{"id":"99ef5c79-1189-416c-95ea-d330c821631e","arxiv_id":"1908.08946","paper_version":3,"verdict":"REJECT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The Karman-Howarth and Corrsin closures are claimed to be K=u^3 sqrt((1-f)/2) f' and G=u theta^2 sqrt((1-f)/2) f_theta', obtained from the Liouville theorem plus an assumed uniform distribution of longitudinal velocity increments.","lead":"This paper proposes a derivation of non-diffusive closures for the von Karman-Howarth and Corrsin equations of isotropic turbulence, using the Liouville theorem and an assumed statistical independence between velocity and particle separation. The formulas are compact and parameter-free, but they rest on an unproven assumption that velocity increments are uniformly distributed, so the derivation is not established.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (26)'s uniform law for U contradicts Eq. (10): homogeneity gives ⟨U⟩=0, while Eq. (27) requires a positive mean, so the 1/2 factor entering the closures (46) is unsupported.","rationale":"The reader's weakest_assumption identifies the correct load-bearing point, and the problem is even sharper than 'real turbulence increment pdfs are not flat.' As defined in Eq. (10), U is the longitudinal velocity increment at separation xi. Ensemble averaging under the paper's own homogeneity hypothesis gives ⟨U⟩ = 0 identically, while Eq. (27) gives a positive mean whenever f(r) < 1. This is an internal inconsistency with the definition of U, not merely an inaccurate phenomenological choice. The subsequent identity ⟨U^2⟩ = 2u^2(1-f) in Eq. (45) is the standard structure-function relation, but it does not repair the mean. The closure formulas (46) depend linearly on ⟨U⟩ through Eq. (44); replacing the unjustified 1/2 with the homogeneous value 0 would make K = G = 0. The paper's limitation paragraph restricts the results to fully developed chaos, but it does not remove the contradiction, since fully developed homogeneous isotropic turbulence still satisfies ⟨U⟩ = 0. I also examined the route through Eqs. (38)-(44): the Liouville theorem supplies the form ⟨P_xi ∂f/∂xi · dot(xi)⟩ but not the uniform law; the uniform law is an extra assumption inconsistent with the defining kinematics. A conditional interpretation would require an explicit conditioning and would invalidate the use of the unconditional f(r) in Eq. (45). Thus the central claim, that Eq. (46) is rigorously derived from Liouville and statistical independence, is not established. The reader's REJECT verdict is supported; no verdict adjustment is needed.","tokens_in":16539,"tokens_out":10365,"duration_ms":112897,"concrete_test":"In a 256^3 forced DNS of homogeneous isotropic turbulence, compute for several separations r the ensemble average of U = (u(t,x+r)-u(t,x))·r/r and the longitudinal correlation f(r). Homogeneity predicts ⟨U⟩ = 0 to statistical precision, whereas Eq. (27) predicts u sqrt((1-f)/2) > 0. Additionally, histogram U at fixed r: Eq. (26) requires a flat distribution on (-U_S/2, U_S), whereas DNS shows a non-flat, zero-mean distribution. If the measured mean is zero and the histogram is non-flat, the central coefficient of the closures is falsified. An analytical Gaussian field with the same f(r) suffices to demonstrate the same contradiction without DNS.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction collapses at Eqs. (26)-(27). The paper asserts that U = dot(xi)·xi/xi = (u(t,x+xi)-u(t,x))·xi/xi is uniformly distributed on (-U_S/2, U_S), whence ⟨U⟩_xi = 1/2 sqrt(⟨U^2⟩_xi). But U is exactly the longitudinal velocity increment. Under the paper's own hypothesis of homogeneous turbulence, ensemble-averaging the definition (10) gives ⟨U⟩ = (⟨u(t,x+xi)⟩ - ⟨u(t,x)⟩)·xi/xi = 0 for every fixed xi, and therefore also for the isotropic P_xi average. Equation (27), by contrast, gives ⟨U⟩ = u sqrt((1-f)/2) > 0 whenever f(r) < 1. Thus the uniform law contradicts Eq. (10), not just real-turbulence phenomenology. Isotropy makes the direction xi/|xi| uniform on the sphere; it does not make the increment amplitude U uniform. A conditional interpretation restricted to diverging material lines is never defined, and it is incompatible with Eq. (45), which fixes ⟨U^2⟩ using the unconditional f(r). Because the closure (46) is linear in ⟨U⟩ through Eq. (44), replacing the unjustified coefficient 1/2 with the correct homogeneous value 0 would make K = G = 0. The claimed rigorous derivation from Liouville and statistical independence therefore fails at this step.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript claims to close the von Kármán–Howarth and Corrsin equations for homogeneous isotropic turbulence using the Liouville theorem together with an assumed statistical independence between the velocity/temperature fields and the material separation vector. It derives non-diffusive closure formulas, Eqs. (46), in which the triple correlations are expressed as products of the longitudinal correlation gradient and a scale-dependent propagation speed c_T = u sqrt((1-f)/2). The paper also analyzes Loitsianskii, Saffman–Birkhoff, and Corrsin invariants and the corresponding decay laws. The central derivation rests on the uniform distribution of the longitudinal velocity increment U, Eq. (26), and on the factorization P = F P_xi, Eq. (12).","tokens_in":16916,"tokens_out":4228,"duration_ms":47124,"significance":"If correct, the result would be significant: it would provide parameter-free, non-diffusive closures for the two classic two-point correlation equations, with testable consequences for the energy cascade and for invariant-based decay laws. The manuscript is clearly organized, states its assumptions explicitly, and the algebraic passage from Eq. (38) to Eq. (46) is internally consistent. However, the derivation fails at a load-bearing point: the assumed uniform law for U contradicts the definition of U together with homogeneity, and the closure formulas are essentially a restatement of that unsupported distributional ansatz. Because the central claim rests on this inconsistency, the paper cannot be accepted in its present form.","major_comments":[{"comment":"The uniform distribution of U contradicts the definition of U in Eq. (10) together with the homogeneity hypothesis. U is the longitudinal velocity increment (u(t,x+xi)-u(t,x))·xi/|xi|. Under homogeneity, the expectation of u(t,x+xi) with respect to F equals the expectation of u(t,x), so <U> = 0 for every fixed xi, and therefore also after averaging with P_xi. Equation (27) instead gives <U>_xi = (1/2) sqrt(<U^2>_xi) > 0 whenever f(r) < 1. Thus Eq. (26) is not a consequence of isotropy and incompressibility; isotropy makes the direction xi/|xi| uniform on the sphere, not the amplitude U uniform. Since the closure (46) is linear in <U>_xi through Eq. (44), replacing the unjustified coefficient 1/2 with the homogeneity-required value 0 would give K = G = 0. This is an internal inconsistency, not merely a disagreement with empirical turbulence phenomenology.","section":"Section 3, Eqs. (26)-(27)"},{"comment":"The derivation of the support (-U_S/2, U_S) does not justify uniformity. Equation (17) asserts <dV/dt>_xi = 0 for finite separation vectors, but incompressibility only guarantees dV/dt = 0 in the infinitesimal-volume limit, not for finite xi; the representation in Eq. (20) with Ak and epsilon is itself an additional ansatz. Even if the proposed interval were accepted, a bounded support does not imply a flat distribution on that interval, so the argument does not establish Eq. (26).","section":"Section 3, Eqs. (17)-(24)"},{"comment":"The closure formula is algebraically forced by the uniform ansatz rather than derived from independent statistical principles. Substituting Eq. (45) into Eq. (27) converts Eq. (44) into Eq. (46); the square-root factor (1-f)/2 is exactly the content of the assumed uniform distribution. The paper offers no benchmark, consistency test, or independent argument that would distinguish this distribution from other laws with the same second moment, so the claimed proof from Liouville's theorem and statistical independence is not established.","section":"Section 5, Eqs. (44)-(46)"},{"comment":"The factorization P = F P_xi is asserted as the crucial hypothesis of fully developed turbulence, but no argument is given for the statistical independence of xi from u and theta. Since this assumption is load-bearing for Eqs. (38)-(43), the statement in the Introduction and Conclusion that the present proof is 'more general and rigorous' than the earlier Lyapunov-based derivations is unsupported; the present derivation swaps one unproven ansatz for another.","section":"Section 3, Eq. (12)"}],"minor_comments":[{"comment":"The phrase 'exact relationship' used for Eq. (32) overstates the status of that relation, since it is obtained only under the delta-distribution model (13) and the fully developed chaos hypothesis.","section":"Abstract and Introduction"},{"comment":"There are several typographical issues, including 'litarature' in the Introduction and the malformed email address 'nicola.dedivitiis@uni roma1.it'.","section":"Throughout"},{"comment":"The skewness H3 is said to be in agreement with literature data, but no comparison plot or quantitative benchmark is shown in this manuscript; the reader is referred to previous works, which limits the self-containedness of the validation claim.","section":"Section 5, Eq. (48)"},{"comment":"The asymptotic conditions m > 4, n > 2, and m > 2 are stated as conditions for invariance, but the interplay with the arbitrary constants c_u and c_theta in Eq. (52) is only sketched; a more explicit derivation of the boundary-term vanishing would improve clarity.","section":"Section 6, Eqs. (53)-(55)"}],"recommendation":"reject","confidential_remarks":"The central contradiction at Eqs. (26)-(27) is fatal for the proposed closure, and it is not repairable within the scope of this manuscript because the correct homogeneous value <U>=0 would nullify the derived K and G. I would not recommend further review of this version. The paper also leans heavily on the author's previous works for validation, but the technical flaw is independent of the citation pattern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper's new proof of the von Kármán–Howarth and Corrsin closures breaks at Eq. (26), where the longitudinal velocity increment U is asserted to be uniformly distributed on (-U_S/2, U_S). That distribution has positive mean, but under the paper's own homogeneity assumption the ensemble average of U is exactly zero for every fixed separation r. This is an internal contradiction with Eq. (10), and since the closure formulas are linear in ⟨U⟩, it is fatal to the claimed rigorous derivation.\n\nWhat is genuinely new here is modest. The closures K = u^3 sqrt((1-f)/2) f' and G = u θ^2 sqrt((1-f)/2) f_θ' were already published in the author's earlier Lyapunov-based papers; the paper states this openly. The claimed novelties are a Lyapunov-free route through the Liouville theorem and an exact relation, Eq. (32), between correlations computed with the material-line distribution and with the velocity distribution. That relation is essentially an orientation average, and it is reasonable under isotropy. The algebra from Eq. (38) to Eq. (46) is internally consistent, and the invariant analysis in Section 6 is a plausible extension conditional on the closure.\n\nThe soft spot is Eq. (26). The uniform law does not follow from isotropy and incompressibility. Isotropy only makes the direction ξ/|ξ| uniform on the sphere; it does not make the amplitude of (u(x+ξ)-u(x))·ξ/ξ uniform. Real turbulence increment pdfs are not flat. The deeper problem is sign: the uniform distribution has positive mean, while homogeneity forces ⟨U⟩=0. The prose about trajectory divergence and incompressibility is not a defined conditional average, and Eq. (45) fixes ⟨U²⟩ using the unconditional f(r). So the derivation is not salvageable as written. Replacing the 1/2 factor with the homogeneous value 0 makes K and G vanish.\n\nTo give credit: the paper is clearly written, the author is transparent about coincidence with earlier work and about the limits of validity. The closure formulas themselves have had some empirical success—they reproduce the -3/7 derivative skewness and a reasonable Kolmogorov constant in previous tests. But the present proof does not establish them.\n\nFor peer review: I would send it out. The problem is important, the author is serious, and a referee could push for a clear statement of the statistical assumption for U, perhaps reframing it as an empirical closure rather than a theorem. Expect major revision or rejection. I would not cite it as a proof, though I might mention the closure as an empirical model. For a reading group, it is a useful case study in how a distributional assumption can silently carry an entire derivation.","headline":"The new derivation fails because the assumed uniform law for the longitudinal increment contradicts homogeneity; the closure formulas are left as empirical models, not derived results.","tokens_in":17377,"tokens_out":4147,"would_cite":false,"duration_ms":38769,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["47.27.Gs"],"model":"deepseek-v4-flash","headline":"This paper derives parameter-free, non-diffusive closure formulas for the von Kármán–Howarth and Corrsin equations, showing that the energy cascade is a propagation of correlations at a scale-dependent speed.","keywords":["homogeneous isotropic turbulence","von Kármán-Howarth equation","Corrsin equation","closure problem","Liouville theorem","energy cascade","correlation propagation","self-similar decay"],"falsifier":"Measure the probability density of the longitudinal velocity increment at a fixed separation $r$ in high-Reynolds-number homogeneous isotropic turbulence, in an experiment or a direct numerical simulation. If the pdf is not flat over its support, or if $\\langle U\\rangle \\neq \\tfrac{1}{2}\\sqrt{\\langle U^2\\rangle}$, then the closure formulas (46) fail quantitatively.","tokens_in":16330,"feed_emoji":"🌊","tokens_out":9310,"duration_ms":82953,"temperature":0.7,"pith_summary":"This paper aims to close the von Kármán–Howarth and Corrsin equations, the evolution equations for longitudinal velocity and temperature correlations in homogeneous isotropic turbulence, without eddy-viscosity assumptions. Using the Liouville theorem and a statistical-independence hypothesis between the velocity field and the material separation vectors, it derives algebraic formulas that express the triple-correlation transfer terms in terms of the second-order correlations and their gradients. The resulting closures are non-diffusive and parameter-free; they describe the energy cascade as a spatial propagation of correlations at speed $c_T = u\\sqrt{(1-f)/2}$. The paper then shows that, under these closures, the Loitsianskii, Saffman–Birkhoff, and Corrsin invariants exist under stated decay conditions, and that self-similar decay can be described by the fluctuating intensities and correlation scales alone. A sympathetic reader would care because the derivation offers a first-principles route to closing two central equations of turbulence theory, free of fitted constants.","feed_headline":"Turbulence cascade is correlation propagation, not diffusion","feed_subtitle":"A Liouville-theorem derivation closes the von Kármán–Howarth and Corrsin equations with no free parameters.","key_machinery":"The load-bearing machinery is the factorization $P=F\\,P_\\xi$ of the joint distribution into a velocity–temperature part and a material-separation part, together with the Liouville equation for $P_\\xi$. With the separation vector concentrated on the sphere $|\\xi|=r$ by the strongly peaked distribution (13), the longitudinal velocity difference $U=\\dot\\xi\\cdot\\xi/\\xi$ is argued to be uniformly distributed on $(-U_S/2,\\,U_S)$ (Eq. 26), giving $\\langle U\\rangle_\\xi = \\tfrac{1}{2}\\sqrt{\\langle U^2\\rangle_\\xi}$. That identity, combined with $\\langle U^2\\rangle_\\xi=2u^2(1-f)$, converts the transfer terms into the closure formulas (46). The paper's claimed novelty is the exact relation (32) connecting spatial correlations computed with the velocity distribution function to those computed with the separation-line distribution function.","core_discovery":"On the paper's own terms, the central discovery is that the two unclosed transfer terms in the von Kármán–Howarth and Corrsin equations can be written exactly as $K(r)=u^3\\sqrt{(1-f)/2}\\,\\partial f/\\partial r$ and $G(r)=u\\theta^2\\sqrt{(1-f)/2}\\,\\partial f_\\theta/\\partial r$, where $f$ and $f_\\theta$ are the longitudinal velocity and temperature correlations and $u$ and $\\theta$ are their r.m.s. fluctuations. These formulas are derived, not postulated, from the Liouville equation for the joint distribution of velocity, temperature, and material separation, under the hypotheses of full statistical independence of the rapidly fluctuating separation vectors from the velocity and temperature fields, together with homogeneity, isotropy, and incompressibility. They contain no empirical constants and no second derivatives of the correlations, so the cascade is not a diffusion process but a wave-like spatial propagation with local speed $c_T = u\\sqrt{(1-f)/2}$. The paper further claims that these closures reproduce the negative skewness of velocity differences, Kolmogorov's law, and scalar spectra, and that they lead to clean conditions for the Loitsianskii and Saffman–Birkhoff invariants.","pith_inferences":["Editorial inference: the same Liouville-based route, if valid, should in principle produce analogous closures for anisotropic or inhomogeneous two-point statistics whenever a separation-vector distribution can be defined; the paper explicitly restricts itself to homogeneous isotropic turbulence.","Editorial inference: the flat-distribution assumption (26) is directly testable against measured or simulated probability densities of longitudinal velocity increments at fixed separations; a non-flat pdf would change the factor $1/2$ and hence shift every prediction quantitatively.","Editorial inference: if the equivalence (32) holds, surface-averaged second-order structure functions computed on spheres of radius $r$ must equal ensemble-averaged ones in fully developed isotropic turbulence, a quantitative prediction that simulations could check.","Editorial inference: the closure's success at reproducing known inertial-range results suggests it might also apply to other transported passive scalars, with the same square-root factor carried by the relevant correlation."],"forward_implications":["The von Kármán–Howarth and Corrsin equations become closed partial differential equations with no free empirical constants.","The energy cascade is reinterpreted as a spatial propagation of correlations at scale-dependent speed $c_T=u\\sqrt{(1-f)/2}$, rather than as a diffusive process.","The longitudinal velocity-difference skewness is fixed at $H_3(0)=-3/7$, which the paper reports as being in good agreement with prior measurements and simulations.","Under self-similarity and the invariant conditions, decay reduces to ordinary differential equations for $u$, $\\lambda_T$, $\\theta$, and $\\lambda_\\theta$, giving explicit power laws such as $u^2\\propto (1+4\\nu t/\\lambda_T^2(0))^{-5/2}$ when the Loitsianskii invariant holds.","The same closures imply Kolmogorov-like inertial-range behavior and scalar spectra consistent with Kolmogorov, Obukhov–Corrsin, and Batchelor scalings."],"supporting_citations":[{"why":"Defines the von Kármán–Howarth equation that this paper closes.","marker":"[1]"},{"why":"Provides the standard framework of homogeneous turbulence and the energy balance behind the transfer term.","marker":"[2]"},{"why":"States the Corrsin equation for temperature correlations that receives the $G$ closure.","marker":"[3]"},{"why":"Gives the temperature-spectrum context and the triple correlation terms that must be closed.","marker":"[4]"},{"why":"Prior derivation of the same velocity closure via finite-scale Lyapunov analysis; this paper reproduces it without invoking Lyapunov theory.","marker":"[29]"},{"why":"Prior derivation of the temperature closure; the present $G$ formula coincides with it.","marker":"[30]"},{"why":"Earlier Lagrangian formulation whose proof the present Liouville-only route is meant to corroborate.","marker":"[31]"},{"why":"The author's review of the statistical Lyapunov theory, used for the self-similarity and Kolmogorov-constant claims.","marker":"[45]"},{"why":"Provides the uniform-distribution result for finite-scale Lyapunov exponents that motivates Eq. (26).","marker":"[57]"},{"why":"Analysis showing the Loitsianskii integral need not be invariant, used to frame the paper's invariant conditions.","marker":"[64]"}],"fun_headline_variants":["Turbulence cascade is correlation propagation, not diffusion","Liouville theorem closes turbulence equations with zero parameters","Cascade is a wave: exact closures from Liouville theorem","No diffusion: turbulence correlations propagate, new closure"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that, for each fixed separation distance, the longitudinal velocity difference has a flat probability distribution over its possible range, which produces the factor $1/2$ entering the closure formulas; if actual increment distributions are not flat, the central formulas fail.","fun_headline_variants_meta":{"raw":{"variants":["Turbulence cascade is correlation propagation, not diffusion","Liouville theorem closes turbulence equations with zero parameters","Cascade is a wave: exact closures from Liouville theorem","No diffusion: turbulence correlations propagate, new closure"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000414,"raw_usage":{"total_tokens":2203,"prompt_tokens":1071,"completion_tokens":1132,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":687,"completion_tokens_details":{"reasoning_tokens":1070}},"tokens_in":687,"tokens_out":1132,"duration_ms":9302,"temperature":1.0,"reasoning_tokens":1070,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:27:14.393843+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the probability density of the longitudinal velocity increment at a fixed separation $r$ in high-Reynolds-number homogeneous isotropic turbulence, in an experiment or a direct numerical simulation. If the pdf is not flat over its support, or if $\\langle U\\rangle \\neq \\tfrac{1}{2}\\sqrt{\\langle U^2\\rangle}$, then the closure formulas (46) fail quantitatively.","supporting_citations":[{"cited_title":", On the Statistical Theory of Isotropic Turbulence., Proc","cited_arxiv_id":null,"evidence_quote":"Defines the von Kármán–Howarth equation that this paper closes."},{"cited_title":", Statistics of finite scale local Lyapunov exponents in fully developed homogeneous isotropic turbulence, Advances in Math- ematical Physics , vol","cited_arxiv_id":null,"evidence_quote":"Provides the uniform-distribution result for finite-scale Lyapunov exponents that motivates Eq. (26)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Analysis showing the Loitsianskii integral need not be invariant, used to frame the paper's invariant conditions."}],"review_version":1}