{"id":"45aafd9d-6fb2-4f62-b82c-d69587c694c1","arxiv_id":"2507.14524","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"WPTS, a wave-particle hybrid turbulence model, reproduces round-jet statistics at Re=5000 on a grid with 2% of DNS cells, matching DNS and experimental data.","lead":"This paper tests a wave-particle hybrid method for turbulence on a round jet at Reynolds number 5000, using a grid with only 2% of the cells a direct simulation would need. It reports close agreement with DNS and experiments for the jet's spreading, mean velocity, and Reynolds stresses, hinting at a cheaper path for engineering turbulent flow simulations.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2%-grid accuracy claim is not causally attributed: no same-grid particle-free or standard-LES baseline is provided, so the wave-particle closure's role in the successful jet prediction is untested.","rationale":"Read in good faith, this is a competent validation letter: one WPTS configuration at Re=5000, Ma=0.6 on 84^3 cells gives Bu=5.69 versus DNS 5.50 and experiments 5.80/6.06, collapsed mean profiles, and reasonable second-order statistics. That is genuine positive evidence that the package works for this test case. My concern is interpretational rather than accusatory: the abstract's claim is that the wave-particle closure, with constants reused from [35], produces the 2%-grid accuracy. Because the wave solver is a high-order GKS and the inflow is strongly forced, the same accuracy might be obtained without particles or with a standard LES closure. The absence of an ablation or same-cost baseline leaves the causal claim undetermined. I therefore agree with the reader's CONDITIONAL verdict but shift the emphasis from closure-constant sensitivity to causal attribution; the constants sweep is useful but secondary, because even a successful sweep would not show that the particles are the active mechanism. I also flag for the authors that Eq. (18) appears to have the centerline decay ratio inverted (it should read Uc/Ue = Bu D/(x - x0u) in the standard convention), and Eq. (9) shows 1 - e^{Δt/τn} where the negative exponent is required; these typos should be corrected but are not the basis of my verdict. The proposed ablation test directly settles the load-bearing attribution issue.","tokens_in":10954,"tokens_out":13350,"duration_ms":170966,"concrete_test":"Run the identical jet case on the same 84^3 grid, same inlet perturbations, and same time horizon with (i) the stochastic particle component disabled (pure high-order GKS) and (ii) particles replaced by a standard Smagorinsky SGS model with the same C_s. Compare Bu from Eq. (18), mean velocity profiles, and Reynolds-stress profiles at x=25, 30, 35, and report the resolved versus particle-borne stress split. If the particle-free or Smagorinsky baseline matches DNS within the scatter of Figs. 5-6, the wave-particle closure's attribution fails; if both baselines are clearly worse, the central claim is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's central claim is that WPTS on 84^3 cells — about 2% of the DNS grid — reproduces round-jet mean flow and Reynolds stresses. For this claim to mean what it says, the wave-particle closure carried over from [35], with constants C_s^2=0.05, Ep=0.8, k=0.98 and C0=0.5, must be the active ingredient producing the agreement. The paper never tests that attribution. The wave component is a fifth-order WENO-AO gas-kinetic scheme (Sec. 2), which already provides implicit dissipation on coarse grids, and the inflow is strongly forced by dual-mode plus broadband perturbations (Sec. 3.2). Validation is exclusively against DNS/experiments; no coarse-grid baseline or particle ablation is run. Section 3.3.2 reports particle density but not the split of resolved versus particle-borne Reynolds stress, and the paper itself notes that far-field particles retained through Ep may have a non-negligible influence on the statistics. Without a same-cost baseline, the figures could reflect the wave solver and inlet forcing rather than the particle transport, so the specific claim that the wave-particle closure is responsible for the accuracy is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper applies the wave-particle turbulence simulation (WPTS) method to a spatially developing round jet at Re=5000 and Ma=0.6 on an 84^3 grid said to contain roughly 2% of the cells used in the reference DNS. The method couples a fifth-order WENO-AO gas-kinetic wave solver with stochastic fluid particles whose non-equilibrium transport is intended to model subgrid turbulent kinetic energy. The authors report a centerline velocity decay constant Bu=5.69, radial mean-velocity profiles, and anisotropic Reynolds stress profiles at three axial stations, all stated to collapse and agree with DNS and experimental data; they also present a sensitivity test on the inlet perturbation frequency ratio f.","tokens_in":11380,"tokens_out":5354,"duration_ms":66192,"significance":"If the central claim holds, the paper is a useful and rather striking demonstration: a coarse-grid wave-particle closure, with constants carried over from the authors' prior mixing-layer study rather than refit to the jet, reproduces the mean flow and second-order statistics of a spatially developing round jet. Strengths include the use of external DNS and experimental data for validation, a falsifiable quantitative prediction (Bu=5.69 versus literature values 5.50-6.06), and the adaptive spatial distribution of the particle component. The weaknesses identified below concern statistical and grid-convergence support, sensitivity of the closure constants, and attribution of the accuracy to the wave-particle mechanism as opposed to the underlying high-order wave solver and inlet forcing; these gaps currently prevent the 2%-grid efficiency claim from being fully established.","major_comments":[{"comment":"The quantitative comparisons, including Bu=5.69 and the Reynolds stress profiles in Figure 6, come from a single realization and a 23Te averaging window, with no confidence intervals, block-averaged error bars, or a second realization. The agreement with DNS (Bu=5.50) and experiments (5.80, 6.06) cannot be assessed as 'excellent' unless the sampling uncertainty of the present statistics is quantified; please report estimated statistical error bars on Bu and on the stress profiles, for example by block averaging over the 23Te window.","section":"§3.3.1, Table 1"},{"comment":"The central claim that the computation uses only 2% of the DNS cells is not accompanied by a grid-convergence or resolution-sensitivity study, and the reference DNS grid size is not stated. Without at least one additional grid level showing that Bu and the Reynolds stress profiles are insensitive to the mesh, the reported accuracy is a single-grid result rather than a demonstrated property of the method; the 2% efficiency figure therefore needs explicit support.","section":"§3.2, §3.3.1"},{"comment":"The turbulent collision time closure in Eq. (16) and the sampling constant C0=0.5 are load-bearing: they determine the wave-particle split, the particle lifetime, and hence the subgrid transport. The constants Cs^2=0.05, Ep=0.8, and k=0.98 are taken from the authors' prior study [35] with no sensitivity analysis in this paper. A small perturbation study of these constants (for example +/-20% in Ep and Cs^2) is needed to support the claim that the closure transfers to the round jet without recalibration.","section":"§2, Eq. (16)"},{"comment":"No baseline computation is reported that isolates the role of the wave-particle mechanism. The wave component is the fifth-order WENO-AO gas-kinetic scheme, which supplies its own implicit dissipation on coarse grids, and the inflow is strongly forced by dual-mode plus broadband perturbations (Section 3.2). A same-grid computation with the particle transport disabled, or with the particle flux set to zero, would show whether the observed accuracy is attributable to WPTS or to the wave solver and inlet conditioning; this attribution matters because the abstract's claim is specifically about WPTS.","section":"§2, §3.3.1"},{"comment":"The paper itself notes that low-density far-field particles retained through the Ep term may have a non-negligible influence on fluid evolution and turbulent statistics, but this influence is never quantified. Please report, at the three statistical stations x=25,30,35, the split of the resolved versus particle-borne contribution to the sampled Reynolds stresses, or run an Ep=0 sensitivity case, so that the reader can judge whether the reported statistics are robust to particle retention in the far field.","section":"§3.3.2"}],"minor_comments":[{"comment":"The sentence 'corresponding to approximately 2% that employed in DNS study [18]' should state the DNS grid size explicitly, since the 2% figure is central to the paper's efficiency claim.","section":"§3.2"},{"comment":"The notation DN(.,.) in Eq. (9) is not defined in this paper; please define the sampling operator or restate the formula from [35].","section":"§2, Eq. (9)"},{"comment":"The symbol ωp in Eq. (16) is not defined in the text; please state its meaning and range.","section":"§2, Eq. (16)"},{"comment":"The treatment of the particle transport time tf is confusing: Eq. (8) says sampled particles are evolved with tf=Δt, while Eq. (10) computes tf for surviving particles as min[-τn ln(η), Δt]. Please clarify the order of these operations and the definition of tf for newly sampled versus surviving particles.","section":"§2, Eqs. (8) and (10)"},{"comment":"The sensitivity test changes only the inlet frequency ratio f, but the virtual origin changes from x0u=3.44 to 1.34 while Bu changes only slightly; the discussion should state more explicitly that the inlet frequency primarily affects the virtual origin and transition region rather than the fully developed decay rate, as the manuscript currently gestures at this point.","section":"§3.3.3"},{"comment":"The closure details and all model coefficients are attributed to the arXiv preprint [35]; either cite a published version if available or include a short summary of the closure derivation so that the present paper is more self-contained.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper relies heavily on the authors' own preprint [35] for the closure model and constants; this is self-citation rather than circularity, because the jet validation uses external DNS and experimental data, but it does mean the model's prior validation is not yet peer-reviewed in the cited form. The missing grid-sensitivity and ablation studies are the main concerns; for a short validation letter the 23Te window may be acceptable, but the absence of any error estimate makes the quantitative claim difficult to verify. The editor may also wish to confirm that the companion preprint [35] is not being concurrently considered in a way that would duplicate the method description."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a straightforward method validation: WPTS, a wave-particle turbulence closure introduced in the authors' earlier mixing-layer preprint, is now applied to a spatially developing round jet at Re=5000. That is the genuinely new part, and it is worth having. The comparison against Sharan and Bellan's DNS and against Hussein and Panchapakesan's experiments covers mean velocity, centerline decay rate (Bu=5.69, between DNS 5.50 and experiments 5.80/6.06), and anisotropic Reynolds stresses, with profiles collapsing at three axial stations. The closure constants were carried over from the prior paper without refitting to jet data, which is a good sign for transferability.\n\nThe soft spots are real but they are gaps in validation, not signs of a wrong result. The central claim that 2% of DNS cells suffices is not causally attributed. There is no same-cost coarse-grid baseline (pure GKS or standard LES on the same grid), and no particle ablation, so the reader cannot tell whether the agreement comes from the particle transport model or from the fifth-order WENO-AO wave solver and the strong dual-mode inlet forcing. The paper itself notes in Section 3.3.2 that far-field particles retained through Ep may have a non-negligible influence on the statistics, which is honest but also a warning. Statistics come from a single realization with 23Te averaging; no confidence intervals, no grid-convergence study. The hand-set parameters (Cs^2=0.05, Ep=0.8, k=0.98, C0=0.5) are not varied; only the inlet frequency is tested. These are all addressable in revision.\n\nThe model itself is heuristic, but the evaluation is not circular: the predictions are tested against external DNS and experiments. The method is a plausible engineering-style LES alternative, not a new theory of turbulence. The citation pattern is normal for a methodological line, with self-citations to the prior WPTS and UGKWP papers, and the external references are appropriate.\n\nWho gets value from this: CFD researchers interested in cheap hybrid wave-particle closures for shear flows, and anyone assessing whether WPTS merits adoption. It deserves a serious referee, but the referee should push for a same-cost baseline and uncertainty quantification before the 2% claim is accepted as stated. My recommendation: send to peer review, with the expectation of major revision.\n\nReading group: maybe. I would not build on it without seeing the baseline, but it is a reasonable validation data point.","headline":"Useful round-jet validation of the WPTS closure, but the 2%-grid accuracy claim lacks a same-cost baseline that would show the particle component is actually doing the work.","tokens_in":11783,"tokens_out":1886,"would_cite":false,"duration_ms":24531,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["76F10","76F65"],"pacs":["47.27.wg"],"model":"deepseek-v4-flash","headline":"The wave-particle method WPTS reproduces the round jet at Re = 5000 — decay rate, mean velocity, Reynolds stresses — on a grid with only 2% of the cells DNS needs.","keywords":["wave-particle turbulence simulation","round jet","non-equilibrium transport","subgrid kinetic energy","self-similarity","Reynolds stresses","coarse-grid turbulence modeling","gas-kinetic scheme"],"falsifier":"Re-run the same jet on the same $84^3$ grid with the closure constants varied one at a time — for example $E_p = 0.6$ and $1.0$, $C_s^2 = 0.03$ and $0.10$, $C_0 = 0.3$ and $0.7$, $k = 0.95$ and $0.99$ — and recompute $B_u$ and the Reynolds-stress profiles at $x = 25, 30, 35$. If the decay rate or the stress levels move materially outside the DNS and experimental bands, the reported accuracy is tied to the chosen operating point rather than to the closure itself. A complementary check is to apply the identical constants to another spatially developing shear flow, such as a plane channel or a backward-facing step at $\\mathrm{Re} \\approx 5000$, and see whether the mean flow and second-order statistics still match reference data.","tokens_in":10766,"feed_emoji":"🌀","tokens_out":16283,"duration_ms":128088,"temperature":0.7,"pith_summary":"This letter applies the wave-particle turbulence simulation (WPTS) method to a spatially developing round jet at Reynolds number $\\mathrm{Re} = 5000$, and claims that on a grid with only 2% of the cells a direct numerical simulation requires, the method reproduces the jet's fully developed behaviour. The validation targets are the linear decay of centreline velocity, with decay constant $B_u = 5.69$ falling between the DNS value 5.50 and the experimental values 5.80 and 6.06; the collapse of mean axial and radial velocity profiles; and the anisotropic Reynolds-stress components. The broader point the authors are trying to establish is that one wave-particle closure, carried over without recalibration from their earlier mixing-layer study, captures shear-driven turbulence on coarse grids, so the result is not specific to one flow geometry. A sympathetic reader would care because, if the claim holds, engineering-scale jet flows that are out of reach for DNS become affordable on ordinary grids.","feed_headline":"Jet turbulence predicted on a grid using 2% of DNS cells","feed_subtitle":"Wave-particle closure yields the right decay rate and Reynolds stresses at Re=5000 without a finer mesh.","key_machinery":"The load-bearing object is the turbulent collision time $\\tau_t$ and the wave-particle split it governs. Its model, $e^{-\\Delta t/\\tau_t} = (1-\\omega_p)e^{-\\Delta t/\\tau_{mac}} + \\omega_p E_p$, blends a macroscale Smagorinsky-type relaxation time $\\tau_{mac}$ with a particle-retention fraction $E_p = 0.8$, using coefficients $C_s^2 = 0.05$ and $k = 0.98$; this model decides how much of each cell's fluid is sampled into stochastic particles, $W^{hp}_i = e^{-\\Delta t/\\tau_n} W^h_i$, and how long those particles free-stream before being reabsorbed. Sampled particles receive a velocity perturbation drawn from the local turbulent kinetic energy scaled by $C_0 = 0.5$, are transported under the pressure gradient, and are deleted with their mass, momentum, and energy merged back into the wave field. The $\\tau_t$ model therefore plays three roles at once: it determines where particles exist, how far they travel, and how much resolved kinetic energy is handed to the subgrid (particle) representation.","core_discovery":"The paper's central discovery is that turbulent flow can be split, at grid resolution, into a wave component that carries the resolved Navier-Stokes dynamics and a stochastic particle component that carries unresolved subgrid kinetic energy through free transport, with the split controlled by a modelled turbulent collision time; and that this decomposition delivers quantitative turbulence statistics on a very coarse grid. For a round jet at $\\mathrm{Re}_j = 5000$ and $Ma = 0.6$ on an $84^3$ grid, the method captures the linear centreline-velocity decay with $B_u = 5.69$, mean axial and radial velocity profiles that collapse onto DNS and experimental data at $x = 25, 30, 35$, and Reynolds-stress profiles in reasonable agreement, with only the cross-stress term slightly elevated. The decomposition is adaptive: where the grid resolves the flow, the particle field vanishes and WPTS reduces to the gas-kinetic scheme solving the Navier-Stokes equations, while in the shear layer near the jet exit particles appear in proportion to the resolved strain. Switching the inlet perturbation frequency ratio from $f = 2.40$ to $f = 2.22$ shifts the transition region but leaves the self-similar statistics essentially unchanged.","pith_inferences":["The particle-concentration field shown in the paper tracks the resolved shear — dense near the jet exit, thin downstream — which suggests WPTS particle density could double as a built-in indicator of where the grid fails to resolve the flow and could guide adaptive mesh refinement; the paper does not make this suggestion.","The '2% of DNS cells' comparison counts grid points, not cost: particle sampling, transport, and bookkeeping add overhead per cell, so a wall-clock or CPU-hour comparison against a well-tuned LES on the same grid would sharpen (or erode) the efficiency claim.","The four closure constants are carried from one flow to the next without recalibration, and the paper reports no sensitivity study; a systematic sweep over $E_p$, $C_s^2$, $k$, and $C_0$, or a third canonical flow such as a channel or backward-facing step, would reveal whether 2%-grid accuracy is a property of the closure or of its current operating point."],"forward_implications":["The same closure constants ($C_s^2 = 0.05$, $E_p = 0.8$, $k = 0.98$, $C_0 = 0.5$) that worked for the mixing layer also work for the round jet, which is the evidence that the method's coarse-grid accuracy is not specific to one flow geometry.","For this jet, WPTS yields a centerline decay constant $B_u = 5.69$, bracketed by the DNS value 5.50 and the experimental values 5.80 and 6.06; the model therefore sits inside the spread of established data without recalibration.","The collapse of mean-velocity and Reynolds-stress profiles at three downstream stations supports using WPTS to study the self-similar region of round jets on coarse grids.","Because WPTS reverts to the gas-kinetic scheme wherever the grid resolves the flow, a coarse-grid simulation needs no global decision about where turbulence modelling applies; the particle field appears and disappears locally.","The near-insensitivity of the self-similar statistics to the inlet frequency ratio ($f = 2.40$ vs $2.22$) indicates the reported decay rate and stresses are not an artifact of the chosen forcing."],"supporting_citations":[{"why":"Introduces the WPTS method and its turbulent-collision-time model; supplies the closure constants ($C_s^2 = 0.05$, $E_p = 0.8$, $k = 0.98$, $C_0 = 0.5$) that the jet simulation inherits unchanged, so without it the present simulation has no closure.","marker":"[35]"},{"why":"The direct numerical simulation at $\\mathrm{Re}_j = 5000$ that defines the 2%-grid cell-count baseline and supplies the reference decay rate, mean-velocity profiles, Reynolds stresses, and the hyperbolic-tangent inflow profile adopted here.","marker":"[18]"},{"why":"The experiment supplying the benchmark centreline decay constant $B_u = 5.80$, against which the predicted $B_u = 5.69$ is compared.","marker":"[7]"},{"why":"The experiment supplying $B_u = 6.06$ and the measured mean-velocity and fluctuation profiles used as comparison data.","marker":"[16]"},{"why":"Supplies the dual-mode (axisymmetric plus helical) inlet perturbation with $St_D = 0.5$ and $f = 2.40$ used to trigger transition, and the alternative frequency $f = 2.22$ used in the sensitivity test.","marker":"[6]"},{"why":"The unified gas-kinetic wave-particle framework whose coupled wave-particle evolution and free-transport flux treatment the WPTS method is built upon.","marker":"[13]"},{"why":"The gas-kinetic BGK scheme that WPTS automatically reverts to in well-resolved cells, which guarantees recovery of the Navier-Stokes solution in laminar or resolved regions.","marker":"[28]"}],"fun_headline_variants":["Wave-particle turbulence model nails jet on 2% of DNS cells","Jet turbulence predicted at 2% DNS grid cost via wave-particle split","Wave-particle closure gives jet physics on 2% of DNS cells","Round jet turbulence captured on coarse grid with wave-particle model","Wave-particle method mimics jet turbulence at 2% DNS grid"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is a single hand-set formula for the turbulent collision time, $e^{-\\Delta t/\\tau_t} = (1-\\omega_p)e^{-\\Delta t/\\tau_{mac}} + \\omega_p E_p$ with $C_s^2 = 0.05$, $E_p = 0.8$, $k = 0.98$, and sampling constant $C_0 = 0.5$, carried over unchanged from the authors' earlier mixing-layer study; if this closure is not a general property of turbulence but is tuned to that flow, the 2%-grid accuracy reported for the jet will not transfer to other configurations.","fun_headline_variants_meta":{"raw":{"variants":["Wave-particle turbulence model nails jet on 2% of DNS cells","Jet turbulence predicted at 2% DNS grid cost via wave-particle split","Wave-particle closure gives jet physics on 2% of DNS cells","Round jet turbulence captured on coarse grid with wave-particle model","Wave-particle method mimics jet turbulence at 2% DNS grid"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000742,"raw_usage":{"total_tokens":3353,"prompt_tokens":1029,"completion_tokens":2324,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":645,"completion_tokens_details":{"reasoning_tokens":2235}},"tokens_in":645,"tokens_out":2324,"duration_ms":474021,"temperature":1.0,"reasoning_tokens":2235,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:53:44.071821+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the same jet on the same $84^3$ grid with the closure constants varied one at a time — for example $E_p = 0.6$ and $1.0$, $C_s^2 = 0.03$ and $0.10$, $C_0 = 0.3$ and $0.7$, $k = 0.95$ and $0.99$ — and recompute $B_u$ and the Reynolds-stress profiles at $x = 25, 30, 35$. If the decay rate or the stress levels move materially outside the DNS and experimental bands, the reported accuracy is tied to the chosen operating point rather than to the closure itself. A complementary check is to apply the identical constants to another spatially developing shear flow, such as a plane channel or a backward-facing step at $\\mathrm{Re} \\approx 5000$, and see whether the mean flow and second-order statistics still match reference data.","supporting_citations":[{"cited_title":"Investigation of high-pressure turbulent jets using direct numerical simulation","cited_arxiv_id":null,"evidence_quote":"The direct numerical simulation at $\\mathrm{Re}_j = 5000$ that defines the 2%-grid cell-count baseline and supplies the reference decay rate, mean-velocity profiles, Reynolds stresses, and the hyperbolic-tangent inflow profile adopted here."},{"cited_title":"Velocity measurements in a high-Reynolds- number, momentum-conserving, axisymmetric, turbulent jet","cited_arxiv_id":null,"evidence_quote":"The experiment supplying the benchmark centreline decay constant $B_u = 5.80$, against which the predicted $B_u = 5.69$ is compared."},{"cited_title":"Turbulence measurements in axisymmetric jets of air and helium","cited_arxiv_id":null,"evidence_quote":"The experiment supplying $B_u = 6.06$ and the measured mean-velocity and fluctuation profiles used as comparison data."},{"cited_title":"Simulation of the blooming phenomenon in forced circular jets","cited_arxiv_id":null,"evidence_quote":"Supplies the dual-mode (axisymmetric plus helical) inlet perturbation with $St_D = 0.5$ and $f = 2.40$ used to trigger transition, and the alternative frequency $f = 2.22$ used in the sensitivity test."},{"cited_title":"Unified gas-kinetic wave-particle methods I: Continuum and rarefied gas flow","cited_arxiv_id":null,"evidence_quote":"The unified gas-kinetic wave-particle framework whose coupled wave-particle evolution and free-transport flux treatment the WPTS method is built upon."},{"cited_title":"A gas-kinetic BGK scheme for the Navier–Stokes equations and its connection with artificial dissipation and Godunov method","cited_arxiv_id":null,"evidence_quote":"The gas-kinetic BGK scheme that WPTS automatically reverts to in well-resolved cells, which guarantees recovery of the Navier-Stokes solution in laminar or resolved regions."}],"review_version":1}