Pith. sign in

REVIEW 3 major objections 6 minor 39 references

MARUT keeps high-order compressible CFD, AMR, and finite-rate chemistry fully on the GPU and scales near-linearly across multi-GPU nodes.

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 →

MARUT delivers a fully GPU-resident high-order DG CFD framework with AMR and finite-rate chemistry that validates on canonical compressible and reacting benchmarks and scales across multiple NVIDIA GPUs.

T0 review reviewed 2026-07-14 challenge →

load-bearing objection Solid GPU-native high-order DG stack with real engineering advances; the “exascale-ready / near-linear multi-GPU” branding outruns the four-GPU fixed-mesh evidence. the 3 major comments →

arxiv 2605.26388 v4 pith:KF7TLHNM submitted 2026-05-25 physics.comp-ph math-phmath.MP

MARUT: An Exascale-Ready, GPU-Accelerated High-Order CFD Framework with AMR for High-Speed Flows and Finite-Rate Chemistry

classification physics.comp-ph math-phmath.MP
keywords High-Fidelity CFDCompressible FlowsMulti-GPUMPI ParallelizationAdaptive Mesh RefinementFinite-Rate ChemistrySpectral Discontinuous GalerkinAI-compatible CFD
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.

The reading

MARUT is a multi-GPU CFD framework that runs high-order spectral discontinuous Galerkin discretizations, viscous terms, adaptive mesh refinement, and finite-rate two-temperature chemistry entirely on NVIDIA GPUs. The design keeps the forest mesh, connectivity, fluxes, and source terms device-resident so that time stepping does not shuttle data back to the host. Dynamic AMR concentrates resolution on shocks, shear layers, and reaction fronts while a face-only MPI exchange overlaps halo traffic with interior work. Across canonical subsonic-to-hypersonic and chemically reacting benchmarks the solver matches established reference solutions, and on up to four GPUs it reports high strong- and weak-scaling efficiency. The paper presents this combination as a practical route to high-fidelity, multiscale compressible-flow simulation on heterogeneous machines and as a modular base for later AI-compatible workflows.

Core claim

A fully GPU-resident high-order spectral DG solver with device-side AMR (GPUForest) and on-device finite-rate / two-temperature chemistry can resolve shocks, boundary layers and reacting interfaces while achieving near-linear multi-GPU strong scaling and close agreement with classical compressible-flow benchmarks.

What carries the argument

GPUForest: a device-resident forest-of-quadtrees/octrees that stores tree connectivity, Morton ordering, refinement levels and curvilinear geometry entirely in GPU memory, so that indicator evaluation, 2:1 balancing, L2 solution transfer and connectivity rebuild run as pure device kernels without host round-trips.

Load-bearing premise

That scaling measured on at most four InfiniBand-linked GPUs with CPU-staged MPI is enough to claim the design will stay efficient at true exascale, where load imbalance from dynamic AMR and interconnect latency may dominate.

What would settle it

Run the same AMR-enabled hypersonic reacting case on hundreds of GPUs (or a full multi-node system) and check whether strong-scaling efficiency collapses once communication or rebalancing cost exceeds the interior kernel time.

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

Share X Bluesky LinkedIn Reddit HN

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

3 major / 6 minor

Summary. MARUT is a Julia/Trixi-based multi-GPU CFD framework for compressible flows from subsonic to hypersonic regimes, including finite-rate multi-species chemistry and two-temperature thermal nonequilibrium. It implements a fully GPU-resident spectral DG pipeline (entropy-stable split-form volume integrals, Hennemann–Gassner shock capturing, mortar interfaces, BR1 viscous terms, boundary kernels) on curvilinear P4estMesh geometries, a device-resident forest-of-trees AMR layer (GPUForest), and per-node Strang-split chemistry/Landau–Teller source kernels. Single-GPU validations cover a Mach-3 cylinder with AMR, subsonic and supersonic Taylor–Green vortex (including AMR), ONERA M6 Cp vs experiment, and a nonequilibrium reactive blast. Performance is reported via CPU–GPU wall-time comparisons and strong/weak scaling on up to four L40S GPUs for a fixed-mesh 2D RAE 2822 case.

Significance. If the accuracy and GPU-resident design claims hold, MARUT is a useful, modular high-order platform that advances beyond TrixiCUDA.jl by supporting curvilinear unstructured meshes, full viscous operators, GPUForest AMR, and on-device finite-rate/two-temperature chemistry in one stack. The external-reference validations (van Rees TGV, Chapelier TENO TGV, Schmitt–Charpin ONERA M6, Grossman-type blast) and the documented single-GPU AMR and chemistry kernels are genuine strengths. The Julia implementation and AI-compatibility discussion are secondary but relevant. The main significance risk is branding: “exascale-ready” and “near-linear multi-GPU scaling” for the full AMR+chemistry feature set are not yet demonstrated at the scale the abstract and conclusions assert.

major comments (3)
  1. Abstract, §5.2, and Conclusions claim near-linear strong scaling and an “exascale-ready” design for the full multi-GPU framework with AMR and finite-rate chemistry. The only multi-GPU numbers (strong/weak scaling to 4 L40S GPUs, up to ~91.5% efficiency) use a static-mesh 2D RAE 2822 viscous case with face-only MPI and no dynamic GPUForest adaptation and no chemistry source kernels. §5.1 correctly lists the hard distributed-AMR costs (prolongation/restriction, 2:1 balance, SFC rebalancing, refluxing, subcycling), but those operations are never timed under multi-rank load imbalance. Either provide multi-GPU scaling with active GPUForest (and ideally chemistry) or substantially qualify the abstract/title/conclusion claims so they match the evidence.
  2. §4.4 (nonequilibrium blast) is the sole reactive validation: a 2D closed-domain air-5 blast on a fixed 64×64, P=7 mesh. The paper’s central multiphysics claim includes hypersonic multi-species nonequilibrium with stiff sources. A second reactive case with quantitative comparison (e.g., shock-standoff, species profiles, or a published 1D/2D nonequilibrium reference with tabulated errors) would better support the chemistry accuracy claim; qualitative contour agreement alone is thin for that pillar of the contribution list.
  3. §3.1.4 and Table 3 describe a fully GPU-resident AMR cycle, and single-GPU AMR is shown for the cylinder and TGV Case 3. There is no multi-GPU demonstration that GPUForest rebalancing, mortar rebuild, and L2 projection preserve conservation and load balance across ranks when refinement is highly localized (e.g., bow shock + wake). Without that, the claim that GPUForest “overcomes the primary computational bottleneck in adaptive multiscale simulations” on multi-GPU systems remains an architectural statement rather than a measured result.
minor comments (6)
  1. Title and abstract use “Exascale-Ready”; §5 only reaches four GPUs with CPU-staged InfiniBand MPI. Prefer “multi-GPU scalable” or “designed for exascale architectures” unless larger-scale data are added.
  2. §4.1: times are written as “t = 0.1, 1, 3, and 5 s” while the nondimensional setup has no physical time unit; drop “s” or state the nondimensionalization.
  3. Figure 2 caption and body: “147,801 elements (≈9.46×10^6 DOFs)” is clear; ensure all DOF counts consistently state whether they are grid points or conserved-variable DOFs (cylinder text mixes both).
  4. §3.1.3: Newton FD perturbation ε and Strang-split settings are free parameters; a short sensitivity note (or default table) would help reproducibility of the reactive blast.
  5. References [35,36] are the authors’ own DRL flow-control papers; they are peripheral to the solver contribution and could be moved to a brief “applications” sentence to avoid over-weighting self-citation.
  6. Occasional notation slips (e.g., Ms vs M in TGV; “TrixiCUDA.jl” vs “Trixi ecosystem”) should be unified in a final pass.

Circularity Check

0 steps flagged

No circular derivation: accuracy claims rest on external experimental/DNS benchmarks and standard DG/AMR operators; self-citations are peripheral and non-load-bearing.

full rationale

MARUT is a methods/software paper whose load-bearing claims are (i) a fully GPU-resident spectral DG + BR1 + GPUForest AMR + Park-air-5/two-temperature source pipeline and (ii) measured agreement with external references plus multi-GPU timings. The governing equations (Sec. 2), entropy-stable split-form DGSEM residual (App. B, Eqs. 9–11, 37–42), BR1 viscous operator (Sec. 3.1.2), Strang-split chemistry (Sec. 3.1.3, Eqs. 24–25), and Löhner/Hennemann indicators are standard constructions taken from the literature (Gassner, Ranocha, Bassi–Rebay, Park, etc.) and implemented, not redefined in terms of the target results. Validation cases (cylinder, TGV vs. van Rees DNS and Chapelier TENO, ONERA M6 Cp vs. Schmitt–Charpin experiment, reactive blast vs. Grossman setup) compare solver output to independent external data; no free parameters are fitted to a subset and then re-predicted. The only self-citations ([35,36]) appear in the conclusions as prior application examples of the same code to DRL flow control; they are not invoked to justify uniqueness, stability, or any numerical result of the present paper. Scaling numbers (Sec. 5) are direct wall-time measurements on ≤4 GPUs, not circular predictions. Over-branding as “exascale-ready” is an aspirational claim, not a circular derivation. Hence zero circularity under the stated criteria.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 2 invented entities

The work is an engineering CFD framework: governing physics and high-order DG theory are taken from the literature; free parameters are numerical controls (AMR thresholds, CFL, blending, positivity floors); the main invented software entity is GPUForest. Claims rest on standard continuum models plus implementation choices, not on new physical postulates.

free parameters (5)
  • AMR indicator thresholds and level controller (e.g. cylinder 0.02/0.05; TGV 0.1/0.2; ℓ_max)
    Hand-chosen Löhner/Hennemann thresholds and max levels control where refinement occurs and thus cost and local accuracy; not derived from first principles.
  • CFL / SSP time-step safety factors (e.g. CFL=0.5 ONERA; CFL=0.05 reactive blast)
    Stability multipliers chosen per case; affect reported wall times and whether stiff chemistry remains stable.
  • Hennemann–Gassner blending bounds and shock-indicator field (ρp)
    α_e range and indicator quantity are tunable numerical parameters that switch DG vs FV behavior near shocks.
  • Positivity floors (ρ,p ≥ 10^{-6} or similar)
    Ad hoc floors keep the scheme robust; they are not physical constants.
  • Newton FD perturbation ε and Strang-split chemistry substep settings
    Implicit source solver tolerances/perturbations are implementation knobs that can affect stiff reactive accuracy.
axioms (5)
  • domain assumption Compressible Navier–Stokes (and multi-species two-temperature extension) with Newtonian stress, Fourier heat flux, and ideal-gas / mixture EOS adequately model the target regimes.
    §2 and §2.1 adopt continuum closures standard in hypersonic CFD; continuum breakdown and radiation are out of scope.
  • domain assumption Park air-5 kinetics with Ta=√(T Tv) for dissociation and Landau–Teller/Millikan–White relaxation correctly represent finite-rate thermochemistry for the blast case.
    §2.1 and §3.1.3; mechanism and rates taken from Park literature, not re-derived.
  • domain assumption Entropy-stable split-form DGSEM + Hennemann–Gassner subcell blending + BR1 viscous fluxes yield a stable, accurate semi-discretization on curvilinear elements.
    §3 and Appendices B–C rely on established DG/SBP theory (Gassner, Hennemann, Bassi–Rebay, Ranocha).
  • domain assumption 2:1 balanced forest-of-octrees with Jacobian-aware L2 projection preserves conservation sufficiently for the reported diagnostics.
    §3.1.4 AMR cycle; conservation under nonconforming projection is assumed and checked only via integral diagnostics (e.g. TGV).
  • standard math Standard SBP/GLL calculus and SSP-RK convex-combination stability transfer hold for the implemented kernels.
    Appendices B–C invoke classical summation-by-parts and SSP theory.
invented entities (2)
  • GPUForest no independent evidence
    purpose: GPU-resident forest-of-quadtree/octree holding connectivity, Morton order, levels, and curvilinear metrics so AMR and solvers never leave device memory.
    Primary software invention enabling the paper’s GPU-resident AMR claim; independent evidence is only the authors’ performance/accuracy runs, not an external standard.
  • MARUT solver stack (Julia multi-GPU kernel pipeline) no independent evidence
    purpose: Unified end-to-end GPU path for hyperbolic, parabolic, chemistry, and AMR on unstructured meshes.
    The named framework is the contribution; falsifiable only via released code and third-party runs.

reviewed 2026-07-14 · how reviews work

0 comments
Cite this review

Pith. "Pith review of MARUT: An Exascale-Ready, GPU-Accelerated High-Order CFD Framework with AMR for High-Speed Flows and Finite-Rate Chemistry." pith.science (2026). https://pith.science/paper/KF7TLHNM

@misc{pith2026260526388,
  author       = {Pith},
  title        = {Pith review of: MARUT: An Exascale-Ready, GPU-Accelerated High-Order CFD Framework with AMR for High-Speed Flows and Finite-Rate Chemistry},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KF7TLHNM}},
  note         = {Machine review of arXiv:2605.26388}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We present MARUT, a scalable multi-GPU computational fluid dynamics (CFD) framework designed for high-fidelity simulations of compressible flows spanning subsonic to hypersonic regimes, including chemically reacting nonequilibrium flows with finite-rate chemistry and adaptive mesh refinement (AMR). The framework addresses a central challenge in contemporary scientific computing: the development of numerically accurate and computationally scalable algorithms capable of resolving strongly nonlinear, multiscale flow physics on emerging heterogeneous supercomputing architectures. Built around a distributed-memory MPI-parallel infrastructure and implemented natively on NVIDIA GPUs, MARUT combines high-order spectral discontinuous Galerkin discretisations with strong-stability-preserving Runge--Kutta time integration to achieve low-dissipation and high-resolution representation of shocks, vortical structures and reactive interfaces. Dynamic AMR further enables efficient concentration of computational resources in localized regions of physical complexity, thereby substantially reducing computational cost while preserving solution fidelity. MARUT is designed to maintain strong parallel efficiency through GPU-resident computations and scalable MPI communication strategies, achieving near-linear strong scaling across multiple GPUs. The solver is validated against a broad suite of canonical benchmark problems involving inviscid, viscous, and reactive compressible flows, including subsonic, transonic, supersonic, and hypersonic configurations with multi-species nonequilibrium chemistry. The numerical predictions show close agreement with established reference solutions. Beyond its immediate performance characteristics, the framework reflects the broader transition of computational science towards modular, adaptive and AI-compatible simulation ecosystems.

Figures

Figures reproduced from arXiv: 2605.26388 by Ameya D. Jagtap, Trishit Mondal.

Figure 1
Figure 1. Figure 1: High-level architecture of the MARUT solver. The spatial-discretisation components configure the GPU￾resident RHS kernels, which are advanced in time by the SSP-RK integrator under the control of runtime callbacks. MARUT provides a robust AMR framework based on hierarchical tree-structured meshes, enabling dynamically adaptive high-order DG simulations with excellent conservation and parallel scalability p… view at source ↗
Figure 2
Figure 2. Figure 2: Mach 3 supersonic Euler flow past a cylinder in a channel: density field (left column) and AMR mesh (right [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Subsonic TGV (Case 3, Ms = 0.1, Re = 1600) at t = 10, ≈ 120,000 active elements: AMR mesh (left), density slice (middle), and iso-surface of Q = 0.01 coloured by velocity magnitude (right) [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Subsonic TGV: kinetic energy Ek(t) (left) and dissipation rate ε(t) = −dEk/dt (right) for the three grids of Sec. 4.2 (Case 1: 123 , Case 2: 243 , Case 3: 83 with AMR up to ℓmax = 3), against pseudo-spectral DNS reference data from van Rees et al. [26]. 14 [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Supersonic TGV (Ms = 1.25, Re = 1600) at t = 11, approximately 140,000 active elements: same panel layout as [PITH_FULL_IMAGE:figures/full_fig_p015_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Supersonic TGV: Ek(t) (left), solenoidal dissipation εs(t) (middle), and dilatational dissipation εd(t) (right) for the same three grids as in [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: (Left) Three-dimensional computational domain bounding the ONERA M6 wing, with extents expressed [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Surface pressure coefficient Cp distributions on the ONERA M6 wing at M∞ = 0.84 for the seven spanwise stations (z/b) ∈ {0.20, 0.44, 0.65, 0.80, 0.90, 0.95, 0.99}. Solid lines: MARUT solution at t = 9.0 convective time units on the static 217,088-element P = 3 mesh; black markers: Schmitt and Charpin wind-tunnel data [29] (downward triangles: lower/pressure surface, upward triangles: upper/suction surface)… view at source ↗
Figure 9
Figure 9. Figure 9: Iso-surfaces of the Q-criterion coloured by velocity magnitude for the transonic ONERA M6 wing at M∞ = 0.84, showing the tip-vortex and the trailing-edge shear layer. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Non-equilibrium reactive blast wave at tend = 2 × 10−4 s on a 64 × 64 grid with P = 7 spectral DG and SSPRK54. Top row, left to right: atomic nitrogen ρN, molecular nitrogen ρN2 , nitric oxide ρNO. Bottom row: atomic oxygen ρO, molecular oxygen ρO2 , and translational–rotational temperature T (in K). Color scales are individual to each panel. The cylindrical shock has propagated from the initial interface… view at source ↗
Figure 11
Figure 11. Figure 11: 2D Euler cylinder benchmark: wall-time per SSPRK54 step vs. number of grid points, comparing CPU [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: 3D Navier–Stokes TGV benchmark: wall-time per SSPRK54 step vs. number of grid points, for a 64-thread [PITH_FULL_IMAGE:figures/full_fig_p021_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: MARUT multi-GPU representation. The mesh is partitioned across ranks, and each GPU advances its own partition through an identical pipeline; ghost-cell exchanges overlap with the interior compute on each rank. 5.2 Strong and Weak Scaling The parallel efficiency of MARUT is assessed through both strong and weak scaling experiments on up to four NVIDIA L40S GPUs interconnected via InfiniBand, using CPU-stag… view at source ↗
Figure 14
Figure 14. Figure 14: Strong scaling on up to four L40S GPUs. (Left) Wall time per SSPRK54 step versus GPU count for three [PITH_FULL_IMAGE:figures/full_fig_p023_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Weak scaling on up to four L40S GPUs. (Left) Weak scaling efficiency versus GPU count for three [PITH_FULL_IMAGE:figures/full_fig_p024_15.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

39 extracted references

  1. [1]

    Direct numerical simulation: a tool in turbulence research.Annual review of fluid mechanics, 30(1):539–578, 1998

    Parviz Moin and Krishnan Mahesh. Direct numerical simulation: a tool in turbulence research.Annual review of fluid mechanics, 30(1):539–578, 1998

  2. [2]

    Progress in shock wave/boundary layer interactions.Progress in Aerospace Sciences, 72:80–99, 2015

    Datta V Gaitonde. Progress in shock wave/boundary layer interactions.Progress in Aerospace Sciences, 72:80–99, 2015

  3. [3]

    Cfd vision 2030 study: a path to revolutionary computational aerosciences

    Jeffrey P Slotnick, Abdollah Khodadoust, Juan Alonso, David Darmofal, William Gropp, Elizabeth Lurie, and Dimitri J Mavriplis. Cfd vision 2030 study: a path to revolutionary computational aerosciences. Technical report, 2014

  4. [4]

    On the utility of gpu accelerated high-order methods for unsteady flow simulations: A comparison with industry-standard tools.Journal of Computational Physics, 334:497–521, 2017

    Brian C Vermeire, Freddie D Witherden, and Peter E Vincent. On the utility of gpu accelerated high-order methods for unsteady flow simulations: A comparison with industry-standard tools.Journal of Computational Physics, 334:497–521, 2017

  5. [5]

    Tvb runge-kutta local projection discontinuous galerkin finite element method for conservation laws

    Bernardo Cockburn and Chi-Wang Shu. Tvb runge-kutta local projection discontinuous galerkin finite element method for conservation laws. ii. general framework.Mathematics of computation, 52(186):411–435, 1989. 34

  6. [6]

    Hesthaven and Tim Warburton.Nodal Discontinuous Galerkin Methods: Algorithms, Analysis, and Applications, volume 54 ofTexts in Applied Mathematics

    Jan S. Hesthaven and Tim Warburton.Nodal Discontinuous Galerkin Methods: Algorithms, Analysis, and Applications, volume 54 ofTexts in Applied Mathematics. Springer, New York, 2008

  7. [7]

    Nodal discontinuous galerkin methods on graphics processors.Journal of Computational Physics, 228(21):7863–7882, 2009

    Andreas Kl¨ ockner, Tim Warburton, Jeff Bridge, and Jan S Hesthaven. Nodal discontinuous galerkin methods on graphics processors.Journal of Computational Physics, 228(21):7863–7882, 2009

  8. [8]

    Local adaptive mesh refinement for shock hydrodynamics.Journal of computational Physics, 82(1):64–84, 1989

    Marsha J Berger and Phillip Colella. Local adaptive mesh refinement for shock hydrodynamics.Journal of computational Physics, 82(1):64–84, 1989

  9. [9]

    Trixi.jl: Adaptive high-order numerical simulations of hyperbolic PDEs in Julia

    Michael Schlottke-Lakemper, Gregor J Gassner, Hendrik Ranocha, Andrew R Winters, Jesse Chan, and Andr´ es Rueda-Ram´ ırez. Trixi.jl: Adaptive high-order numerical simulations of hyperbolic PDEs in Julia. https: //github.com/trixi-framework/Trixi.jl, 2025

  10. [10]

    Aiaa, 1989

    John David Anderson.Hypersonic and high temperature gas dynamics. Aiaa, 1989

  11. [11]

    Nonequilibrium hypersonic aerothermodynamics.Nonequilibrium hypersonic aerothermodynamics by Park, page 29860, 1989

    Chul Park. Nonequilibrium hypersonic aerothermodynamics.Nonequilibrium hypersonic aerothermodynamics by Park, page 29860, 1989

  12. [12]

    Computation of weakly ionized hypersonic flows in thermo- chemical nonequilibrium.Journal of Thermophysics and heat transfer, 5(3):266–273, 1991

    Graham V Candler and Robert W MacCormack. Computation of weakly ionized hypersonic flows in thermo- chemical nonequilibrium.Journal of Thermophysics and heat transfer, 5(3):266–273, 1991

  13. [13]

    Wilcox, and Omar Ghattas

    Carsten Burstedde, Lucas C. Wilcox, and Omar Ghattas. p4est: Scalable algorithms for parallel adaptive mesh refinement on forests of octrees.SIAM Journal on Scientific Computing, 33(3):1103–1133, 2011

  14. [14]

    An adaptive finite element scheme for transient problems in CFD.Computer Methods in Applied Mechanics and Engineering, 61(3):323–338, 1987

    Rainald L¨ ohner. An adaptive finite element scheme for transient problems in CFD.Computer Methods in Applied Mechanics and Engineering, 61(3):323–338, 1987

  15. [15]

    Rueda-Ram´ ırez, Florian J

    Sebastian Hennemann, Andr´ es M. Rueda-Ram´ ırez, Florian J. Hindenlang, and Gregor J. Gassner. A provably entropy stable subcell shock capturing approach for high order split form DG for the compressible Euler equations. Journal of Computational Physics, 426:109935, 2021

  16. [16]

    Kopriva.Implementing Spectral Methods for Partial Differential Equations: Algorithms for Scientists and Engineers

    David A. Kopriva.Implementing Spectral Methods for Partial Differential Equations: Algorithms for Scientists and Engineers. Scientific Computation. Springer, Dordrecht, 2009

  17. [17]

    Fisher and Mark H

    Travis C. Fisher and Mark H. Carpenter. High-order entropy stable finite difference schemes for nonlinear conservation laws: Finite domains.Journal of Computational Physics, 252:518–557, 2013

  18. [18]

    Gassner, Andrew R

    Gregor J. Gassner, Andrew R. Winters, and David A. Kopriva. Split form nodal discontinuous Galerkin schemes with summation-by-parts property for the compressible Euler equations.Journal of Computational Physics, 327:39–66, 2016

  19. [19]

    Comparison of some entropy conservative numerical fluxes for the Euler equations.Journal of Scientific Computing, 76(1):216–242, 2018

    Hendrik Ranocha. Comparison of some entropy conservative numerical fluxes for the Euler equations.Journal of Scientific Computing, 76(1):216–242, 2018

  20. [20]

    Michael G. Edwards. The dominant wave-capturing flux: A finite-volume scheme without decomposition for systems of hyperbolic conservation laws.Journal of Computational Physics, 218(1):275–294, 2006

  21. [21]

    Lax, and Bram van Leer

    Amiram Harten, Peter D. Lax, and Bram van Leer. On upstream differencing and Godunov-type schemes for hyperbolic conservation laws.SIAM Review, 25(1):35–62, 1983

  22. [22]

    David A. Kopriva. A conservative staggered-grid Chebyshev multidomain method for compressible flows. II. A semi-structured method.Journal of Computational Physics, 128(2):475–488, 1996

  23. [23]

    Bassi and S

    F. Bassi and S. Rebay. A high-order accurate discontinuous finite element method for the numerical solution of the compressible Navier–Stokes equations.Journal of Computational Physics, 131(2):267–279, 1997

  24. [24]

    The viscosity of gases and molecular force.Philosophical Magazine, Series 5, 36(223):507–531, 1893

    William Sutherland. The viscosity of gases and molecular force.Philosophical Magazine, Series 5, 36(223):507–531, 1893

  25. [25]

    G. M. Morton. A computer oriented geodetic data base and a new technique in file sequencing. Technical report, IBM Ltd., Ottawa, Ontario, Canada, 1966. 35

  26. [26]

    Gmsh: A 3-d finite element mesh generator with built-in pre-and post-processing facilities.International journal for numerical methods in engineering, 79(11):1309–1331, 2009

    Christophe Geuzaine and Jean-Fran¸ cois Remacle. Gmsh: A 3-d finite element mesh generator with built-in pre-and post-processing facilities.International journal for numerical methods in engineering, 79(11):1309–1331, 2009

  27. [27]

    van Rees, Anthony Leonard, D

    Wim M. van Rees, Anthony Leonard, D. I. Pullin, and Petros Koumoutsakos. A comparison of vortex and pseudo-spectral methods for the simulation of periodic vortical flows at high Reynolds numbers.Journal of Computational Physics, 230(8):2794–2805, 2011

  28. [28]

    H3PC: Hypersonic, high-order, high-performance code with adaptive mesh refinement and real chemistry.arXiv preprint arXiv:2511.17551, 2025

    Ahmad Peyvan, Khemraj Shukla, and George Em Karniadakis. H3PC: Hypersonic, high-order, high-performance code with adaptive mesh refinement and real chemistry.arXiv preprint arXiv:2511.17551, 2025

  29. [29]

    Lusher, William Van Noordt, Christoph Wenzel, Tobias Gibis, Pascal Mossier, Andrea Beck, Guido Lodato, Christoph Brehm, Matteo Ruggeri, Carlo Scalo, and Neil Sandham

    Jean-Baptiste Chapelier, David J. Lusher, William Van Noordt, Christoph Wenzel, Tobias Gibis, Pascal Mossier, Andrea Beck, Guido Lodato, Christoph Brehm, Matteo Ruggeri, Carlo Scalo, and Neil Sandham. Comparison of high-order numerical methodologies for the simulation of the supersonic Taylor–Green vortex flow.Physics of Fluids, 36(5):055146, 2024

  30. [30]

    Pressure distributions on the onera m6-wing at transonic mach numbers, experimental data base for computer program assessment.AGARD AR-138, 1979

    V Schmitt. Pressure distributions on the onera m6-wing at transonic mach numbers, experimental data base for computer program assessment.AGARD AR-138, 1979

  31. [31]

    Flux-split algorithms for flows with non-equilibrium chemistry and vibrational relaxation.Journal of Computational Physics, 88(1):131–168, 1990

    B Grossman and P Cinnella. Flux-split algorithms for flows with non-equilibrium chemistry and vibrational relaxation.Journal of Computational Physics, 88(1):131–168, 1990

  32. [32]

    Evaluating modern gpu interconnect: Pcie, nvlink, nv-sli, nvswitch and gpudirect.IEEE Transactions on Parallel and Distributed Systems, 31(1):94–110, 2019

    Ang Li, Shuaiwen Leon Song, Jieyang Chen, Jiajia Li, Xu Liu, Nathan R Tallent, and Kevin J Barker. Evaluating modern gpu interconnect: Pcie, nvlink, nv-sli, nvswitch and gpudirect.IEEE Transactions on Parallel and Distributed Systems, 31(1):94–110, 2019

  33. [33]

    Adaptive mesh refinement for hyperbolic partial differential equations

    Marsha J Berger and Joseph Oliger. Adaptive mesh refinement for hyperbolic partial differential equations. Journal of computational Physics, 53(3):484–512, 1984

  34. [34]

    Amrex: Block-structured adaptive mesh refinement for multiphysics applications.The International Journal of High Performance Computing Applications, 35(6):508–526, 2021

    Weiqun Zhang, Andrew Myers, Kevin Gott, Ann Almgren, and John Bell. Amrex: Block-structured adaptive mesh refinement for multiphysics applications.The International Journal of High Performance Computing Applications, 35(6):508–526, 2021

  35. [35]

    Shocks under control: Taming transonic compressible flow over an rae2822 airfoil with deep reinforcement learning.arXiv preprint arXiv:2511.07564, 2025

    Trishit Mondal, Ricardo Vinuesa, and Ameya D Jagtap. Shocks under control: Taming transonic compressible flow over an rae2822 airfoil with deep reinforcement learning.arXiv preprint arXiv:2511.07564, 2025

  36. [36]

    Trishit Mondal and Ameya D. Jagtap. Hypersonic flow control: Generalized deep reinforcement learning for hypersonic intake unstart control under uncertainty.Physics of Fluids, 38(4):046112, 2026

  37. [37]

    A review of reaction rates and thermodynamic and transport properties for an 11-species air model for chemical and thermal nonequilibrium calculations to 30000 k

    Roop N Gupta, Jerrold M Yos, Richard A Thompson, and Kam-Pui Lee. A review of reaction rates and thermodynamic and transport properties for an 11-species air model for chemical and thermal nonequilibrium calculations to 30000 k. 1990

  38. [38]

    Spiteri and Steven J

    Raymond J. Spiteri and Steven J. Ruuth. A new class of optimal high-order strong-stability-preserving time discretization methods.SIAM Journal on Numerical Analysis, 40(2):469–491, 2002

  39. [39]

    Efficient implementation of essentially non-oscillatory shock-capturing schemes

    Chi-Wang Shu and Stanley Osher. Efficient implementation of essentially non-oscillatory shock-capturing schemes. Journal of Computational Physics, 77(2):439–471, 1988. 36

This paper was first reviewed by grok-4.5 on July 14, 2026.