pith. sign in

arxiv: 2602.13601 · v2 · pith:IXQDYO2Onew · submitted 2026-02-14 · ⚛️ physics.flu-dyn

Resolving Cryogenic and Hypersonic Rarefied Flows via Deep Learning-Accelerated Lennard-Jones DSMC

Ahmad Shoja Sani , Ehsan Roohi , Stefan Stefanov This is my paper

Pith reviewed 2026-05-25 07:09 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn
keywords Direct Simulation Monte CarloLennard-JonesDeepONetrarefied gascryogenic flowhypersonic flowscattering anglevariable effective diameter
0
0 comments X

The pith

A DeepONet predicts Lennard-Jones deflection angles to accelerate DSMC by 40% while a viscosity-based diameter model captures attractive-force effects in rarefied flows.

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

The authors aim to make the Lennard-Jones potential usable in DSMC for rarefied flows by solving two problems: defining collision rates and computing scattering angles efficiently. They create a variable effective diameter model that matches local viscosity to set collision probabilities correctly across temperatures. They also train a DeepONet to output deflection angles directly from impact parameters and relative speeds, avoiding repeated numerical integrations. If successful, this lets simulations capture how attractive molecular forces change flow behavior in cold or high-speed conditions, such as lower friction and wider wakes, which standard models miss. This matters for accurate modeling of spacecraft reentry or cryogenic vacuum systems.

Core claim

By formulating a Variable Effective Diameter model through local Chapman-Enskog viscosity matching, the framework supplies a finite collision-rate closure for the LJ potential in Bird's DSMC algorithms. The DeepONet surrogate then predicts the LJ deflection angle from high-fidelity data, achieving a mean wrapped-angle error of 1.6 times 10 to the minus 3 radians and accelerating the collision step by 40 percent. Validation in helium and argon shocks, cryogenic Couette flow, and hypersonic cylinder flows confirms agreement with VHS at high temperatures while revealing LJ-specific reductions in shear stress and increases in wake size, with further checks in diffusion problems.

What carries the argument

Variable Effective Diameter model derived from Chapman-Enskog viscosity matching for collision selection, together with a Deep Operator Network surrogate replacing the Matsumoto-Koura integral for scattering angles.

If this is right

  • The VED model yields collision rates that produce transport properties consistent with viscosity in the examined regimes.
  • LJ-DSMC predicts reduced shear stress relative to VHS when attractive forces are active in cryogenic flows.
  • Hypersonic cylinder flows develop larger wakes under the LJ potential than under VHS.
  • The surrogate maintains accuracy in diffusion benchmarks outside pure viscosity-controlled regimes.

Where Pith is reading between the lines

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

  • The method could be applied to other long-range potentials to avoid custom numerical scattering calculations in DSMC.
  • Larger computational domains or higher particle counts become feasible due to the 36 percent wall-time reduction.
  • Coupling the surrogate with adaptive sampling might further improve accuracy in regions with extreme velocity distributions.

Load-bearing premise

The Variable Effective Diameter model obtained from local Chapman-Enskog viscosity matching supplies a finite and physically valid DSMC collision-rate closure for the LJ potential in the cryogenic and hypersonic regimes examined.

What would settle it

Running an independent molecular dynamics simulation of cryogenic supersonic Couette flow and finding that the shear stress or velocity profile from the VED-LJ DSMC deviates substantially from the MD result.

Figures

Figures reproduced from arXiv: 2602.13601 by Ahmad Shoja Sani, Ehsan Roohi, Stefan Stefanov.

Figure 1
Figure 1. Figure 1: Schematic of the DeepONet architecture. 3.3 Training dataset and sampling strategy Both machine-learning surrogate models are trained on high-fidelity scattering data generated from the exact Lennard–Jones formulation described in Section 2. The training dataset consists of pairs [PITH_FULL_IMAGE:figures/full_fig_p017_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Predictive performance of the DeepONet: 𝜒𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑vs. 𝜒𝑡𝑟𝑢𝑒. 3.4 Integration into the DSMC collision procedure Once trained, the surrogate models are integrated directly into the DSMC collision algorithm by replacing the numerical evaluation of the LJ scattering integral during the velocity transformation step. The pair selection procedure and the overall structure of the DSMC algorithm are described in Se… view at source ↗
read the original abstract

Integrating the physically realistic Lennard--Jones (LJ) potential into Direct Simulation Monte Carlo (DSMC) remains challenging because the long-range potential complicates collision-rate definition and makes repeated scattering-angle evaluation expensive. This study develops an LJ--DSMC framework built around two methodological advances and a transport-level validation of the resulting collision kernel. First, a generalized collision-selection treatment is formulated for Bird's DSMC algorithms (DSMC1, DSMC1S, and DS2V) through a Variable Effective Diameter (VED) model obtained from local Chapman--Enskog viscosity matching. This viscosity-consistent pair-selection model provides a finite DSMC collision-rate closure for the LJ potential and is validated in helium and argon normal shocks, cryogenic supersonic Couette flow, and hypersonic cylinder flows. The results show agreement with VHS in high-temperature repulsive regimes, but reveal clear LJ effects, including reduced shear stress and larger cryogenic wakes, when attractive forces become important. Second, the computational bottleneck of the accepted LJ binary-scattering step is removed by training a Deep Operator Network (DeepONet) to predict the LJ deflection angle from high-fidelity scattering data, replacing the numerical Matsumoto--Koura integral while preserving the standard elastic post-collision update. The surrogate gives a bulk mean wrapped-angle error of \(1.6\times10^{-3}\,\mathrm{rad}\) and a 99th-percentile error of \(9.9\times10^{-3}\,\mathrm{rad}\), accelerates the collision subroutine by 40\%, and reduces total wall time by 36\%. Finally, the same DeepONet--LJ scattering kernel is tested beyond viscosity-controlled flows through diffusion benchmarks.

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 / 2 minor

Summary. The manuscript develops an LJ-DSMC framework for rarefied flows. It introduces a Variable Effective Diameter (VED) model derived from local Chapman-Enskog viscosity matching to define collision rates for the Lennard-Jones potential within Bird's DSMC algorithms, and trains a DeepONet surrogate to predict LJ deflection angles from scattering data, replacing the Matsumoto-Koura integral. Validation is reported in helium/argon normal shocks, cryogenic supersonic Couette flow, and hypersonic cylinder flows, with claims of agreement with VHS in repulsive regimes but LJ-specific effects (reduced shear stress, larger wakes) when attractive forces matter; the DeepONet achieves a mean wrapped-angle error of 1.6e-3 rad, 99th-percentile error of 9.9e-3 rad, 40% collision-subroutine speedup, and 36% total wall-time reduction. Diffusion benchmarks are used to test beyond viscosity-controlled regimes.

Significance. If the VED closure holds, the framework enables efficient inclusion of realistic long-range LJ physics in DSMC for cryogenic and hypersonic rarefied flows where attractive forces alter transport properties. The quantified DeepONet performance (specific error metrics and speedups) and extension to diffusion benchmarks constitute clear computational and methodological advances that could be adopted in existing DSMC codes.

major comments (2)
  1. [Abstract (VED formulation and validation cases)] The VED collision-rate closure is obtained from local Chapman-Enskog viscosity matching (Abstract). This assumes a local Maxwellian to equate viscosity, yet the highlighted test cases (cryogenic Couette and hypersonic cylinder flows) feature strong translational non-equilibrium and attractive-force effects. The manuscript must show that the resulting pair-selection probability remains consistent with the true LJ collision integral evaluated on the actual non-equilibrium distributions; otherwise the reported shear-stress reductions and wake-size changes rest on an uncontrolled approximation rather than a controlled closure.
  2. [Validation sections (normal shocks, Couette, cylinder)] Table or figure data for the flow validations (normal shocks, Couette, cylinder) are not provided with full baseline comparisons, error bars, or quantitative metrics for shear stress and wake properties. Without these, it is not possible to judge whether the claimed LJ-specific deviations from VHS exceed numerical or modeling uncertainty.
minor comments (2)
  1. The abstract states quantitative error metrics for the DeepONet but does not include the corresponding full data tables or baseline comparisons needed to assess robustness across regimes.
  2. Notation for the wrapped-angle error and the precise definition of the VED should be cross-referenced to the relevant equations for clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the referee's insightful comments on our manuscript. We address each major comment below and outline the revisions we will make.

read point-by-point responses

  1. Referee: [Abstract (VED formulation and validation cases)] The VED collision-rate closure is obtained from local Chapman-Enskog viscosity matching (Abstract). This assumes a local Maxwellian to equate viscosity, yet the highlighted test cases (cryogenic Couette and hypersonic cylinder flows) feature strong translational non-equilibrium and attractive-force effects. The manuscript must show that the resulting pair-selection probability remains consistent with the true LJ collision integral evaluated on the actual non-equilibrium distributions; otherwise the reported shear-stress reductions and wake-size changes rest on an uncontrolled approximation rather than a controlled closure.

    Authors: The VED closure is formulated using local cell-averaged properties to determine the effective diameter, following the standard practice in DSMC for models like VHS where collision rates are based on local temperature and density. This local Maxwellian assumption for the rate is an inherent feature of the method to enable efficient sampling. The non-equilibrium effects are captured through the particle velocities and the scattering. Our validations in the non-equilibrium cases (shocks, Couette, cylinder) show physically consistent results with expected LJ attractive force effects. To strengthen the justification, we will include in the revision a direct numerical check: sampling particle pairs from the actual simulated non-equilibrium distribution in a representative cell from the cylinder flow and comparing the VED selection probability to the integral over the true LJ cross-section. revision: yes

  2. Referee: [Validation sections (normal shocks, Couette, cylinder)] Table or figure data for the flow validations (normal shocks, Couette, cylinder) are not provided with full baseline comparisons, error bars, or quantitative metrics for shear stress and wake properties. Without these, it is not possible to judge whether the claimed LJ-specific deviations from VHS exceed numerical or modeling uncertainty.

    Authors: We acknowledge the need for more quantitative presentation. In the revised version, we will add tables summarizing the key quantities (e.g., shock thickness, shear stress profiles, wake recirculation length) with comparisons to VHS, including mean differences and standard deviations from ensemble runs to provide uncertainty estimates. Error bars will be added to relevant figures, and baseline data will be explicitly tabulated. revision: yes

Circularity Check

1 steps flagged

VED collision-rate closure is constructed via Chapman-Enskog viscosity matching, rendering validation in viscosity-controlled flows partly tautological

specific steps
  1. fitted input called prediction [Abstract]
    "a generalized collision-selection treatment is formulated for Bird's DSMC algorithms (DSMC1, DSMC1S, and DS2V) through a Variable Effective Diameter (VED) model obtained from local Chapman--Enskog viscosity matching. This viscosity-consistent pair-selection model provides a finite DSMC collision-rate closure for the LJ potential and is validated in helium and argon normal shocks, cryogenic supersonic Couette flow, and hypersonic cylinder flows. The results show agreement with VHS in high-temperature repulsive regimes, but reveal clear LJ effects, including reduced shear stress and larger cryo-"

    The VED pair-selection probability is explicitly fitted to reproduce the Chapman-Enskog viscosity; therefore the reported shear-stress reduction and wake-size changes in the viscosity-governed test cases follow directly from the matching definition rather than constituting an independent test of LJ physics.

full rationale

The paper's load-bearing closure for LJ-DSMC collision rates is obtained by construction from local Chapman-Enskog viscosity matching, then applied to the very class of flows (normal shocks, Couette, cylinder) whose dynamics are controlled by that same viscosity. This satisfies the fitted-input-called-prediction pattern for the transport results (shear stress, wakes) while the DeepONet surrogate and diffusion benchmarks remain independent. The central claim therefore contains a self-referential component rather than a fully first-principles derivation.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Central claim rests on the domain assumption that viscosity matching yields a valid collision closure and on the empirical generalization of a neural surrogate trained on unspecified high-fidelity scattering data.

free parameters (1)
  • DeepONet weights and architecture hyperparameters
    Determined by training on high-fidelity LJ scattering data; specific values not reported in abstract.
axioms (1)
  • domain assumption Chapman-Enskog theory supplies the reference viscosity used to define the Variable Effective Diameter
    Invoked to obtain the collision-rate closure for the LJ potential.

pith-pipeline@v0.9.0 · 5847 in / 1260 out tokens · 38933 ms · 2026-05-25T07:09:25.888521+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

55 extracted references · 55 canonical work pages

  1. [1]

    Roohi, E., Akhlaghi, H., & Stefanov, S. (2025). Advances in Direct Simulation Monte Carlo: From Micro-Scale to Rarefied Flow Phenomena. Springer

    work page 2025

  2. [2]

    Bird, Approach to Translational Equilibrium in a Rigid Sphere Gas, Phys

    G.A. Bird, Approach to Translational Equilibrium in a Rigid Sphere Gas, Phys. Fluids 6 (1963) 1518–1519. https://doi.org/10.1063/1.1710976

    work page doi:10.1063/1.1710976 1963

  3. [3]

    G.A Bird , 1981, Monte -Carlo simulation in an engineering context, In: Rarefied gas dynamics; International Symposium, 12th, Charlottesville, VA, July 7-11, 1980, Technical Papers. Part 1. (A82 -13026 03 -77) New York, American Institute of Aeronautics and Astronautics, 1981, p. 239-255

    work page 1981

  4. [4]

    Bird, Definition of mean free path for real gases, Phys

    G.A. Bird, Definition of mean free path for real gases, Phys. Fluids 26 (1983) 3222 –3223. https://doi.org/10.1063/1.864095

    work page doi:10.1063/1.864095 1983

  5. [5]

    Chapman, T.G

    S. Chapman, T.G. Cowling, The Mathematical Theory of Non-uniform Gases: An Account 59 of the Kinetic Theory of Viscosity, Thermal Conduction and Diffusion in Gases, Cambridge University Press, 1970. https://books.google.com/books?id=Cbp5JP2OTrwC

    work page 1970

  6. [6]

    Koura, H

    K. Koura, H. Matsumoto, T. Shimada, A test of equivalence of the variable-hard-sphere and inverse-power-law models in the direct -simulation Monte Carlo method, Phys. Fluids A Fluid Dyn. 3 (1991) 1835–1837. https://doi.org/10.1063/1.857912

    work page doi:10.1063/1.857912 1991

  7. [7]

    Koura, H

    K. Koura, H. Matsumoto, Variable soft sphere molecular model for inverse‐power‐law or Lennard‐Jones potential, Phys. Fluids A Fluid Dyn. 3 (1991) 2459 –2465. https://doi.org/10.1063/1.858184

    work page doi:10.1063/1.858184 1991

  8. [8]

    Hassan, D.B

    H.A. Hassan, D.B. Hash, A generalized hard -sphere model for Monte Carlo simulation, Phys. Fluids A Fluid Dyn. 5 (1993) 738–744. https://doi.org/10.1063/1.858656

    work page doi:10.1063/1.858656 1993

  9. [9]

    Fan, A generalized soft-sphere model for Monte Carlo simulation, Phys

    J. Fan, A generalized soft-sphere model for Monte Carlo simulation, Phys. Fluids 14 (2002) 4399–4405. https://doi.org/10.1063/1.1521123

    work page doi:10.1063/1.1521123 2002

  10. [10]

    Hirschfelder, R.B

    J.O. Hirschfelder, R.B. Bird, E.L. Spotz, The Transport Properties for Non -Polar Gases, J. Chem. Phys. 16 (1948) 968–981. https://doi.org/10.1063/1.1746696

    work page doi:10.1063/1.1746696 1948

  11. [11]

    Pham -Van-Diep, D

    G. Pham -Van-Diep, D. Erwin, E.P. Muntz, Nonequilibrium Molecular Motion in a Hypersonic Shock Wave, Science 80. 245 (1989) 624 –626. https://doi.org/10.1126/SCIENCE.245.4918.624

    work page doi:10.1126/science.245.4918.624 1989

  12. [12]

    Lengrand, J

    J.C. Lengrand, J. Allegre, A. Chpoun, M. Raffin, Rarefied hypersonic flow over a sharp flat plate: numerical and experimental results, Prog. Astronaut. Aeronaut. 160 (1994) 276

    work page 1994

  13. [13]

    Holden, Holden, M

    M.S. Holden, Holden, M. S., Development and code evaluation studies in hypervelocity flows in the lens facility, ESASP 367 (1995) 319. https://ui.adsabs.harvard.edu/abs/1995ESASP.367..319H/abstract (accessed June 18, 2024)

    work page 1995

  14. [14]

    Matsumoto, K

    H. Matsumoto, K. Koura, Comparison of velocity distribution functions in an argon shock wave between experiments and Monte Carlo calculations for Lennard-Jones potential, Phys. Fluids A Fluid Dyn. 3 (1991) 3038–3045. https://doi.org/10.1063/1.857848

    work page doi:10.1063/1.857848 1991

  15. [15]

    Ayyaswamy, A

    V. Ayyaswamy, A. Alexeenko, DSMC Collision Model for the Lennard -Jones Potential: Efficient Algorithm and Verification, in: 42nd AIAA Thermophys. Conf., American Institute of Aeronautics and Astronautics, Reston, Virigina, 2011. https://doi.org/10.2514/6.2011-3313

    work page doi:10.2514/6.2011-3313 2011

  16. [16]

    Venkattraman, A.A

    A. Venkattraman, A.A. Alexeenko, Binary scattering model for Lennard -Jones potential: 60 Transport coefficients and collision integrals for non -equilibrium gas flow simulations, Phys. Fluids 24 (2012). https://doi.org/10.1063/1.3682375

    work page doi:10.1063/1.3682375 2012

  17. [17]

    Sharipov, J.L

    F. Sharipov, J.L. Strapasson, Direct simulation Monte Carlo method for an arbitrary intermolecular potential, Phys. Fluids 24 (2012). https://doi.org/10.1063/1.3676060

    work page doi:10.1063/1.3676060 2012

  18. [18]

    Sharipov, F.C

    F. Sharipov, F.C. Dias, Ab initio simulation of planar shock waves, Comput. Fluids 150 (2017) 115–122. https://doi.org/10.1016/j.compfluid.2017.04.002

    work page doi:10.1016/j.compfluid.2017.04.002 2017

  19. [19]

    Stefanov, On DSMC Calculations of Rarefied Gas Flows with Small Number of Particles in Cells, SIAM J

    S.K. Stefanov, On DSMC Calculations of Rarefied Gas Flows with Small Number of Particles in Cells, SIAM J. Sci. Comput. 33 (2011) 677 –702. https://doi.org/10.1137/090751864

    work page doi:10.1137/090751864 2011

  20. [20]

    Roohi, S

    E. Roohi, S. Stefanov, A. Shoja -Sani, H. Ejraei, A generalized form of the Bernoulli Trial collision scheme in DSMC: Derivation and evaluation, J. Comput. Phys. 354 (2018) 476 –

    work page 2018

  21. [21]

    https://doi.org/10.1016/j.jcp.2017.10.033

    work page doi:10.1016/j.jcp.2017.10.033 2017

  22. [22]

    Bird, Molecular Gas Dynamics and the Direct Simulation of Gas Flows, Clarendon Press, 1994

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

    work page 1994

  23. [23]

    Weaver, A.A

    A.B. Weaver, A.A. Alexeenko, Revised Variable Soft Sphere and Lennard -Jones Model Parameters for Eight Common Gases up to 2200 K, J. Phys. Chem. Ref. Data 44 (2015). https://doi.org/10.1063/1.4921245

    work page doi:10.1063/1.4921245 2015

  24. [24]

    Raissi, P

    M. Raissi, P. Perdikaris, G.E. Karniadakis, Physics -informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, J. Comput. Phys. 378 (2019) 686 –707. https://doi.org/10.1016/j.jcp.2018.10.045

    work page doi:10.1016/j.jcp.2018.10.045 2019

  25. [25]

    Cuomo, V.S

    S. Cuomo, V.S. Di Cola, F. Giampaolo, G. Rozza, M. Raissi, F. Piccialli, Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next, J. Sci. Comput. 92 (2022) 88. https://doi.org/10.1007/s10915-022-01939-z

    work page doi:10.1007/s10915-022-01939-z 2022

  26. [26]

    Farea, O

    A. Farea, O. Yli -Harja, F. Emmert -Streib, Understanding Physics -Informed Neural Networks: Techniques, Applications, Trends, and Challenges, AI 5 (2024) 1534 –1557. https://doi.org/10.3390/ai5030074

    work page doi:10.3390/ai5030074 2024

  27. [27]

    C. Zhao, F. Zhang, W. Lou, X. Wang, J. Yang, A comprehensive review of advances in physics-informed neural networks and their applications in complex fluid dynamics, Phys. Fluids 36 (2024). https://doi.org/10.1063/5.0226562

    work page doi:10.1063/5.0226562 2024

  28. [28]

    L. Lu, P. Jin, G. Pang, Z. Zhang, G.E. Karniadakis, Learning nonlinear operators via 61 DeepONet based on the universal approximation theorem of operators, Nat. Mach. Intell. 3 (2021) 218–229. https://doi.org/10.1038/s42256-021-00302-5

    work page doi:10.1038/s42256-021-00302-5 2021

  29. [29]

    S. Wang, H. Wang, P. Perdikaris, Learning the solution operator of parametric partial differential equations with physics -informed DeepONets, Sci. Adv. 7 (2021). https://doi.org/10.1126/sciadv.abi8605

    work page doi:10.1126/sciadv.abi8605 2021

  30. [30]

    T. Xiao, M. Frank, Using neural networks to accelerate the solution of the Boltzmann equation, J. Comput. Phys. 443 (2021) 110521. https://doi.org/10.1016/j.jcp.2021.110521

    work page doi:10.1016/j.jcp.2021.110521 2021

  31. [31]

    De Florio, E

    M. De Florio, E. Schiassi, B.D. Ganapol, R. Furfaro, Physics -Informed Neural Networks for rarefied -gas dynamics: Poiseuille flow in the BGK approximation, Zeitschrift Für Angew. Math. Und Phys. 73 (2022) 126. https://doi.org/10.1007/s00033-022-01767-z

    work page doi:10.1007/s00033-022-01767-z 2022

  32. [32]

    Miller, N

    S.T. Miller, N. V. Roberts, S.D. Bond, E.C. Cyr, Neural-network based collision operators for the Boltzmann equation, J. Comput. Phys. 470 (2022) 111541. https://doi.org/10.1016/j.jcp.2022.111541

    work page doi:10.1016/j.jcp.2022.111541 2022

  33. [33]

    Roohi, A

    E. Roohi, A. Mahdavi, Shock -Aware Physics -Guided Fusion -DeepONet Operator for Rarefied Micro-Nozzle Flows, (2025). http://arxiv.org/abs/2510.17887 (accessed February 9, 2026)

    work page arXiv 2025

  34. [34]

    Zheng, H

    J. Zheng, H. Hu, J. Huang, B. Zhao, H. Huang, CF -DeepONet: Deep operator neural networks for solving compressible flows, Aerosp. Sci. Technol. 163 (2025) 110329. https://doi.org/10.1016/j.ast.2025.110329

    work page doi:10.1016/j.ast.2025.110329 2025

  35. [35]

    Peyvan, V

    A. Peyvan, V. Kumar, G.E. Karniadakis, Fusion -DeepONet: A data -efficient neural operator for geometry -dependent hypersonic and supersonic flows, J. Comput. Phys. 544 (2026) 114432. https://doi.org/10.1016/j.jcp.2025.114432

    work page doi:10.1016/j.jcp.2025.114432 2026

  36. [36]

    N. D. Ball, J. F. MacArt, J. Sirignano, Online optimisation of machine learning collision models to accelerate direct molecular simulation of rarefied gas flows, Journal of Computational Physics, 549, (2026), 114601

    work page 2026

  37. [37]

    Roohi, A

    E. Roohi, A. Shoja -Sani, Data-driven surrogate modeling of DSMC solutions using deep neural networks, Aerosp. Sci. Technol. 168 (2026) 110785. https://doi.org/10.1016/j.ast.2025.110785

    work page doi:10.1016/j.ast.2025.110785 2026

  38. [38]

    Tatsios, A.K

    G. Tatsios, A.K. Chinnappan, A. Kamal, N. Vasileiadis, S.Y. Docherty, C. White, L. Gibelli, M.K. Borg, J.R. Kermode, D.A. Lockerby, A DSMC-CFD coupling method using surrogate modelling for low-speed rarefied gas flows, J. Comput. Phys. 520 (2025) 113500. 62 https://doi.org/10.1016/j.jcp.2024.113500

    work page doi:10.1016/j.jcp.2024.113500 2025

  39. [39]

    Roohi, A

    E. Roohi, A. Mahdavi, Analysis of the Rarefied Flow at Micro -Step using a DeepONet Surrogate Model with a Physics -Guided Zonal Loss Function, (2025). http://arxiv.org/abs/2509.17254 (accessed February 9, 2026)

    work page arXiv 2025

  40. [40]

    Q.-D. To, C. Soize, Size effects of rarefied gas flows in nanochannels by simulations based on statistical surrogate model for atomistic collisions, Phys. Rev. E 112 (2025) 025303. https://doi.org/10.1103/mcz3-4fkd

    work page doi:10.1103/mcz3-4fkd 2025

  41. [41]

    Stefanov, Particle Monte Carlo algorithms with small number of particles in grid cells, in: Lect

    S.K. Stefanov, Particle Monte Carlo algorithms with small number of particles in grid cells, in: Lect. Notes Comput. Sci. (Including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), 2011: pp. 110–117. https://doi.org/10.1007/978-3-642-18466-6_12

    work page doi:10.1007/978-3-642-18466-6_12 2011

  42. [42]

    Mason, JT Vanderslice in Atomic and Molecular Processes, DR Bates, Ed, (1962)

    E.A. Mason, JT Vanderslice in Atomic and Molecular Processes, DR Bates, Ed, (1962)

    work page 1962

  43. [43]

    Koura, Null -collision technique in the direct -simulation Monte Carlo method, Phys

    K. Koura, Null -collision technique in the direct -simulation Monte Carlo method, Phys. Fluids 29 (1986) 3509–3511. https://doi.org/10.1063/1.865826

    work page doi:10.1063/1.865826 1986

  44. [44]

    Koura, A sensitive test for accuracy in evaluation of molecular collision number in the direct-simulation Monte Carlo method, Phys

    K. Koura, A sensitive test for accuracy in evaluation of molecular collision number in the direct-simulation Monte Carlo method, Phys. Fluids A Fluid Dyn. 2 (1990) 1287 –1289. https://doi.org/10.1063/1.857577

    work page doi:10.1063/1.857577 1990

  45. [45]

    Muntz, L.N

    E.P. Muntz, L.N. Harnett, Molecular Velocity Distribution Function Measurements in a Normal Shock Wave, Phys. Fluids 12 (1969) 2027 –2035. https://doi.org/10.1063/1.1692308

    work page doi:10.1063/1.1692308 1969

  46. [46]

    Holtz, E.P

    T. Holtz, E.P. Muntz, Molecular velocity distribution functions in an argon normal shock wave at Mach number 7, Phys. Fluids 26 (1983) 2425 –2436. https://doi.org/10.1063/1.864428

    work page doi:10.1063/1.864428 1983

  47. [47]

    Erwin, G.C

    D.A. Erwin, G.C. Pham -Van-Diep, E.P. Muntz, Nonequilibrium gas flows. I: A detailed validation of Monte Carlo direct simulation for monatomic gases, Phys. Fluids A Fluid Dyn. 3 (1991) 697–705. https://doi.org/10.1063/1.858075

    work page doi:10.1063/1.858075 1991

  48. [48]

    Goshayeshi, E

    B. Goshayeshi, E. Roohi, S. Stefanov, A novel simplified bernoulli trials collision scheme in the direct simulation monte carlo with intelligence over particle distances, Phys. Fluids 27 (2015) 107104. https://doi.org/10.1063/1.4933251

    work page doi:10.1063/1.4933251 2015

  49. [49]

    Lofthouse, I.D

    A.J. Lofthouse, I.D. Boyd, M.J. Wright, Effects of continuum breakdown on hypersonic aerothermodynamics, Phys. Fluids 19 (2007) 027105. https://doi.org/10.1063/1.2710289

    work page doi:10.1063/1.2710289 2007

  50. [50]

    Shoja-sani, E

    A. Shoja-sani, E. Roohi, S. Stefanov, Evaluation of the generalized bernoulli trial-transient 63 adaptive subcell ( GBT -TAS ) collision scheme in treating rarefied gas flows, Comput. Fluids 213 (2020) 104740. https://doi.org/10.1016/j.compfluid.2020.104740

    work page doi:10.1016/j.compfluid.2020.104740 2020

  51. [51]

    Goshayeshi, E

    B. Goshayeshi, E. Roohi, S. Stefanov, DSMC simulation of hypersonic flows using an improved SBT -TAS technique, J. Comput. Phys. 303 (2015) 28 –44. https://doi.org/10.1016/j.jcp.2015.09.027

    work page doi:10.1016/j.jcp.2015.09.027 2015

  52. [52]

    Shoja-sani, E

    A. Shoja-sani, E. Roohi, M. Javani, H. Akhlaghi, S. Stefanov, Efficient collision algorithms in DSMC for rarefied gas dynamics: Markovian NTC-pre-scan and Bernoulli-trial schemes, Vacuum 247 (2026) 115138. https://doi.org/10.1016/j.vacuum.2026.115138

    work page doi:10.1016/j.vacuum.2026.115138 2026

  53. [53]

    Bird, The DSMC method, CreateSpace Independent Publishing Platform, 2013

    G.A. Bird, The DSMC method, CreateSpace Independent Publishing Platform, 2013

    work page 2013

  54. [54]

    Roohi, & A

    S., Stefanov, E. Roohi, & A. Shoja-Sani, (2022), A novel transient-adaptive subcell algorithm with a hybrid application of different collision techniques in direct simulation Monte Carlo (DSMC). Physics of Fluids, 34(9)

    work page 2022

  55. [55]

    Shoja-Sani, A., Roohi, E., & Stefanov, S. (2020). Evaluation of the generalized Bernoulli trial-transient adaptive subcell (GBT-TAS) collision scheme in treating rarefied gas flows. Computers & Fluids, 213, 104740

    work page 2020