pith. sign in

arxiv: 2605.29211 · v1 · pith:ZGNSMCXYnew · submitted 2026-05-28 · ⚛️ physics.flu-dyn

Tail observability and fourth-order closure recovery in physics-informed neural networks for Bhatnagar-Gross-Krook normal shocks

Ehsan Roohi This is my paper

Pith reviewed 2026-06-29 05:58 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn
keywords physics-informed neural networksBGK equationnormal shocksmoment closuretail observabilitydiscrete velocity methodnonequilibrium distributionfourth-order moments
0
0 comments X

The pith

A shock-local correction aligned with unobserved tail-weighted cancellation reduces relative R_xx^cl error to 0.112 in PINN BGK Mach-2 shocks while preserving lower moments.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that accurate density, velocity and temperature profiles in a physics-informed neural network for BGK normal shocks do not guarantee accurate fourth-order closure terms because those terms arise from velocity-weighted projections sensitive to high-velocity tails. Using independent discrete-velocity method references, the authors find that anchoring on heat flux and normal stress together stabilizes the nonequilibrium layer, yet R_xx^cl still shows order-unity error. DVM diagnostics trace the discrepancy to a sign-changing, tail-weighted cancellation that lower moments observe only weakly. Adding a shock-local closure correction that targets this missing projection brings the relative error in R_xx^cl down to 1.12 times 10 to the minus 1 without disturbing the recovered lower moments. The work also shows that distribution-function probe losses are diagnostic rather than constitutive and that the obstruction is anisotropic tail weighting rather than polynomial degree alone.

Core claim

In a stationary Mach-2 normal shock the positive macro-micro PINN recovers rho, u_x, T, q_x, sigma_xx and m_xxx^cl but leaves R_xx^cl with order-unity error. DVM diagnostics show that R_xx^cl is controlled by a sign-changing, tail-weighted cancellation weakly observed by lower moments. A shock-local closure correction aligned with this missing projection reduces the relative R_xx^cl error to 1.12 times 10 to the minus 1 while preserving the lower moments.

What carries the argument

The shock-local closure correction aligned with the sign-changing tail-weighted cancellation in the nonequilibrium distribution that controls R_xx^cl.

If this is right

  • Sparse joint anchoring of heat flux and normal stress stabilises the primary nonequilibrium layer.
  • Residual-only, macro-only and single-moment variants fail in distinct ways.
  • Optional distribution-function probe losses are diagnostic rather than constitutive.
  • The obstruction to fourth-order recovery is anisotropic sign-changing tail weighting rather than polynomial degree alone.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same tail-observability limitation is likely to appear in multi-dimensional or unsteady shock configurations where higher moments matter.
  • PINN kinetic solvers for other nonequilibrium problems may require explicit tail-sensitive anchoring beyond standard moment losses.
  • The diagnostic value of probe losses suggests a general strategy for verifying whether a learned distribution captures the projections needed for closure.

Load-bearing premise

The sign-changing, tail-weighted cancellation diagnosed by DVM is the dominant and correctable source of the R_xx^cl discrepancy rather than an artifact of the chosen PINN representation or the specific Mach-2 stationary shock setup.

What would settle it

Applying the identical PINN architecture and correction to a Mach-3 or Mach-1.5 stationary BGK shock and checking whether the relative R_xx^cl error remains near 0.112 or returns to order unity.

Figures

Figures reproduced from arXiv: 2605.29211 by Ehsan Roohi.

Figure 1
Figure 1. Figure 1: Illustration of high-order non-observability under low-order moment constraints. A fourth-order [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Polynomial and Maxwellian-weighted moment kernels used to illustrate the velocity-space origin [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Cumulative contribution from the high-peculiar-speed region [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Base strong shock-tube case S0: comparison between the positive macro–micro neural solver and [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Velocity-space diagnostics from the refined conservative DVM normal-shock reference. (a) Line [PITH_FULL_IMAGE:figures/full_fig_p016_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Mach-2 stationary BGK normal shock: PINN with a shock-local tail-closure head compared with [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Data-efficiency of joint 𝑞–𝜎 anchoring. The best low-data setting, qsig16/m32, uses 16 shock-layer locations for the nonequilibrium moment anchors and approaches the denser full42/m160 upper-bound. This plot supports the interpretation of the method as sparse moment regularisation rather than dense supervised learning. 21 [PITH_FULL_IMAGE:figures/full_fig_p021_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Supplementary scalar fourth-order excess diagnostic. (a) Direct comparison between the DVM– [PITH_FULL_IMAGE:figures/full_fig_p028_8.png] view at source ↗
read the original abstract

Closure-level accuracy in neural kinetic shock solvers is not guaranteed by accurate density, velocity and temperature profiles, because the relevant observables are velocity-weighted projections of the nonequilibrium distribution. We study this observability problem for one-dimensional Bhatnagar--Gross--Krook (BGK) shock waves using a positive macro--micro physics-informed neural network (PINN) in which the distribution is represented as a local Maxwellian multiplied by a bounded exponential correction. Independent discrete-velocity method (DVM) references are used for validation. Shock-tube tests show that sparse joint anchoring of heat flux and normal stress stabilises the primary nonequilibrium layer, whereas residual-only, macro-only and single-moment variants fail in distinct ways. In a stationary Mach-2 normal shock, a flux-locked compact model recovers $\rho$, $u_x$, $T$, $q_x$, $\sigma_{xx}$ and $m_{xxx}^{cl}$, but leaves $R_{xx}^{cl}$ with order-unity error. DVM diagnostics show that $R_{xx}^{cl}$ is controlled by a sign-changing, tail-weighted cancellation weakly observed by lower moments. A shock-local closure correction aligned with this missing projection reduces the relative $R_{xx}^{cl}$ error to $1.12\times10^{-1}$ while preserving the lower moments. A common-initialisation ablation shows that optional distribution-function probe losses are diagnostic rather than constitutive. A supplementary DVM--PINN comparison for the scalar fourth-order excess $\Delta$ shows that the obstruction is anisotropic, sign-changing tail weighting rather than fourth-order polynomial degree alone.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper claims that accurate recovery of density, velocity and temperature in a positive macro-micro PINN for BGK shocks does not guarantee closure-level accuracy for higher moments because of velocity-weighted tail projections; using DVM validation it shows that joint anchoring of heat flux and normal stress stabilizes the nonequilibrium layer, while a post-hoc shock-local correction aligned with a DVM-diagnosed sign-changing tail cancellation reduces relative error in R_xx^cl to 1.12×10^{-1} for a stationary Mach-2 shock while preserving lower moments.

Significance. If the tail-observability diagnosis and correction generalize beyond the single Mach-2 stationary case, the work would usefully highlight a structural limitation in neural kinetic solvers and provide a practical route to fourth-order closure recovery. The use of independent DVM references for quantitative validation is a clear strength; the reported error reduction is concrete but remains tied to one specific setup.

major comments (2)
  1. [Mach-2 stationary shock results] Mach-2 stationary shock results (abstract and main text): the shock-local closure correction is introduced after DVM diagnostics of the sign-changing tail cancellation and reduces relative R_xx^cl error to 1.12×10^{-1}; however, no tests are reported for different Mach numbers, moving shocks, or altered collision models, so the claim that this alignment recovers general tail observability rests on a single-case demonstration.
  2. [Abstract and supplementary DVM-PINN comparison] Abstract and supplementary DVM-PINN comparison: the central assertion that the obstruction is 'anisotropic, sign-changing tail weighting rather than fourth-order polynomial degree alone' is supported only by the Mach-2 DVM diagnostics; without additional cases the distinction between setup-specific artifact and general mechanism remains untested and load-bearing for the observability claim.
minor comments (1)
  1. [Abstract] The abstract states that 'optional distribution-function probe losses are diagnostic rather than constitutive' but provides no quantitative ablation metrics; a brief table or sentence with the common-initialisation error values would clarify this point.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful review and constructive comments on the scope of the study. The manuscript presents a detailed diagnostic analysis for the stationary Mach-2 BGK shock using PINN and independent DVM validation to identify the tail-weighted observability issue. We address each major comment below.

read point-by-point responses

  1. Referee: [Mach-2 stationary shock results] Mach-2 stationary shock results (abstract and main text): the shock-local closure correction is introduced after DVM diagnostics of the sign-changing tail cancellation and reduces relative R_xx^cl error to 1.12×10^{-1}; however, no tests are reported for different Mach numbers, moving shocks, or altered collision models, so the claim that this alignment recovers general tail observability rests on a single-case demonstration.

    Authors: We agree that all quantitative results, including the error reduction to 1.12×10^{-1} and the shock-local correction, are demonstrated exclusively for the stationary Mach-2 case. The manuscript frames this as a canonical example to diagnose the anisotropic tail-weighting mechanism via DVM, rather than asserting immediate generality of the correction. The text does not claim the alignment recovers tail observability for arbitrary Mach numbers or moving shocks; such extensions would require new simulations and lie outside the present scope. revision: no

  2. Referee: [Abstract and supplementary DVM-PINN comparison] Abstract and supplementary DVM-PINN comparison: the central assertion that the obstruction is 'anisotropic, sign-changing tail weighting rather than fourth-order polynomial degree alone' is supported only by the Mach-2 DVM diagnostics; without additional cases the distinction between setup-specific artifact and general mechanism remains untested and load-bearing for the observability claim.

    Authors: The supplementary comparison of the scalar fourth-order excess Δ versus the tensorial R_xx^cl is performed within the Mach-2 DVM data to illustrate that the obstruction arises from anisotropic, sign-changing tail weighting rather than polynomial degree. We acknowledge this remains a single-configuration demonstration. We will revise the abstract and relevant passages to state that the mechanism is diagnosed for the Mach-2 stationary shock, avoiding any implication of untested generality. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation relies on independent external DVM benchmarks

full rationale

The paper's central results use independent discrete-velocity method (DVM) references for validation and diagnostics, with the shock-local closure correction introduced after DVM analysis of tail-weighted cancellation rather than being forced by the PINN equations or any self-referential fitting. No load-bearing self-citations, fitted inputs renamed as predictions, or self-definitional steps are present; the derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Based solely on abstract; no explicit free parameters, axioms or invented entities are quantified. The bounded exponential correction and the shock-local closure correction are introduced without independent evidence outside the DVM comparison.

invented entities (1)
  • shock-local closure correction no independent evidence
    purpose: to compensate for the unobserved tail-weighted cancellation in R_xx^cl
    Introduced after DVM diagnostics to reduce the reported error; no falsifiable prediction outside the present study is given.

pith-pipeline@v0.9.1-grok · 5824 in / 1239 out tokens · 25292 ms · 2026-06-29T05:58:43.182580+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Closure-channel identifiability and two-channel recovery in monatomic kinetic normal shocks

    astro-ph.HE 2026-05 unverdicted novelty 6.0

    Heat-flux budget observes only projected fourth-order channel S, not separate R_cl_xx and Δ; scalar-excess budget supplies the missing channel for two-channel reconstruction with low error in BGK and Shakhov models.

Reference graph

Works this paper leans on

42 extracted references · 40 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    MacArt, and Justin Sirignano

    Nicholas Daultry Ball, Jonathan F. MacArt, and Justin Sirignano. Online optimisation of machine learning collision models to accelerate direct molecular simulation of rarefied gas flows. Journal of Computational Physics, 530: 0 114601, 2026. doi:10.1016/j.jcp.2025.114601

    work page doi:10.1016/j.jcp.2025.114601 2026

  2. [2]

    P. L. Bhatnagar, E. P. Gross, and M. Krook. A model for collision processes in gases. i. small amplitude processes in charged and neutral one-component systems. Physical Review, 94 0 (3): 0 511--525, 1954. doi:10.1103/PhysRev.94.511

    work page doi:10.1103/physrev.94.511 1954

  3. [3]

    Graeme A. Bird. Molecular Gas Dynamics and the Direct Simulation of Gas Flows. Clarendon Press, Oxford, 1994. URL https://global.oup.com/academic/product/9780198561958

    work page arXiv 1994

  4. [4]

    Flow over an espresso cup: inferring three-dimensional velocity and pressure fields from tomographic background oriented schlieren via physics-informed neural networks

    Shengze Cai, Zhicheng Wang, Frederik Fuest, Young Jin Jeon, Callum Gray, and George Em Karniadakis. Flow over an espresso cup: inferring three-dimensional velocity and pressure fields from tomographic background oriented schlieren via physics-informed neural networks. Journal of Fluid Mechanics, 915: 0 A102, 2021. doi:10.1017/jfm.2021.135

    work page doi:10.1017/jfm.2021.135 2021

  5. [5]

    The Boltzmann Equation and Its Applications

    Carlo Cercignani. The Boltzmann Equation and Its Applications . Springer, 1988. doi:10.1007/978-1-4612-1039-9

    work page doi:10.1007/978-1-4612-1039-9 1988

  6. [6]

    Sydney Chapman and T. G. Cowling. The Mathematical Theory of Non-Uniform Gases. Cambridge University Press, 3 edition, 1970. Cambridge Core https://www.cambridge.org/us/universitypress/subjects/mathematics/fluid-dynamics-and-solid-mechanics/mathematical-theory-non-uniform-gases-account-kinetic-theory-viscosity-thermal-conduction-and-diffusion-gases

    1970

  7. [7]

    Structure and asymptotic preserving deep neural surrogates for uncertainty quantification in multiscale kinetic equations

    Wei Chen, Giacomo Dimarco, and Lorenzo Pareschi. Structure and asymptotic preserving deep neural surrogates for uncertainty quantification in multiscale kinetic equations. arXiv preprint arXiv :2506.10636 , 2025. doi:10.48550/arXiv.2506.10636

    work page doi:10.48550/arxiv.2506.10636 2025

  8. [8]

    Numerical methods for kinetic equations

    Giacomo Dimarco and Lorenzo Pareschi. Numerical methods for kinetic equations. Acta Numerica, 23: 0 369--520, 2014. doi:10.1017/S0962492914000063

    work page doi:10.1017/s0962492914000063 2014

  9. [9]

    Physics-informed neural networks for solving reynolds-averaged navier--stokes equations

    Hamidreza Eivazi, Mojtaba Tahani, Philipp Schlatter, and Ricardo Vinuesa. Physics-informed neural networks for solving reynolds-averaged navier--stokes equations. Physics of Fluids, 34 0 (7): 0 075117, 2022. doi:10.1063/5.0095270

    work page doi:10.1063/5.0095270 2022

  10. [10]

    Physics-informed deep-learning applications to experimental fluid mechanics

    Hamidreza Eivazi, Yuning Wang, and Ricardo Vinuesa. Physics-informed deep-learning applications to experimental fluid mechanics. Measurement Science and Technology, 35 0 (7): 0 075303, 2024. doi:10.1088/1361-6501/ad3fd3

    work page doi:10.1088/1361-6501/ad3fd3 2024

  11. [11]

    A benchmark study of kinetic models for shock waves

    Fei Fei, Haihong Liu, Zhaohui Liu, and Jun Zhang. A benchmark study of kinetic models for shock waves. AIAA Journal, 58 0 (6): 0 2596--2608, 2020. doi:10.2514/1.J059029

    work page doi:10.2514/1.j059029 2020

  12. [12]

    A class of asymptotic-preserving schemes for kinetic equations and related problems with stiff sources

    Francis Filbet and Shi Jin. A class of asymptotic-preserving schemes for kinetic equations and related problems with stiff sources. Journal of Computational Physics, 229 0 (20): 0 7625--7648, 2010. doi:10.1016/j.jcp.2010.06.017

    work page doi:10.1016/j.jcp.2010.06.017 2010

  13. [13]

    Observable-augmented manifold learning for multi-source turbulent flow data

    Kai Fukami and Kunihiko Taira. Observable-augmented manifold learning for multi-source turbulent flow data. Journal of Fluid Mechanics, 1010: 0 R4, 2025. doi:10.1017/jfm.2025.383

    work page doi:10.1017/jfm.2025.383 2025

  14. [14]

    On the kinetic theory of rarefied gases

    Harold Grad. On the kinetic theory of rarefied gases. Communications on Pure and Applied Mathematics, 2 0 (4): 0 331--407, 1949. doi:10.1002/cpa.3160020403

    work page doi:10.1002/cpa.3160020403 1949

  15. [15]

    Xiao-Jun Gu and David R. Emerson. A high-order moment approach for capturing non-equilibrium phenomena in the transition regime. Journal of Fluid Mechanics, 636: 0 177--216, 2009. doi:10.1017/S002211200900768X

    work page doi:10.1017/s002211200900768x 2009

  16. [16]

    Hirschfelder, Charles F

    Joseph O. Hirschfelder, Charles F. Curtiss, and R. Byron Bird. Molecular Theory of Gases and Liquids. Wiley, 1954. URL https://www.wiley.com/en-us/Molecular+Theory+of+Gases+and+Liquids-p-9780471400653

    1954

  17. [17]

    Efficient asymptotic-preserving schemes for some multiscale kinetic equations

    Shi Jin. Efficient asymptotic-preserving schemes for some multiscale kinetic equations. SIAM Journal on Scientific Computing, 21 0 (2): 0 441--454, 1999. doi:10.1137/S1064827598334599

    work page doi:10.1137/s1064827598334599 1999

  18. [18]

    Asymptotic-preserving neural networks for multiscale time-dependent linear transport equations

    Shi Jin, Zheng Ma, and Keke Wu. Asymptotic-preserving neural networks for multiscale time-dependent linear transport equations. arXiv preprint arXiv:2111.02541, 2021. doi:10.48550/arXiv.2111.02541

    work page doi:10.48550/arxiv.2111.02541 2021

  19. [19]

    Asymptotic-preserving neural networks for multiscale kinetic equations

    Shi Jin, Zheng Ma, and Keke Wu. Asymptotic-preserving neural networks for multiscale kinetic equations. arXiv preprint arXiv:2306.15381, 2023. doi:10.48550/arXiv.2306.15381

    work page doi:10.48550/arxiv.2306.15381 2023

  20. [20]

    Karniadakis and Ali Beskok

    George E. Karniadakis and Ali Beskok. Micro Flows: Fundamentals and Simulation. Springer, 2001. doi:10.1007/978-0-387-28676-1

    work page doi:10.1007/978-0-387-28676-1 2001

  21. [21]

    A Theory-guided Weighted $L^2$ Loss for solving the BGK model via Physics-informed neural networks

    Gyounghun Ko, Sung-Jun Son, Seung Yeon Cho, and Myeong-Su Lee. A theory-guided weighted L2 loss for solving the BGK model via physics-informed neural networks. arXiv preprint arXiv :2604.04971 , 2026. doi:10.48550/arXiv.2604.04971

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2604.04971 2026

  22. [22]

    David Levermore

    C. David Levermore. Moment closure hierarchies for kinetic theories. Journal of Statistical Physics, 83: 0 1021--1065, 1996. doi:10.1007/BF02179552

    work page doi:10.1007/bf02179552 1996

  23. [23]

    A bi-fidelity based asymptotic-preserving neural network for the semiconductor boltzmann equation and its inverse problem

    Liu Liu, Xueyu Zhu, and Zhenyi Zhu. A bi-fidelity based asymptotic-preserving neural network for the semiconductor boltzmann equation and its inverse problem. Journal of Computational Physics, page 115076, 2026. doi:10.1016/j.jcp.2026.115076

    work page doi:10.1016/j.jcp.2026.115076 2026

  24. [24]

    Physics-informed neural networks for solving forward and inverse flow problems via the Boltzmann--BGK formulation

    Qin Lou, Xuhui Meng, and George Em Karniadakis. Physics-informed neural networks for solving forward and inverse flow problems via the Boltzmann--BGK formulation. Journal of Computational Physics, 447: 0 110676, 2021. doi:10.1016/j.jcp.2021.110676

    work page doi:10.1016/j.jcp.2021.110676 2021

  25. [25]

    Probabilistic neural networks for fluid flow surrogate modeling and data recovery

    Romit Maulik, Kai Fukami, Nesar Ramachandra, Koji Fukagata, and Kunihiko Taira. Probabilistic neural networks for fluid flow surrogate modeling and data recovery. Physical Review Fluids, 5 0 (10): 0 104401, 2020. doi:10.1103/PhysRevFluids.5.104401

    work page doi:10.1103/physrevfluids.5.104401 2020

  26. [26]

    Discrete velocity model and implicit scheme for the BGK equation of rarefied gas dynamics

    Luc Mieussens. Discrete velocity model and implicit scheme for the BGK equation of rarefied gas dynamics. Mathematical Models and Methods in Applied Sciences, 10 0 (8): 0 1121--1149, 2000. doi:10.1142/S0218202500000562

    work page doi:10.1142/s0218202500000562 2000

  27. [27]

    Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression

    Masaki Morimoto, Kai Fukami, Romit Maulik, Ricardo Vinuesa, and Koji Fukagata. Assessments of epistemic uncertainty using Gaussian stochastic weight averaging for fluid-flow regression. Physica D: Nonlinear Phenomena, 440: 0 133454, 2022. doi:10.1016/j.physd.2022.133454

    work page doi:10.1016/j.physd.2022.133454 2022

  28. [28]

    Separable physics-informed neural networks for solving the BGK model of the Boltzmann equation

    Jaemin Oh, Seung Yeon Cho, Seok-Bae Yun, Eunbyung Park, and Youngjoon Hong. Separable physics-informed neural networks for solving the BGK model of the Boltzmann equation. SIAM Journal on Scientific Computing, 47 0 (2): 0 C451--C474, 2025. doi:10.1137/24M1668809

    work page doi:10.1137/24m1668809 2025

  29. [29]

    Stochastic numerics for the Boltzmann equation

    Sergej Rjasanow and Wolfgang Wagner. Stochastic numerics for the Boltzmann equation. Springer Series in Computational Mathematics, 37, 2005. doi:10.1007/3-540-27689-0

    work page doi:10.1007/3-540-27689-0 2005

  30. [30]

    Neural networks for rarefied gas dynamics: Relaxation problem, polyatomic shock waves, and hypersonic cylinder flow

    Ehsan Roohi, Ahmad Shoja-Sani, and Fatemeh Ebrahimzadeh Azghadi. Neural networks for rarefied gas dynamics: Relaxation problem, polyatomic shock waves, and hypersonic cylinder flow. Physics of Fluids, 38 0 (5): 0 057108, 2026 a . doi:10.1063/5.0334590

    work page doi:10.1063/5.0334590 2026

  31. [31]

    Physics constrained neural collision operators for hard sphere surrogates and ab initio angle prediction in direct simulation Monte Carlo

    Ehsan Roohi, Ahmad Shoja-Sani, and Stefan Stefanov. Physics constrained neural collision operators for hard sphere surrogates and ab initio angle prediction in direct simulation Monte Carlo . Physics of Fluids, 38 0 (5): 0 057123, 2026 b . doi:10.1063/5.0328463

    work page doi:10.1063/5.0328463 2026

  32. [32]

    Homogeneous relaxation and shock wave problems: Assessment of the simplified and generalized bernoulli trial collision schemes

    Ahmad Shoja-Sani, Ehsan Roohi, and Stefan Stefanov. Homogeneous relaxation and shock wave problems: Assessment of the simplified and generalized bernoulli trial collision schemes. Physics of Fluids, 33: 0 032004, 2021. doi:10.1063/5.0039071

    work page doi:10.1063/5.0039071 2021

  33. [33]

    Molecular Gas Dynamics: Theory, Techniques, and Applications

    Yoshio Sone. Molecular Gas Dynamics: Theory, Techniques, and Applications. Birkh\"auser, 2007. doi:10.1007/978-0-8176-4573-3

    work page doi:10.1007/978-0-8176-4573-3 2007

  34. [34]

    Macroscopic Transport Equations for Rarefied Gas Flows

    Henning Struchtrup. Macroscopic Transport Equations for Rarefied Gas Flows. Springer, 2005. doi:10.1007/3-540-32386-4

    work page doi:10.1007/3-540-32386-4 2005

  35. [35]

    Regularization of Grad 's 13 moment equations: Derivation and linear analysis

    Henning Struchtrup and Manuel Torrilhon. Regularization of Grad 's 13 moment equations: Derivation and linear analysis. Physics of Fluids, 15 0 (9): 0 2668--2680, 2003. doi:10.1063/1.1597472

    work page doi:10.1063/1.1597472 2003

  36. [36]

    Physics-constrained Bayesian neural network for fluid flow reconstruction with sparse and noisy data

    Luning Sun and Jian-Xun Wang. Physics-constrained Bayesian neural network for fluid flow reconstruction with sparse and noisy data. Theoretical and Applied Mechanics Letters, 10 0 (3): 0 161--169, 2020. doi:10.1016/j.taml.2020.01.031

    work page doi:10.1016/j.taml.2020.01.031 2020

  37. [37]

    F. G. Tcheremissine. Conservative evaluation of Boltzmann collision integral in discrete ordinates approximation. Computers & Mathematics with Applications, 35 0 (1--2): 0 215--221, 1998. doi:10.1016/S0898-1221(97)00269-1

    work page doi:10.1016/s0898-1221(97)00269-1 1998

  38. [38]

    Modeling nonequilibrium gas flow based on moment equations

    Manuel Torrilhon. Modeling nonequilibrium gas flow based on moment equations. Annual Review of Fluid Mechanics, 48: 0 429--458, 2016. doi:10.1146/annurev-fluid-122414-034259

    work page doi:10.1146/annurev-fluid-122414-034259 2016

  39. [39]

    Regularized 13-moment equations: shock structure calculations and comparison to Burnett models

    Manuel Torrilhon and Henning Struchtrup. Regularized 13-moment equations: shock structure calculations and comparison to Burnett models. Journal of Fluid Mechanics, 513: 0 171--198, 2004. doi:10.1017/S0022112004009907

    work page doi:10.1017/s0022112004009907 2004

  40. [40]

    A unified gas-kinetic scheme for continuum and rarefied flows

    Kun Xu and Jian-Cheng Huang. A unified gas-kinetic scheme for continuum and rarefied flows. Journal of Computational Physics, 229 0 (20): 0 7747--7764, 2010. doi:10.1016/j.jcp.2010.06.032

    work page doi:10.1016/j.jcp.2010.06.032 2010

  41. [41]

    Filtered partial differential equations: a robust surrogate constraint in physics-informed deep learning framework

    Dashan Zhang, Yuntian Chen, and Shiyi Chen. Filtered partial differential equations: a robust surrogate constraint in physics-informed deep learning framework. Journal of Fluid Mechanics, 999: 0 A40, 2024. doi:10.1017/jfm.2024.471

    work page doi:10.1017/jfm.2024.471 2024

  42. [42]

    Kerswell, and P

    Lu Zhu, Xianyang Jiang, Adrien Lefauve, Rich R. Kerswell, and P. F. Linden. New insights into experimental stratified flows obtained through physics-informed neural networks. Journal of Fluid Mechanics, 981: 0 R1, 2024. doi:10.1017/jfm.2024.49

    work page doi:10.1017/jfm.2024.49 2024