Pith. sign in

REVIEW 2 major objections 5 minor 52 references

A parallel DIG-augmented DSMC solver on unstructured meshes cuts near-continuum rarefied-flow cost by hundreds of times while matching standard DSMC.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-10 11:30 UTC pith:3NFHXDXK

load-bearing objection Solid engineering paper: parallel unstructured DIG–DSMC with real 3D cost wins and shipped code; novelty is the stack, not the theory. the 2 major comments →

arxiv 2607.08195 v1 pith:3NFHXDXK submitted 2026-07-09 physics.comp-ph physics.flu-dyn

Parallel simulation of rarefied gas flows on unstructured meshes using the DIG-augmented DSMC method

Tao Huang , Liyan Luo , Hong Deng , Lei Wu This is my paper
classification physics.comp-ph physics.flu-dyn
keywords DSMCDIGrarefied gas dynamicsunstructured meshasymptotic preservingparallel computingload balancinghypersonic flow
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Standard DSMC becomes prohibitively expensive in the near-continuum regime because cell size and time step must stay smaller than the molecular mean free path and collision time, and long statistical averaging is needed to quiet noise. This paper builds a fully parallel three-dimensional solver that keeps ordinary DSMC particle transport and collisions but periodically hands sampled macroscopic moments and higher-order closures to synthetic continuum equations, then feeds the corrected macroscopic state back into the particles. On unstructured meshes the implementation uses ghost cells for local particle tracking, batch MPI migration, and METIS-based dynamic load balancing so that work stays balanced across shock layers and wakes. Validation on lid-driven cavity, sphere, Apollo capsule and International Space Station flows shows agreement with SPARTA while the number of cells and sampling steps drops sharply; for cavity flow at Kn = 0.01 the method reports roughly 360-fold fewer core hours and 50-fold lower memory. The practical payoff is high-fidelity rarefied aerodynamics over complex geometries that would otherwise be out of reach.

Core claim

A parallel DIG-augmented DSMC solver on three-dimensional unstructured meshes reaches the same macroscopic fields as standard DSMC while, because of DIG’s intermittent macroscopic corrections and asymptotic-preserving property, requiring far fewer cells and sampling steps and therefore far less memory and wall-clock time, especially as the Knudsen number falls.

What carries the argument

The DIG intermittent coupling cycle: after Ns DSMC steps, higher-order stress and heat-flux corrections (HoTs) are extracted from exponentially averaged particle moments, supplied as fixed closures to the macroscopic synthetic equations, and the resulting low-order macroscopic state is used to resample particle number, velocity and rotational energy; on unstructured meshes this is stabilized by a local-cell Knudsen admissibility test and neighbor reconstruction, and parallelized by ghost-cell local tracking plus graph-partition dynamic load balancing.

Load-bearing premise

The higher-order stress and heat-flux closures sampled from short DSMC windows, after a cell-Knudsen filter and neighbor reconstruction for noisy tiny cells, remain accurate enough that the macroscopic solve and particle resampling do not bias the final steady fields.

What would settle it

Run the same 3D lid-driven cavity or sphere case at Kn = 0.01 with DIG and with a fully resolved SPARTA mesh; if density, velocity and temperature profiles (or surface heat flux) differ systematically beyond statistical noise once both are statistically steady, the claimed accuracy of the HoT closures fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper presents a parallel DIG-augmented DSMC solver for three-dimensional rarefied gas flows on unstructured meshes. It retains standard DSMC particle transport, collisions (VSS/NTC), surface interactions and Borgnakke–Larsen energy exchange, while intermittently coupling to macroscopic synthetic equations that incorporate higher-order stress and heat-flux closures (HoTs) sampled from DSMC. Parallelization uses a hybrid MPI layout with one storage rank and compute ranks, ghost/virtual cells for local face-based particle tracking, batched inter-rank migration, and METIS graph partitioning with particle-count or timing-based vertex weights plus Kuhn–Munkres label matching for dynamic load balancing. Validation covers 3-D lid-driven cavity, hypersonic sphere, Apollo capsule and ISS flows across Kn = 1–0.01, demonstrating macroscopic agreement with SPARTA (or GSIS when full 3-D SPARTA is prohibitive) together with large reductions in cell count, sampling steps, memory and core-hours, most pronounced near continuum.

Significance. If the reported accuracy and cost figures hold, the work supplies a practical high-performance tool that removes the kinetic-scale cell-size and time-step restrictions of conventional DSMC while preserving its particle framework and physico-chemical modeling capability. The combination of asymptotic-preserving DIG coupling, unstructured-mesh support, ghost-cell batch migration and dynamic load balancing addresses a genuine bottleneck for complex-geometry rarefied flows (re-entry, vacuum systems). Explicit release of the C++ code further raises the contribution’s value for reproducibility and community adoption. The multi-case empirical evidence (field/surface comparisons plus Tables 1–5) is stronger than purely algorithmic claims that lack external baselines.

major comments (2)
  1. [§5.2, Table 2] §5.2 and Table 2 (Kn = 0.01 sphere): the only SPARTA reference is axisymmetric, while DIG is fully 3-D. The paper correctly notes that a 3-D SPARTA run is prohibitive, yet the central cost-reduction claim would be more robust if an order-of-magnitude estimate of the equivalent 3-D SPARTA cell count / core-hours (based on the same mean-free-path resolution criterion used for Kn = 0.1) were supplied, or if a short statement clarified that the axisymmetric comparison already under-states the 3-D savings.
  2. [§3.2, Eq. (19)] §3.1–3.2, Eqs. (14)–(19) and Fig. 1: the cell-Knudsen admissibility test Kn_Δ,c ≤ Kn_Δ,max and the subsequent multi-layer equal-weight neighbor reconstruction are load-bearing for the quality of the HoTs that close the synthetic equations. The manuscript never reports the numerical value of Kn_Δ,max used in any of the four test cases, nor the fraction of cells that required reconstruction. Without these data (or a brief sensitivity check), it is difficult for a reader to judge how often the safeguard is active and whether residual reconstruction bias remains negligible on highly irregular meshes.
minor comments (5)
  1. [§3–4] Throughout §2–4 the free parameters Ns (=100), na (=1000), η_crit_L (=1.05), N_bal (=10) and α (=0.5) are stated as “typical” or “unless otherwise stated.” A single table listing the exact values employed for each of the four configurations would improve reproducibility.
  2. [Figs. 6, 8] Fig. 6 and Fig. 8: contour-line overlays of DIG versus SPARTA are useful, but the line thickness and color contrast make quantitative differences hard to judge near shocks and walls; adding a few line-cut insets (as already done for the sphere stagnation line) would help.
  3. [Abstract] Abstract and §1: the GitHub link is given, yet the repository name “DIG-appolo” appears to contain a typographical inconsistency with the Apollo geometry; a short README note clarifying the intended spelling would avoid confusion.
  4. [§2.1] Eq. (3) and surrounding text: the inelastic probability formula is written for the VSS model; a one-sentence reminder that the same expression reduces to the VHS limit when α → 1 would aid readers who use only VHS.
  5. [Tables 3, 5] Table 3 and Table 5: parallel-efficiency numbers slightly above 100 % are correctly attributed to stochastic particle fluctuations, but a parenthetical remark that the particle load was deliberately reduced to 10 % of production would make the super-linear entries less surprising.

Circularity Check

1 steps flagged

No significant circularity: empirical performance claims rest on independent SPARTA (and GSIS) benchmarks, not tautological restatements of DIG definitions.

specific steps
  1. self citation load bearing [§1 (Introduction) and §2–3 (DIG formulation); citations [32,34,35,52]]
    "As a hybrid acceleration approach coupling DSMC with macroscopic synthetic equations, the direct intermittent general synthetic iterative scheme (DIG) delivers fast convergence and asymptotic-preserving characteristics... Prior DIG work has verified its acceleration efficacy for monatomic [32], polyatomic [34]... Owing to the tailored constitutive relations... DIG exhibits rapid convergence and asymptotic preservation of the NSF limit."

    The load-bearing assertion that DIG supplies the asymptotic-preserving and fast-convergence properties that justify coarser meshes/fewer samples is justified solely by citations to the same authors’ earlier DIG papers (and the Fourier analysis [52]). Those properties are not re-derived or independently proved in the present manuscript; they are taken as given. The circularity is mild because the paper’s actual claims are empirical agreement and measured speed-ups against external SPARTA, not a new theoretical derivation that collapses to the self-citation.

full rationale

This is a methods/parallel-implementation paper whose central claims are (i) agreement of the new unstructured-mesh DIG-DSMC solver with external SPARTA DSMC fields/surfaces (Figs. 6–9, 11–12) and (ii) measured wall-clock/memory reductions that follow from coarser meshes and fewer samples (Tables 1–2, 4). The DIG coupling itself (HoT extraction Eqs. 14–15, particle resampling, Algorithm 1) and its asymptotic-preserving/fast-convergence properties are imported from the authors’ prior papers, which is normal for an extension; those properties are not re-derived here as novel first-principles results, nor are any fitted parameters re-labeled as predictions. Validation against SPARTA (an independent code) and, where full 3-D SPARTA is prohibitive, against a deterministic GSIS solver on the identical mesh supplies external falsifiability. Parallel machinery (ghost cells, METIS load balancing, Kuhn–Munkres) is orthogonal to the accuracy claim. No self-definitional loop, no uniqueness theorem used to forbid alternatives, and no ansatz smuggled as derivation appear. Residual self-citation burden is minor and non-load-bearing for the reported empirical outcomes.

Axiom & Free-Parameter Ledger

6 free parameters · 4 axioms · 0 invented entities

The work rests on standard DSMC/Boltzmann modeling, prior DIG synthetic-equation theory, and several hand-chosen numerical thresholds that control coupling noise, cell admissibility, and re-partitioning. No new physical entities are postulated; free parameters are numerical controls and standard collision-model constants.

free parameters (6)
  • Ns (outer DIG coupling interval) = 100
    Number of DSMC steps per macroscopic correction; set to 100 by practice to balance noise vs. dissipation (§3.1).
  • na (exponential moving-average window) = 1000
    Smoothing length for sampled moments before HoT extraction; chosen as 1000 (§3.1).
  • Kn_Δ,max (cell admissibility threshold) = case-dependent
    Local cell-Knudsen cutoff deciding whether DSMC samples feed the macroscopic solver or are reconstructed from neighbors; case-dependent numerical parameter (§3.2, Eq. 19).
  • η_crit_L and N_bal (load-balance triggers) = 1.05; 10
    Imbalance ratio threshold and re-partition interval for METIS; defaults 1.05 and 10 (§4.3.1).
  • α (runtime EMA smoothing for cell weights) = 0.5
    Smoothing factor for timing-based METIS vertex weights (§4.3.2).
  • Zr (rotational collision number) = 2.59
    Fixed Borgnakke–Larsen rotational relaxation parameter for N2 (§5).
axioms (4)
  • domain assumption DSMC with VSS/NTC and Borgnakke–Larsen is a valid stochastic approximation of the Boltzmann equation for the monatomic/polyatomic regimes considered.
    Standard kinetic-theory modeling basis for all particle evolution (§2.1).
  • domain assumption Macroscopic conservation laws closed by NSF constitutive relations plus HoTs sampled as DSMC minus NSF (Eqs. 11, 14–15) correctly accelerate and asymptotically preserve NSF while retaining rarefaction effects.
    Core DIG theory imported from prior work [32,34,35]; used throughout §2–3 without re-derivation.
  • ad hoc to paper Equal-weight neighbor reconstruction and multi-layer propagation for cells failing Kn_Δ,c ≤ Kn_Δ,max preserve acceptable macroscopic accuracy for the synthetic solve.
    Stability enhancement introduced in §3.2 / Fig. 1; not derived from first principles.
  • ad hoc to paper Particle count or smoothed wall-clock time is a sufficient vertex weight, and face-crossing counts a sufficient edge weight, for METIS to balance DIG particle work.
    Load model in §4.3.2; engineering choice validated only by reported efficiency curves.

pith-pipeline@v1.1.0-grok45 · 26859 in / 3421 out tokens · 46016 ms · 2026-07-10T11:30:56.746821+00:00 · methodology

0 comments
read the original abstract

While the direct simulation Monte Carlo (DSMC) is a mainstream stochastic particle method for simulating rarefied gas flows, it incurs excessively high computational costs in the near continuum regime. As a hybrid acceleration approach coupling DSMC with macroscopic synthetic equations, the direct intermittent general synthetic iterative scheme (DIG) delivers fast convergence and asymptotic-preserving characteristics, which effectively alleviate the kinetic scale limitations inherent to standard DSMC. In this study, we develop a parallel DIG augmented DSMC solver for three dimensional rarefied gas flow simulations on unstructured meshes. On top of the standard DSMC algorithms for particle transport and collisions, a reliable intermittent coupling framework is constructed to exchange macroscopic flow data between the stochastic DSMC module and deterministic macroscopic synthetic equations. For parallel execution on unstructured grids, we employ a hybrid MPI architecture equipped with ghost cells to enable local particle tracking and batch inter-rank particle migration. A graph partitioning based dynamic load balancing strategy is also integrated to mitigate uneven particle distribution over computational domains. Numerical results demonstrate that the proposed solver achieves satisfactory agreement with the SPARTA DSMC. Leveraging the fast convergence and asymptotic-preserving properties of the DIG method, the required number of spatial cells and statistical sampling steps are drastically decreased, leading to substantial reductions in computational memory and runtime. This work presents an efficient high-performance numerical tool for high-fidelity simulations of rarefied flows over complex geometries. The code is available in the developer repository at the github link.

Figures

Figures reproduced from arXiv: 2607.08195 by Hong Deng, Lei Wu, Liyan Luo, Tao Huang.

Figure 1
Figure 1. Figure 1: Illustration of the local reconstruction procedure for cells rejected by criterion (19). Blue cells [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: MPI process organization in the DIG solver. The global mesh is distributed over [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Particle migration across partition interfaces and batched MPI communication. (a) Particle [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The flowchart of the proposed dynamic mesh re-partitioning procedure. At every [PITH_FULL_IMAGE:figures/full_fig_p016_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Graph-based mesh re-partitioning using METIS in the DIG method. (a) Schematic illustration of [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of density (first column), velocity magnitude (second column), and total temperature [PITH_FULL_IMAGE:figures/full_fig_p019_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Normalized horizontal and vertical velocity profiles on the centerlines of the middle plane [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Macroscopic quantities in the hypersonic flow over a sphere on the [PITH_FULL_IMAGE:figures/full_fig_p022_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: (a-c) The density, streamwise velocity, and temperature along the windward stagnation line in [PITH_FULL_IMAGE:figures/full_fig_p023_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Dynamic mesh re-partitioning for hypersonic flow over a sphere by 40 MPI processes. (a) and [PITH_FULL_IMAGE:figures/full_fig_p026_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: DIG results in the hypersonic flow around an Apollo reentry capsule for [PITH_FULL_IMAGE:figures/full_fig_p027_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Comparisons of density, velocity and temperature from [PITH_FULL_IMAGE:figures/full_fig_p028_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: First row: Geometric model and surface mesh for the hypersonic flow of [PITH_FULL_IMAGE:figures/full_fig_p029_13.png] view at source ↗
Figure 2
Figure 2. Figure 2: Starting from this steady particle state, [PITH_FULL_IMAGE:figures/full_fig_p030_2.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages · 1 internal anchor

  1. [1]

    Votta, A

    R. Votta, A. Schettino, A. Bonfiglioli, Hypersonic high altitude aerothermodynamics of a space re-entry vehicle, Aerospace Science and Technology 25 (1) (2013) 253–265

    work page 2013

  2. [2]

    M. S. Ivanov, S. F. Gimelshein, Computational hypersonic rarefied flows, Annual Review of Fluid Mechanics 30 (1) (1998) 469–505

    work page 1998

  3. [3]

    Tantos, S

    C. Tantos, S. Varoutis, V. Hauer, C. Day, P. Innocente, 3D numerical study of neutral gas dynamics in the DTT particle exhaust using the DSMC method, Nuclear Fusion 64 (1) (2024) 016019

    work page 2024

  4. [4]

    Q. Wang, K. Wang, X. Wu, Z. Gao, Numerical investigation of flow and particles contamination in reticle mini environment for extreme ultraviolet lithography, Journal of Vacuum Science & Technology B 42 (5) (2024) 052602. 32

    work page 2024

  5. [5]

    I.D.Boyd, G.V.Chen, G.V.Candler, Predictingfailureofthecontinuumfluidequationsintransitional hypersonic flows, Physics of Fluids 7 (1) (1995) 210–219

    work page 1995

  6. [6]

    A. J. Lofthouse, L. C. Scalabrin, I. D. Boyd, Velocity slip and temperature jump in hypersonic aerother- modynamics, Journal of Thermophysics and Heat Transfer 22 (1) (2008) 38–49

    work page 2008

  7. [7]

    V. V. Aristov, Direct Methods for Solving the Boltzmann Equation and Study of Nonequilibrium Flows, Springer Dordrecht, 2001

    work page 2001

  8. [8]

    P. Wang, M. T. Ho, L. Wu, Z. Guo, Y. Zhang, A comparative study of discrete velocity methods for low-speed rarefied gas flows, Computers & Fluids 161 (2018) 33–46

    work page 2018

  9. [9]

    T. E. Schwartzentruber, L. C. Scalabrin, I. D. Boyd, A modular particle-continuum numerical method for hypersonic non-equilibrium gas flows, Journal of Computational Physics 225 (1) (2007) 1159–1174

    work page 2007

  10. [10]

    G. A. Bird, Molecular Gas Dynamics and the Direct Simulation of Gas Flows, Oxford University Press, 1994

    work page 1994

  11. [11]

    White, M

    C. White, M. K. Borg, T. J. Scanlon, S. M. Longshaw, B. John, D. R. Emerson, J. M. Reese, dsmc- Foam+: An OpenFOAM based direct simulation Monte Carlo solver, Computer Physics Communica- tions 224 (2018) 22–43

    work page 2018

  12. [12]

    Borgnakke, P

    C. Borgnakke, P. S. Larsen, Statistical collision model for Monte Carlo simulation of polyatomic gas mixture, Journal of Computational Physics 18 (4) (1975) 405–420

    work page 1975

  13. [13]

    K. C. Kannenberg, I. D. Boyd, Strategies for efficient particle resolution in the direct simulation Monte Carlo method, Journal of Computational Physics 157 (2) (2000) 727–745

    work page 2000

  14. [14]

    Z.-X. Sun, Z. Tang, Y.-L. He, W.-Q. Tao, Proper cell dimension and number of particles per cell for DSMC, Computers & Fluids 50 (1) (2011) 1–9

    work page 2011

  15. [15]

    G. Chen, I. D. Boyd, Statistical error analysis for the direct simulation Monte Carlo technique, Journal of Computational Physics 126 (2) (1996) 434–448

    work page 1996

  16. [16]

    N. G. Hadjiconstantinou, A. L. Garcia, M. Z. Bazant, G. He, Statistical error in particle simulations of hydrodynamic phenomena, Journal of Computational Physics 187 (1) (2003) 274–297

    work page 2003

  17. [17]

    Q. Sun, I. D. Boyd, Evaluation of macroscopic properties in the direct simulation Monte Carlo method, Journal of Thermophysics and Heat Transfer 19 (3) (2005) 329–335

    work page 2005

  18. [18]

    Wu, Rarefied Gas Dynamics: Kinetic Modeling and Multi-Scale Simulation, Springer Nature Singa- pore, 2022

    L. Wu, Rarefied Gas Dynamics: Kinetic Modeling and Multi-Scale Simulation, Springer Nature Singa- pore, 2022

    work page 2022

  19. [19]

    Pareschi, G

    L. Pareschi, G. Russo, Time relaxed Monte Carlo methods for the Boltzmann equation, SIAM Journal on Scientific Computing 23 (4) (2002) 1253–1273

    work page 2002

  20. [20]

    W. Ren, H. Liu, S. Jin, An asymptotic-preserving Monte Carlo method for the Boltzmann equation, Journal of Computational Physics 276 (2014) 380–404

    work page 2014

  21. [21]

    Fei, A time-relaxed Monte Carlo method preserving the Navier–Stokes asymptotics, Journal of Computational Physics 486 (2023) 112128

    F. Fei, A time-relaxed Monte Carlo method preserving the Navier–Stokes asymptotics, Journal of Computational Physics 486 (2023) 112128

    work page 2023

  22. [22]

    Degond, G

    P. Degond, G. Dimarco, L. Pareschi, The moment-guided Monte Carlo method, International Journal for Numerical Methods in Fluids 67 (2) (2011) 189–213

    work page 2011

  23. [23]

    M. H. Gorji, M. Torrilhon, P. Jenny, Fokker–Planck model for computational studies of monatomic rarefied gas flows, Journal of Fluid Mechanics 680 (2011) 574–601

    work page 2011

  24. [24]

    Xu, J.-C

    K. Xu, J.-C. Huang, A unified gas-kinetic scheme for continuum and rarefied flows, Journal of Compu- tational Physics 229 (20) (2010) 7747–7764

    work page 2010

  25. [25]

    Y. Zhu, C. Zhong, K. Xu, Implicit unified gas-kinetic scheme for steady state solutions in all flow regimes, Journal of Computational Physics 315 (2016) 16–38

    work page 2016

  26. [26]

    Z. Guo, K. Xu, R. Wang, Discrete unified gas kinetic scheme for all Knudsen number flows: Low-speed isothermal case, Physical Review E 88 (3) (2013) 033305

    work page 2013

  27. [27]

    C. Liu, Y. Zhu, K. Xu, Unified gas-kinetic wave-particle methods I: Continuum and rarefied gas flow, Journal of Computational Physics 401 (2020) 108977. 33

    work page 2020

  28. [28]

    Y. Chen, Y. Zhu, K. Xu, A three-dimensional unified gas-kinetic wave-particle solver for flow compu- tation in all regimes, Physics of Fluids 32 (9) (2020) 096108

    work page 2020

  29. [29]

    W. Long, Y. Wei, K. Xu, Nonequilibrium flow simulations using unified gas-kinetic wave-particle method, AIAA Journal 62 (4) (2024) 1411–1433

    work page 2024

  30. [30]

    W. Su, L. Zhu, P. Wang, Y. Zhang, L. Wu, Can we find steady-state solutions to multiscale rarefied gas flows within dozens of iterations?, Journal of Computational Physics 407 (2020) 109245

    work page 2020

  31. [31]

    W. Su, L. Zhu, L. Wu, Fast convergence and asymptotic preserving of the general synthetic iterative scheme, SIAM Journal on Scientific Computing 42 (6) (2020) B1517–B1540

    work page 2020

  32. [32]

    L. Luo, L. Wu, Multiscale simulation of rarefied gas dynamics via direct intermittent gsis-dsmc coupling, Advances in Aerodynamics 6 (1) (2024) 22

    work page 2024

  33. [33]

    C. D. Hauck, M. P. Laiu, S. R. Schnake, On high-order/low-order and micro-macro methods for implicit time-stepping of the BGK model, SIAM Journal on Scientific Computing 47 (6) (2025) A3566–A3593

    work page 2025

  34. [34]

    L. Luo, T. Huang, Q. Li, L. Wu, Multiscale simulation of rarefied polyatomic gas flow via DIG method, Acta Mechanica Sinica 42 (2) (2026) 725053

    work page 2026

  35. [35]

    L. Luo, J. Zeng, Y. Zhang, W. Li, Q. Li, L. Wu, Enhancing DSMC simulations of rarefied gas mixtures using a fast-converging and asymptotic-preserving scheme, Computer Methods in Applied Mechanics and Engineering 449 (2026) 118508

    work page 2026

  36. [36]

    H. Deng, L. Y. Luo, L. Wu, Efficient simulation of chemical reaction in DSMC, arXiv:2605.13653 (2026)

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  37. [37]

    C. S. Wang-Chang, G. E. Uhlenbeck, Transport Phenomena in Polyatomic Gases, University of Michi- gan Engineering Research Rept. No. CM-681, 1951

    work page 1951

  38. [38]

    Chapman, T

    S. Chapman, T. Cowling, The Mathematical Theory of Non-Uniform Gases, 3rd Edition, Cambridge University Press, 1970

    work page 1970

  39. [39]

    Mason, L

    E. Mason, L. Monchick, Heat conductivity of polyatomic and polar gases, Journal of Chemical Physics 36 (6) (1962) 1622–1639

    work page 1962

  40. [40]

    Q. Li, J. N. Zeng, W. Su, L. Wu, Uncertainty quantification in rarefied dynamics of molecular gas: rate effect of thermal relaxation, J. Fluid Mech. 917 (2021) A58

    work page 2021

  41. [41]

    Jenny, M

    P. Jenny, M. Torrilhon, S. Heinz, A solution algorithm for the fluid dynamic equations based on a stochastic model for molecular motion, Journal of Computational Physics 229 (4) (2009) 1077–1098

    work page 2009

  42. [42]

    B. Hu, L. Y. Luo, L. Wu, Accelerating the Monte Carlo simulation of the Enskog equation for multiscale dense gas flows, J. Comput. Phys. 562 (2026) 114989

    work page 2026

  43. [43]

    Zhang, J

    Y. Zhang, J. Zeng, R. Yuan, W. Liu, Q. Li, L. Wu, Efficient parallel solver for rarefied gas flow using GSIS, Computers & Fluids 281 (2024) 106374

    work page 2024

  44. [44]

    Chordá, J

    R. Chordá, J. A. Blasco, N. Fueyo, An efficient particle-locating algorithm for application in arbitrary 2D and 3D grids, International Journal of Multiphase Flow 28 (9) (2002) 1565–1580

    work page 2002

  45. [45]

    G. B. Macpherson, N. Nordin, H. G. Weller, Particle tracking in unstructured, arbitrary polyhedral meshes for use in CFD and molecular dynamics, Communications in Numerical Methods in Engineering 25 (3) (2009) 263–273

    work page 2009

  46. [46]

    Wu, K.-C

    J.-S. Wu, K.-C. Tseng, Parallel DSMC method using dynamic domain decomposition, International Journal for Numerical Methods in Engineering 63 (1) (2005) 37–76

    work page 2005

  47. [47]

    Zhang, M

    C. Zhang, M. Wen, B. Zhang, J. Lin, H. Liu, A load-decoupling parallel strategy based on shared memory architecture for DSMC to simulate near-continuum gases, Computer Physics Communications 279 (2022) 108466

    work page 2022

  48. [48]

    Shamseddine, I

    M. Shamseddine, I. Lakkis, A novel spatio-temporally adaptive parallel three-dimensional DSMC solver for unsteady rarefied micro/nano gas flows, Computers & Fluids 186 (2019) 1–14

    work page 2019

  49. [49]

    J. Li, X. R. Geng, D. W. Jiang, J. Q. Chen, Dynamic load balance scheme for the DSMC algorithm, in: AIP Conference Proceedings, Vol. 1628, 2014, pp. 288–295

    work page 2014

  50. [50]

    H. W. Kuhn, The Hungarian method for the assignment problem, Naval Research Logistics Quarterly 2 (1-2) (1955) 83–97. 34

    work page 1955

  51. [51]

    Munkres, Algorithms for the assignment and transportation problems, Journal of the Society for Industrial and Applied Mathematics 5 (1) (1957) 32–38

    J. Munkres, Algorithms for the assignment and transportation problems, Journal of the Society for Industrial and Applied Mathematics 5 (1) (1957) 32–38

    work page 1957

  52. [52]

    B. Hu, L. Luo, K. Wang, L. Wu, Fast-converging and asymptotic-preserving DSMC, arXiv (2025). arXiv:2511.19061. 35

    work page arXiv 2025