Petviashvili (KP) equation, and that in the one-point scalar case the second logarithmic derivative satisfies scalar KP-II. The proof starts from the Fredholm determinant formula for the KPZ fixed point obtained in [MQR17] as the 1:2:3 limit of TASEP, rewrites the kernel so that derivatives in t, x, and r act through the differential relations (3.7) and (3.9), computes the logarithmic derivative as tr Q, and then performs a lengthy but explicit operator-algebraic calculation to derive (1.6). Rigorous justification is by trace-class estimates for sufficiently large r followed by real analytic continuation. The paper also presents the Tracy-Widom distributions as self-similar solutions, formal initial data for escarpments, equations for Airy processes, and several KPZ-equation special solutions satisfying KP-II.","tokens_in":24980,"tokens_out":6827,"duration_ms":70789,"significance":"If the main theorem is correct, it establishes a new and surprising link between the integrable probability of the KPZ fixed point and integrable PDE theory. The derivation is honest and explicit: the relevant differential relations are stated, the analytic-continuation argument is described, and the paper flags precisely where rigorous support is imported from earlier work. The examples and the lower-tail heuristics are valuable. The main limitation is that the theorem is proved for the TASEP-derived KPZ fixed point, so the title and abstract's general statement about all random growth in one dimension rests on the unproven KPZ universality conjecture; the authors state this themselves. This does not undermine Theorem 1.1 but means the advertised scope should be qualified.","major_comments":[{"comment":"The general claim that KP governs random growth off a one-dimensional substrate is wider than what is proved. Theorem 1.1 is established for the KPZ fixed point defined through the TASEP 1:2:3 limit in [MQR17], and the only support for the universal statement is the sentence after (1.2): 'It is widely believed that this KPZ fixed point governs the limiting fluctuation for all models in the class.' Since universality is not proved here, the title and abstract overstate the result. This is not an internal inconsistency, but it is load-bearing for the advertised message. Please qualify the statements, for example by saying 'for the TASEP-derived KPZ fixed point' or 'for models in the class where the same fixed point has been proved.'","section":"Title, Abstract, and Section 1 after (1.2)"},{"comment":"The paper's language that distributions 'evolve according to' KP is stronger than the theorem. Theorem 1.1 is a pointwise differential identity for Q and q at t > 0; it does not establish well-posedness of the initial-value problem. The text after (1.9) explicitly says that well-posedness with escarpment initial data is left for future work, and the matrix initial data (1.10) is stated to be insufficient because 'the 0 and ∞ interact' and would need augmentation by 'some description of the rate of convergence.' These caveats should be reflected in the abstract or in the statement of the main claim so that 'evolve according to KP' is not read as a fully justified evolution statement.","section":"Section 1.1, Eqs. (1.9) and (1.10)"}],"minor_comments":[{"comment":"The abstract's first sentence says the logarithmic derivative evolves by KP, but the scalar equation (1.7) is for φ = ∂_r^2 log F, not for ∂_r log F itself. The body is precise; please correct the abstract wording.","section":"Abstract and Section 1, Eq. (1.7)"},{"comment":"The paper states that the new Airy-process equation and the Adler-van Moerbeke equation 'do not appear to be equivalent' and leaves the reconciliation for future work. This is acceptable, but a brief explanation of whether a discrepancy is suspected or merely a different form would help readers.","section":"Section 1.4, Eqs. (1.17) and (1.18)"},{"comment":"The phrase 'the ∞ looks formal' is ambiguous; likely 'the −∞ in the initial data looks formal' is intended. Please clarify.","section":"Section 1.1, after Eq. (1.9)"},{"comment":"The reference [MQR+] is listed as 'In preparation'; if an arXiv or published version exists, please update the citation so readers can access it.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main theorem is strong and the derivation is sound in outline. My recommendation of major revision is motivated by the gap between the advertised universal claim and the TASEP-derived proof, and by the formal status of the initial-value problem. These are fixable by qualification and do not cast doubt on Theorem 1.1 itself. The paper is appropriate for the journal and gives appropriate credit to prior work, including the Le Doussal-Calarabese correction."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The central theorem is real and the derivation holds up. What the paper proves is that the finite-dimensional marginals of the KPZ fixed point, starting from the MQR17 Fredholm determinant, satisfy matrix and scalar KP. The mechanism is clean: the fixed point kernel satisfies the differential relations (3.7)/(3.9), and once you have those, the Fredholm-determinant-to-KP machinery (Pöppe, going back to Zakharov–Shabat) does the rest. The new content is that the kernel actually evolves that way and that random growth is thereby connected to integrable systems; the Tracy-Widom distributions reappearing as self-similar solutions of KP is a satisfying payoff, not an input. There is no fitting anywhere. The proof is honest: the equation is checked for large r where trace norms are small, then extended by real analyticity, and the paper says which parts are imported from where. The acknowledgement of the v1 flat-initial-data error is a good sign; it also shows the mechanism is delicate, and the paper does not lean on it.\n\nSoft spots, in proportion. The title and the abstract's opening sentence claim more than the theorem: what is proved is that the TASEP-derived fixed point satisfies KP. The step from there to \"randomly fluctuating interfaces in one dimension\" is the standard universality conjecture, which the paper explicitly labels as widely believed rather than proved. That is a presentation issue, not a mathematical gap — the body is careful. The second soft spot is the initial data: the escarpment data (1.9) is formal, and the authors state plainly that the matrix initial data (1.10) is insufficient as written because the 0 and ∞ entries interact. Well-posedness of the PDE is left open. That does not threaten Theorem 1.1, but it does mean the PDE side of the story is less complete than the probability side. Minor items: the abstract says \"logarithmic derivative\" where the scalar result is for the second r-derivative; the relation to the Adler–van Moerbeke/Tracy–Widom equations for the Airy process is left unresolved; and the Section 3.3 computation is condensed enough that a referee should actually check the cancellations, especially in the matrix case.\n\nWho is this for: anyone working on KPZ universality or integrable probability. It deserves a serious referee. I would send it out, with a request that the authors soften the universality language and keep the honest caveats.","headline":"A genuinely new and well-supported connection between the KPZ fixed point and the KP equation; the derivation is sound and the main soft spot is the title's universality claim, which the body itself flags as conjectural.","tokens_in":25490,"tokens_out":5131,"would_cite":true,"duration_ms":53191,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K35","35Q53","37K10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The KP equation governs the marginals of random growth interfaces in one dimension.","keywords":["KPZ universality class","KPZ fixed point","Kadomtsev-Petviashvili equation","Fredholm determinants","Tracy-Widom distributions","Airy process","integrability","random growth"],"falsifier":"Compute the 1:2:3-scaled two-point distribution function $F$ of a growth model that is believed to lie in the KPZ universality class but is not known to converge to the same fixed point as TASEP, and check numerically or exactly whether $\\partial_r^2 \\log F$ satisfies the scalar KP-II equation (1.7); a nonzero residual would falsify the claim that KP governs random growth off a one-dimensional substrate.","tokens_in":24486,"feed_emoji":"📈","tokens_out":7362,"duration_ms":69154,"temperature":0.7,"pith_summary":"The paper establishes that the logarithmic derivative of the finite-dimensional marginals of the KPZ fixed point—the conjectural universal scaling limit of one-dimensional random growth—satisfies the Kadomtsev–Petviashvili (KP) equation. Concretely, for an $n$-point height distribution $F$, the matrix $Q$ built from the Fredholm determinant kernel obeys the matrix KP equation, and $D_r \\log F$ equals $\\operatorname{tr} Q$; for one-point marginals the equation reduces to scalar KP-II. This is shown by algebra from the kernel of the Fredholm determinant for the KPZ fixed point, without a physical heuristic. As corollaries, the Tracy–Widom GUE and GOE distributions appear as self-similar solutions of KP/KdV, and several known exact solutions of the KPZ equation also satisfy KP-II. The result gives the scaling limit a closed PDE structure that it was not known to have.","feed_headline":"KP equation governs random growth distributions","feed_subtitle":"Finite-dimensional marginals of the KPZ fixed point satisfy a matrix KP equation, tying 1D growth to integrable systems.","key_machinery":"The load-bearing object is the extended Brownian scattering operator for the KPZ fixed point, defined through Brownian motion killed when it hits the hypograph of the initial height, conjugated by the Airy unitary group $U_t=e^{-t\\partial^3/3}$. Its kernel $K$ obeys three differential relations—$D_r K=(D_1+D_2)K$, $\\partial_t K=-\\tfrac13(D_1^3+D_2^3)K$, and $D_x K=(D_2^2-D_1^2)K$—together with an integration-by-parts identity $[A][B]=-[A D_1 B + D_2 A B]$ for matrix entries evaluated at $(0,0)$. These relations are what force $Q=[(I-K)^{-1}K]$ to satisfy the matrix KP equation; the scalar version emerges by taking traces. The same relations also characterize which Fredholm determinants are KP tau functions, connecting the stochastic growth problem to the KP hierarchy.","core_discovery":"On the paper's own terms, the central discovery is Theorem 1.1: let $K$ be the shifted extended Brownian scattering kernel for the KPZ fixed point, $R=(I-K)^{-1}$, and $Q=[RK]$ the $n\\times n$ matrix of $RK$ evaluated at $(0,0)$; then $Q$ and $q=D_r Q$ solve the matrix KP equation $\\partial_t q + \\tfrac12 D_r q^2 + \\tfrac1{12} D_r^3 q + \\tfrac14 D_x^2 Q + \\tfrac12[q,D_x Q]=0$, and the logarithmic derivative of the $n$-point distribution is $D_r \\log F = \\operatorname{tr} Q$. In the one-point case, $\\varphi=D_r^2 \\log F$ solves the scalar KP-II equation (1.7). The authors emphasize that the result was unexpected and follows essentially by algebra from the kernel's differential relations; they also recover the GUE and GOE Tracy–Widom laws as self-similar solutions and show that several explicit KPZ-equation solutions (narrow wedge, spiked, and two-sided Brownian initial data) satisfy KP-II.","pith_inferences":["Editorial inference: because the proof uses only the three kernel differential relations, any determinantal stochastic process whose kernel satisfies those relations will automatically produce KP solutions; this gives a testable algebraic criterion for discovering new integrable structures in random growth.","Editorial inference: the emergence of matrix KP suggests the full finite-dimensional transition probabilities of the KPZ fixed point may be tau functions of the KP hierarchy, in which case higher-order hierarchy equations would impose additional constraints on multipoint distributions beyond (1.6).","Editorial inference: if the broad universality claim holds, the KP equation becomes a practical numerical tool for approximating growth-model distributions at large scales—solving KP with escarpment initial data could yield predictions for tails and correlations that are currently obtained only from Fredholm determinant expansions."],"forward_implications":["The GUE and GOE Tracy–Widom distributions appear as self-similar solutions of KP/KdV, so the KP equation supplies the 1:2:3 scaling invariance that the Tracy–Widom laws themselves lack.","The multipoint distribution of the Airy$_2$ process satisfies an explicit PDE (Theorem 1.9), giving a partial answer to the longstanding question of whether such a closed equation exists.","For flat initial data, the integrated matrix KdV equation governs the multipoint Airy$_1$ distribution, placing the flat case in the same integrable-systems framework.","Several known exact solutions of the KPZ equation, including narrow-wedge, spiked/half-Brownian, and two-sided Brownian initial data, are shown to produce KP-II solutions for their logarithmic derivatives.","The lower tails of the Tracy–Widom distributions can be recovered directly from the Burgers part of KP, which dominates in that regime."],"supporting_citations":[{"why":"Supplies the Fredholm determinant formula for the KPZ fixed point transition probabilities that the paper differentiates to obtain the KP equation.","marker":"[MQR17]"},{"why":"Provides the one-dimensional precedent that Fredholm determinants of kernels satisfying suitable differential relations solve the Hirota/KP equations.","marker":"[Po89]"},{"why":"Source of the kernel differential relations that lead to scalar KP-II, as the paper notes was rediscovered several times.","marker":"[ZS74]"},{"why":"Identifies the Tracy–Widom GUE distribution, which the paper recovers as a self-similar solution of KP.","marker":"[TW94]"},{"why":"Identifies the Tracy–Widom GOE distribution, which the paper recovers as a self-similar solution of KdV.","marker":"[TW96]"},{"why":"Provides the Fredholm determinant for the narrow-wedge KPZ generating function used in Example 2.1.","marker":"[ACQ11]"},{"why":"Gives the spiked/half-Brownian KPZ generating function formula used to show KP-II in Examples 2.2 and 2.4.","marker":"[BCF14]"},{"why":"Provides the two-sided Brownian modified generating function used in Example 2.3.","marker":"[BCFV15]"},{"why":"Previous PDE for the Airy process multipoint distribution, compared with Theorem 1.9.","marker":"[AM05]"}],"fun_headline_variants":["Matrix KP equation emerges from KPZ fixed point","1D random growth distributions obey KP equation","Surprise: KP equation governs growth marginals","KPZ fixed point solves matrix KP equation","Logarithmic derivatives of KPZ obey KP equation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The broad statement that KP governs random growth off a one-dimensional substrate relies on the unproven universality of the KPZ fixed point: the theorem is proved for the fixed point obtained as the 1:2:3 scaling limit of TASEP, and if some model in the class has a different large-scale limit, the general claim does not follow from this derivation.","fun_headline_variants_meta":{"raw":{"variants":["Matrix KP equation emerges from KPZ fixed point","1D random growth distributions obey KP equation","Surprise: KP equation governs growth marginals","KPZ fixed point solves matrix KP equation","Logarithmic derivatives of KPZ obey KP equation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000242,"raw_usage":{"total_tokens":1498,"prompt_tokens":891,"completion_tokens":607,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":507,"completion_tokens_details":{"reasoning_tokens":537}},"tokens_in":507,"tokens_out":607,"duration_ms":7133,"temperature":1.0,"reasoning_tokens":537,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:46:08.124196+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the 1:2:3-scaled two-point distribution function $F$ of a growth model that is believed to lie in the KPZ universality class but is not known to converge to the same fixed point as TASEP, and check numerically or exactly whether $\\partial_r^2 \\log F$ satisfies the scalar KP-II equation (1.7); a nonzero residual would falsify the claim that KP governs random growth off a one-dimensional substrate.","supporting_citations":[],"review_version":1}