Pith. sign in

REVIEW 2 major objections 5 minor 100 references

AGILE is a GPU-native adaptive mesh refinement framework claiming up to a fivefold cut in cost-to-solution for astrophysical flows, with small 16^3 blocks and scaling to 2048 GPUs.

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 · deepseek-v4-flash

2026-08-01 12:52 UTC pith:Q2JLQLVP

load-bearing objection AGILE is a well-engineered GPU-AMR framework with credible performance claims and broad physics coverage, but the paper leaves a key correctness question unanswered: how coarse-fine interface fluxes are reconciled (if at all). the 2 major comments →

arxiv 2607.19277 v1 pith:Q2JLQLVP submitted 2026-07-21 astro-ph.IM astro-ph.HEastro-ph.SR

Astrophysics on GPUs: introducing AGILE 1.0

Oliver Porth , Adrian Kelly , Olaf Willocx , Hao Wu , Jesse Vos , Yuhao Zhou , H\'ector R. Olivares S\'anchez , Leon Oostrum
show 5 more authors
Johan Hidding Victor Azizi Chun Xia Rony Keppens Jannis Teunissen
This is my paper
classification astro-ph.IM astro-ph.HEastro-ph.SR
keywords adaptive mesh refinementGPU computingOpenACChydrodynamicsmagnetohydrodynamicsrelativistic hydrodynamicsfinite volume methodsastrophysical simulations
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 paper introduces AGILE, a GPU-accelerated adaptive mesh refinement framework for astrophysical conservation laws, and claims it is cost-efficient and scalable enough for production science. On a single current GPU it reaches about two billion cell updates per second, and cost-to-solution drops by up to a factor of five compared with the CPU implementation from which AGILE inherits its mesh handling. The central design claim is that a macro-kernel that loops over AMR blocks and computes each cell's fluxes redundantly—rather than reusing fluxes across threads—makes small 16^3-cell blocks perform as well as larger ones, which is what deeply nested AMR needs. This matters because astrophysical flows span huge dynamic range, and GPU clusters increasingly dominate supercomputing; a framework that combines AMR depth with GPU speed lets established science workflows move to accelerators without rewriting setups.

Core claim

AGILE's central claim is that the main obstacle to GPU AMR—efficient execution with moderate block sizes—can be overcome by restructuring the update loop. Instead of scheduling whole-block operations that create block-sized temporary arrays, AGILE launches one macro-kernel with a gang loop over the space-filling curve of blocks and collapsed vector loops over cells, so all work for a Runge–Kutta substep (primitive-variable conversion, reconstruction, Riemann fluxes, source terms, conservative update) stays local to one vector unit. Fluxes at shared interfaces are computed twice, once by each adjacent cell, but the savings in shared-memory traffic and kernel-launch overhead make the redundanc

What carries the argument

The macro-kernel: a single GPU kernel that treats the AMR block hierarchy as a gang loop and each block's cells as collapsed vector loops, performing every step of one timestep per cell. Its load-bearing details are redundant interface flux computation (each interface flux is computed by both adjacent cells rather than shared through a block-sized array), cell-local register storage, and device-side ghost-cell packing with exactly-sized buffers and one MPI message per neighbor relation. This machinery keeps kernel-launch overhead low and memory traffic minimal, which is what allows small block sizes without performance loss.

Load-bearing premise

The GPU kernels are trusted to reproduce the CPU solver's physics because they reuse its algorithms and pass standard test problems, but no manufactured-solution or order-of-accuracy study checks the device-side macro-kernel and ghost-cell packing for subtle offloading errors.

What would settle it

Run a 3D manufactured-solution or grid-convergence test with AGILE across multiple AMR levels, including a fine-coarse interface with device-side ghost exchange, and monitor conservation of total mass, momentum, and energy; if the measured convergence order falls below the formal reconstruction order, or conserved quantities drift at refinement boundaries in a way the CPU solver does not show, the correctness-inheritance premise fails.

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

If this is right

  • Small 16^3-cell blocks become a viable default for AMR, so refinement can be deep and local without a performance penalty.
  • Cost-to-solution drops by up to about five times on current GPU clusters, letting the same production simulations run for a fraction of the CPU-node budget.
  • Existing setups and data formats from the parent CPU code carry over almost unchanged, lowering the barrier for moving established science workflows to GPUs.
  • Because the kernel is memory-bound, future GPUs optimized for memory-hungry AI workloads should deliver further speedups without code changes.
  • Four physics modules—hydrodynamics, frozen-field hydrodynamics, magnetohydrodynamics, and special-relativistic hydrodynamics—are demonstrated on production-scale runs, with new modules reported to port in about a day.

Where Pith is reading between the lines

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

  • A fair next test would compare full AMR runs against uniform-grid runs at equal effective resolution, since the headline CUPS numbers are uniform-grid benchmarks and do not isolate refinement, load-balancing, or prolongation/restriction overhead.
  • The redundant-flux design trades arithmetic for memory traffic; on GPUs with much higher compute-to-bandwidth ratios the tradeoff could reverse, so a variant that caches fluxes in shared memory would reveal how much of the speed comes from register locality versus avoiding block-sized arrays.
  • A manufactured-solution convergence suite across all four physics modules, especially special-relativistic hydrodynamics with its iterative primitive-variable inversion, would settle the correctness-inheritance question that the standard test problems do not fully address.
  • The reported one-day porting time for a physics module suggests the modular interface itself is the durable asset; if it stays stable, AGILE could become a common GPU backend for the wider family of codes that share its parent CPU lineage.

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. AGILE 1.0 is presented as a GPU-accelerated AMR framework for (near-)conservation laws, developed as a modern-Fortran/OpenACC compatible successor to MPI-AMRVAC. The paper describes the design (macro-kernel cell-by-cell updates, device-side ghost-cell packing, dynamic buffers, OpenACC portability), the physics modules (HD, FFHD, MHD, SRHD), single- and multi-GPU benchmarks up to 2048 GPUs, and six application demonstrations with quantitative diagnostics. The central claims are that AGILE achieves excellent performance on GPUs with moderate block sizes (16^3), scales well, and reduces cost to solution by up to a factor of ~5 relative to the CPU MPI-AMRVAC implementation.

Significance. If the claims are upheld, AGILE is a valuable community resource: it is open-source (GPLv3), has continuous integration on multiple compilers, and provides a relatively low-barrier transition path from MPI-AMRVAC to GPU clusters. The paper includes broad hardware coverage (NVIDIA, AMD, CPU), a roofline analysis, scaling to 2048 GPUs, and cross-code comparisons (MPI-AMRVAC, and discussions of foap4/AthenaK/gPLUTO). The science applications go beyond mere demonstration and include quantitative validation (fractal dimension d=2.541, ram-pressure drag scaling, jet morphology matching MPI-AMRVAC). The main gap is methodological: conservation at AMR coarse-fine boundaries is not documented, and no formal convergence/order-of-accuracy verification of the GPU kernels is provided.

major comments (2)
  1. [Section 2.1.3 / Fig. 2] The AMR dataflow described in Section 2.1.3 and the flow charts in Figures 1–2 do not include a coarse-fine flux correction (refluxing) step. The macro-kernel (Listing 3) updates each cell from fluxes computed from local data, and the ghost-cell exchange packs/unpacks same-resolution, finer, and coarser neighbor data, but no mechanism is described for correcting the coarse-level fluxes at coarse-fine interfaces after fine-level cells are updated. For block-structured AMR without such correction, the scheme is non-conservative across level boundaries. This directly affects the quantitative diagnostics in the multi-level applications (Sections 4.1, 4.3, 4.5, 4.6), such as cloud mass fraction (Figure 16), fractal dimension (Figure 13), and jet propagation. The authors must either (a) document the refluxing routine and where it appears in the code, or (b) add a conservation-error diagnostic
  2. [Sections 2.1.3 and 4] The numerical correctness of the GPU macro-kernel is not verified by a convergence or manufactured-solution study. The paper relies on standard benchmarks, visual agreement, and integrated physical scalings (fractal dimension, ram-pressure drag, jet morphology). While these are necessary and valuable, they are not sufficient to rule out subtle implementation errors in the redundant-flux device kernel or in device-side ghost-cell packing. A formal order-of-accuracy test (e.g., a smooth 3D advection or acoustic-wave problem on multiple AMR levels) would materially strengthen the central 'capable' claim. If such tests exist in the repository or test suite, they should be cited and reported; otherwise they should be added.
minor comments (5)
  1. [Section 3.1] Typo: 'G200' should be 'B200' in the sentence 'the G200 features 4.8TB/s'. The table and surrounding text refer to B200.
  2. [Section 4.6] The MPI-AMRVAC comparison run is said to use 'EPYC Rome' cores on four Snellius nodes, while the CPU benchmark nodes in Section 3.1 are EPYC Genoa/Milan. Please clarify the exact CPU model and node count for the jet comparison, as the cost-to-solution factor depends on this.
  3. [Section 3.1 / Table B1] All CUPS values appear to be single measurements. Given the cost-to-solution claims, a measure of run-to-run variability (or at least a statement that runs are deterministic) would improve confidence. Please indicate whether the reported numbers are representative of a single run or an average over repeats.
  4. [References] The citation format 'Teunissen & et al. (2026)' should be corrected; the reference list entry is 'Teunissen, J. & et al., 2026'.
  5. [Section 5] The conclusion states 'demonstrated reduced cost to solution of up to a factor of 5', while Section 3.1 quotes 'over 5 times reduced runtime and cost' and Section 4.6 quotes a 5.3x wall-clock reduction but a 3.6x cost reduction. Please harmonize these numbers and clarify the exact basis (node cost, wall-clock, or both) for the central cost claim.

Circularity Check

1 steps flagged

Mostly self-contained performance and validation; one cloud-crushing diagnostic fits the drag coefficient, then calls the resulting curve an 'analytic prediction'.

specific steps
  1. fitted input called prediction [Section 4.3, cloud crushing diagnostics and Figure 16]
    "The gray lines show the analytic ram-pressure drag prediction vcl(t) = at with constant acceleration a = 3CD v^2_w/(8χr_c), derived under the assumption vcl << v_w, with drag coefficient CD fitted to the early linear regime of each cooling run. The fitted drag coefficients are consistent across all three density contrasts, CD ∈ {3.06, 3.06, 3.19} respectively, confirming that the early acceleration follows the expected 1/χ scaling of the ram-pressure drag law across the full parameter range."

    The slope of the 'predicted' curve is obtained by fitting CD to the early linear regime of the very same data that the curve is then displayed against, so the early-time agreement is enforced by construction rather than predicted. The only residual content is the consistency of the three fitted CD values across density contrasts, which provides weak support for the 1/χ scaling but does not make the curve a parameter-free analytic prediction. This is a secondary validation diagnostic, not the paper's central capability claim, and the main performance benchmarks remain externally measured.

full rationale

No significant circularity in the central claims. AGILE's capability and cost-efficiency claims rest on measured single-GPU and multi-GPU performance (CUPS, scaling efficiencies, SBU cost comparisons) against external hardware, independent codes such as AthenaK, gPLUTO and foap4, and an openly reproduced CPU implementation in MPI-AMRVAC; these are not fitted parameters or self-referential definitions. The heavy self-citation of MPI-AMRVAC and foap4 is largely attribution of inherited design and algorithms rather than load-bearing evidence for the measured numbers. The science applications are compared to external theoretical scalings (fractal dimension 5/2, cloud-crushing time, analytic drag law) and to previously published setups. The one concrete circular element is the cloud-crushing drag comparison in Section 4.3, where a drag coefficient is fitted to the early data and the resulting curve is labeled an 'analytic ram-pressure drag prediction'; the early-time match is therefore by construction, though the near-constancy of CD across χ retains some independent content. The absence of a documented coarse-fine flux-correction/refluxing step is a potential correctness risk for AMR conservation, but it is not a circularity and is not scored here.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

No new physical particles, forces, or conserved quantities are introduced. The only 'invented' entity is the AGILE software itself, which is not a physical postulate. The free parameters listed above are confined to example applications and diagnostics, not to the core performance or scaling claims.

free parameters (4)
  • log-skew-normal cooling parameters μ and A = μ≈1.19, A≈8.19e-16 g cm^5 s^-3
    Fit to approximate the Colgan-DM cooling curve in the TRML application (Section 4.2). Used in a science example, not in the central performance claim.
  • Cloud drag coefficient C_D = 3.06, 3.06, 3.19
    Fitted to the early linear regime of the cloud velocity in cloud-crushing runs (Section 4.3). A diagnostic parameter for comparing with the analytic ram-pressure drag law.
  • AMR refinement weights = 0.8 tracer / 0.2 density
    Hand-chosen in the cloud-crushing setup (Section 4.3) to concentrate refinement on the cloud; not load-bearing for the framework's performance.
  • GLM divergence-cleaning parameter α_GLM = 0.5
    Fixed to 0.5, within the standard [0,1] range (Section 2.3.3). This is a numerical parameter of the GLM scheme, not fitted to any target.
axioms (4)
  • standard math Finite-volume discretization with Riemann solvers and Runge-Kutta time integration is a correct discretization of the conservation laws in Eq. (1).
    Standard CFD methodology; the paper does not introduce or prove any new numerical scheme.
  • domain assumption OpenACC compilers generate correct device code for the macro-kernel and the nested derived-type ghost-cell structures.
    The paper documents several compiler bugs and workarounds (Section 5); correctness depends on the compilers behaving as assumed on both NVIDIA and AMD targets.
  • domain assumption The physical equations used in the HD, FFHD, MHD, and SRHD modules are the correct models for the presented applications.
    These are established models cited from literature (Dedner et al. 2002; Bergmans et al. 2005; Zhou et al. 2024/2025, etc.), not derived in this paper.
  • domain assumption The uniform-grid benchmark measurements are representative of production AMR workloads.
    Headline CUPS figures come from a uniform double-Kelvin-Helmholtz setup; AMR overhead is only shown anecdotally in the application sections.

pith-pipeline@v1.3.0-alltime-deepseek · 32502 in / 10260 out tokens · 98421 ms · 2026-08-01T12:52:28.297717+00:00 · methodology

0 comments
read the original abstract

We present AGILE, a GPU-enabled adaptive mesh refinement (AMR) framework for the solution of (near-) conservation laws which occur in astro- and solar-physical applications. AGILE is written in modern fortran 2003, inherits a part of its modules and mesh handling from MPI-AMRVAC, and achieves excellent GPU performance via OpenACC offloading. We here discuss the design decisions which enable AGILE to perform cost-efficient and scalable deeply nested AMR simulations with moderate block sizes of e.g. $16^3$ cells. AGILE currently implements several physics modules, ie. hydrodynamics, frozen-field hydrodynamics, magnetohydrodynamics and special-relativistic hydrodynamics and can easily be extended further through its modular design. Besides strong scaling tests to up to 2048 GPUs and standard benchmarks which show consistent performance across a large range of devices and problem sizes, we demonstrate AGILE's capabilities by means of state-of-the art science applications with all currently available physics modules.

Figures

Figures reproduced from arXiv: 2607.19277 by Adrian Kelly, Chun Xia, Hao Wu, H\'ector R. Olivares S\'anchez, Jannis Teunissen, Jesse Vos, Johan Hidding, Leon Oostrum, Olaf Willocx, Oliver Porth, Rony Keppens, Victor Azizi, Yuhao Zhou.

Figure 1
Figure 1. Figure 1: Flow chart of the initialization using host routines. The left col￾umn indicates subroutines called from the top-level function, subsequent subroutine calls are shown correspondingly to the right. Data transfers from host to device are colored green, the opposite direction is colored red. Host routines are marked orange and routines that execute GPU kernels are col￾ored blue. First the base level grid is i… view at source ↗
Figure 3
Figure 3. Figure 3: Main integration kernel looping over blocks via OpenACCs “gang loop” and scheduling operations on a cell-by-cell basis on each vector pro￾cessor. Next to the device data handling and the main update kernel, an￾other performance critical element is the ghostcell exchange between MPI tasks. Efficient buffer storage management is important since GPU memory is limited, in fact our first naive approach (allocat… view at source ↗
Figure 4
Figure 4. Figure 4: Device routine to obtain the hydrodynamic fluxes from variables in primitive form in u (hence e.g. u(iw_e) contains pressure etc.). Note that a developer does not need to add parallelization instructions since the routine is executed on the vector level. hydrodynamics. AGILE is fully modular and exposes generic inter￾faces of fluxes and source terms to the update algorithm. Specific implementations are col… view at source ↗
Figure 5
Figure 5. Figure 5: Single device benchmark with varying problem- and block-sizes. Good performance is already obtained at small block size of 163 , a decreasing number of blocks per SM leads to reduced performance for large block sizes. This is seen clearly for the small test case 1283 resulting in only 8 blocks per device when a block size of 643 is used [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: shows the multi-GPU strong scaling behaviour of AGILE, while in fig. 7 we show the scaling efficiency for both strong and weak scaling, defined as the performance relative to that of a single node. For reference, we also show the scaling behaviour on the LUMI CPU partition, where we are limited to a four times smaller problem size of 512×512×512 on a single node. As with the single-GPU case, the H100 outpe… view at source ↗
Figure 7
Figure 7. Figure 7: Weak and strong scaling efficiency. AGILE scales well on all systems in both strong and weak scaling. 1011 CUPS, using 2048 MI250X GCDs, however scaling efficiency has dropped below 50% at that point. 4 EXAMPLE APPLICATIONS In this section we showcase some relevant applications as demonstra￾tion for the science-readiness of AGILE. We also provide quantitative measures that are compared to theoretical expec… view at source ↗
Figure 8
Figure 8. Figure 8: Evolving AMR level population over simulation time for the double Mach reflection simulation. As more of the initially uniform domain becomes processed by the shock, the level 1 occupancy shrinks in favor of refined grids. At peak, over a million grid blocks are activated which decreases again as the shocked cloud is pushed out of the computational domain. tree, which allow full restarts, we obtain 3 data … view at source ↗
Figure 9
Figure 9. Figure 9: Double Mach reflection structure impacting on a dense cloud. This case shows dynamical adaptive mesh structures using a total of six refinement levels. The simulation used up to 2048 GPUs (256 nodes) on the LUMI cluster. See https://surfdrive.surf.nl/s/eYjk7fzaZnwXX8X for an animated version of the figure. We show the density at 𝑡 = 0.75 in a vertical slice at 𝑧 = 0.25 (top) and density (middle) and pressu… view at source ↗
Figure 10
Figure 10. Figure 10 [PITH_FULL_IMAGE:figures/full_fig_p009_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: A slice of the turbulent shearing system showing the TRML. The absorption of material from the hot phase is characterized by a steady diagonal velocity field. The 𝜌 slice shows SMR blocks boundaries in red, the TRML has room to migrate upwards. See https://surfdrive.surf.nl/s/6S9wrHE9E8Nj3rn for an animated version of the figure [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗
Figure 13
Figure 13. Figure 13: Measured fractal dimension 𝑑 = 2.541, for the same time 𝑡/𝑡sh = 30 shown in Figures 11 and 12 by hydrodynamic instabilities. This decreases monotonically for all runs, indicating that radiative cooling slows but does not completely prevent cloud disruption in the chosen 𝑡cool,c/𝑡cc < 1 parameter range. The comparison between the 𝜒 = 240 runs with and with￾out radiative cooling directly demonstrates the ef… view at source ↗
Figure 14
Figure 14. Figure 14: Cloud morphology and AMR grid structure for the 𝜒 = 240 cloud-crushing run. Projected column density log10 Σ along the 𝑧-axis (left panels) and face-on 𝑥-axis projection (right insets) at 𝑡 ≃ 1 𝑡cc = 6.3 Myr (top row) and 𝑡 ≃ 2.5 𝑡cc = 14.7 Myr (bottom row) [PITH_FULL_IMAGE:figures/full_fig_p012_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Density slices through the 𝜒 = 240 cloud-crushing simulation at 𝑡 ≃ 1.86 𝑡cc = 10.7 Myr, computed with four AMR levels and the TVDLF scheme, showing the shock-compressed cloud core (red) and the turbulent, KH/RT-disrupted tail mixing into the ambient wind (teal). See https://surfdrive.surf. nl/s/5KGAx4Ct8E7anwE for an animated version of the figure. HLL solver, 32 MI250X GCDs) achieved between 1.02–1.22 ×… view at source ↗
Figure 17
Figure 17. Figure 17: Volume rendering of the condensed structures emerging in the ffhd simulation. Filamentary overdensities continuously form and flow along field lines reminiscent of coronal rain. A multi-peaked transfer function was selected to show near isocontours of density and we also show a sample of the frozen-in-time background magnetic field lines. models, the lower atmosphere is represented by a cool chromosphere,… view at source ↗
Figure 16
Figure 16. Figure 16: Quantitative diagnostics for cloud-crushing runs at 𝜒 ∈ {140, 240, 340} on a uniform grid of 1280 × 256 × 256 cells using the HLL solver, with an additional no-radiative-cooling run at 𝜒 = 240 (dashed red). Top: cloud velocity 𝑣25 (𝑡), defined as the velocity at the 25th percentile of the integrated mass distribution along 𝑥; gray lines show the analytic ram￾pressure drag prediction. Bottom: cold cloud ma… view at source ↗
Figure 18
Figure 18. Figure 18: Node-level performance comparison for the FFHD simulation case between running in AGILE and MPI-AMRVAC. Results are shown for the baseline FFHD configuration and for configurations including radiative cool￾ing (RC), hyperbolic thermal conduction (HTC), and both source terms (BOTH). The blue and red curves show the throughput of AGILE and MPI-AMRVAC, respectively, measured in Cell updates / second, while t… view at source ↗
Figure 19
Figure 19. Figure 19: The cooled Orszag-Tang state at times 𝑡 = 0.4 and 𝑡 = 1.8, showing magnetic field lines and strength, a 𝑇 slice, and volumetrically regions of relatively high density indicating the formation of condensations due to cooling instability. 80 60 40 20 0 20 40 60 80 x [kpc] 40 30 20 10 0 10 20 30 40 y [kpc] AMRVAC v3.3 AGILE v1.0 t = 38.58 tcross 6 5 4 3 2 1 0 1 6 5 4 3 2 1 0 1 log10 / b [PITH_FULL_IMAGE:fig… view at source ↗
Figure 20
Figure 20. Figure 20: Propagation of a Lorentz factor Γ = 7 relativistic jet simulation after 38 crossing times. Slice along jet axis comparing MPI-AMRVAC (left) and AGILE (right). See https://surfdrive.surf.nl/s/D7TewDy7ciNX5sk for an animated version of the figure. RASTI 000, 1–20 (2026) [PITH_FULL_IMAGE:figures/full_fig_p016_20.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

100 extracted references · 21 canonical work pages · 2 internal anchors

  1. [1]

    Visualization Handbook , year =

    Ahrens, James and Geveci, Berk and Law, Charles , chapter =. Visualization Handbook , year =

  2. [3]

    T. E. Computing in Science Engineering , title=. 2007 , volume=. doi:10.1109/MCSE.2007.58 , ISSN=

    doi:10.1109/mcse.2007.58 2007

  3. [4]

    Jones, Eric and Oliphant, Travis and Peterson, Pearu , year =

  4. [5]

    Hunter, J. D. , Title =. Computing in Science & Engineering , Volume =

  5. [6]

    Seppänen and Jody Klymak and Damon McDougall and Andrew Straw and Paul Hobson and cgohlke and Tony S Yu and Eric Ma and Adrien F

    Michael Droettboom and Thomas A Caswell and John Hunter and Eric Firing and Jens Hedegaard Nielsen and Antony Lee and Elliott Sales de Andrade and Nelle Varoquaux and David Stansby and Benjamin Root and Phil Elson and Darren Dale and Jae-Joon Lee and Ryan May and Jouni K. Seppänen and Jody Klymak and Damon McDougall and Andrew Straw and Paul Hobson and cg...

  6. [7]

    ,keywords =

    yt: A Multi-code Analysis Toolkit for Astrophysical Simulation Data. ,keywords =. doi:10.1088/0067-0049/192/1/9 ,adsurl =

    doi:10.1088/0067-0049/192/1/9

  7. [8]

    The Astrophysical Journal , volume =

    An, Yifu and Chen, Yuxi and Zhou, Hongyang and Gaenko, Alexander and T. The Astrophysical Journal , volume =. doi:10.3847/1538-4357/adaf15 , urldate =

    work page doi:10.3847/1538-4357/adaf15

  8. [9]

    The Astrophysical Journal , volume =

    Magnetic Reconnection and Hot Spot Formation in Black Hole Accretion Disks , author =. The Astrophysical Journal , volume =. doi:10.3847/1538-4357/ababab , archiveprefix =. 2003.04330 , primaryclass =

    Pith/arXiv arXiv 2003

  9. [10]

    1997 , month = apr, journal =

    Morphology and Dynamics of Relativistic Jets , author =. 1997 , month = apr, journal =. doi:10.1086/303842 , url =

    work page doi:10.1086/303842 1997

  10. [11]

    2011 , month = dec, journal =

    Intracluster Medium Reheating by Relativistic Jets , author =. 2011 , month = dec, journal =. doi:10.1088/0004-637X/743/1/42 , url =. arXiv , arxivurl =:1107.5484 , primaryclass =

    Pith/arXiv arXiv 2011

  11. [12]

    Experimental Results , keywords =

    Shock identification and classification in 2D magnetohydrodynamiccompressible turbulence Orszag-Tang vortex. Experimental Results , keywords =. doi:10.1017/exp.2021.28 , adsurl =

    doi:10.1017/exp.2021.28 2021

  12. [13]

    Journal of Fluid Mechanics , keywords =

    Small-scale structure of two-dimensional magnetohydrodynamic turbulence. Journal of Fluid Mechanics , keywords =. doi:10.1017/S002211207900210X , adsurl =

    doi:10.1017/s002211207900210x

  13. [14]

    , keywords =

    Effect of optically thin cooling curves on condensation formation: Case study using thermal instability. , keywords =. doi:10.1051/0004-6361/202140665 , archivePrefix =. 2107.07569 , primaryClass =

    Pith/arXiv arXiv

  14. [15]

    Computer Physics Communications , keywords =

    A multidimensional grid-adaptive relativistic magnetofluid code. Computer Physics Communications , keywords =. doi:10.1016/j.cpc.2008.05.005 , archivePrefix =. 0807.0713 , primaryClass =

    Pith/arXiv arXiv 2008

  15. [16]

    , keywords =

    Two-fluid implementation in MPI-AMRVAC with applications to the solar chromosphere. , keywords =. doi:10.1051/0004-6361/202243630 , archivePrefix =. 2205.05049 , primaryClass =

    Pith/arXiv arXiv

  16. [17]

    Evolution of the Orszag-Tang vortex system in a compressible medium. II. Supersonic flow. Physics of Fluids B , keywords =. doi:10.1063/1.859953 , adsurl =

    work page doi:10.1063/1.859953

  17. [18]

    , keywords =

    Extension of the MURaM Radiative MHD Code for Coronal Simulations. , keywords =. doi:10.3847/1538-4357/834/1/10 , archivePrefix =. 1609.09818 , primaryClass =

    Pith/arXiv arXiv

  18. [19]

    , keywords =

    Modeling the thermal conduction in the solar atmosphere with the code MANCHA3D. , keywords =. doi:10.1051/0004-6361/202243439 , archivePrefix =. 2205.08846 , primaryClass =

    Pith/arXiv arXiv

  19. [20]

    Geophysical and Astrophysical Fluid Dynamics , keywords =

    Non-Fourier description of heat flux evolution in 3D MHD simulations of the solar corona. Geophysical and Astrophysical Fluid Dynamics , keywords =. doi:10.1080/03091929.2019.1670173 , archivePrefix =. 1811.01572 , primaryClass =

    arXiv 2019

  20. [21]

    and Mignone, A

    Rossi, P. and Mignone, A. and Bodo, G. and Massaglia, S. and Ferrari, A. , year =. Formation of Dynamical Structures in Relativistic Jets: The. Astronomy and Astrophysics , volume =. doi:10.1051/0004-6361:200809687 , url =. arXiv , date-added =:0806.1648 , pages =

    Pith/arXiv arXiv

  21. [22]

    2019 , month = jun, journal =

    Using Synchrotron Emission Modelling of Relativistic Hydrodynamic Jet Simulations to Study the. 2019 , month = jun, journal =. doi:10.1093/mnras/stz734 , url =

    work page doi:10.1093/mnras/stz734 2019

  22. [23]

    and Mart

    Perucho, M. and Mart. Long-Term. 2018 , month = oct, journal =. arXiv , arxivurl =:1810.10968 , url =

    Pith/arXiv arXiv 2018

  23. [24]

    1974 , month = dec, journal =

    A 'twin-Exhaust' Model for Double Radio Sources , author =. 1974 , month = dec, journal =

    1974

  24. [25]

    2015 , month = sep, journal =

    Causality and Stability of Cosmic Jets , author =. 2015 , month = sep, journal =. doi:10.1093/mnras/stv1295 , url =. arXiv , date-added =:1408.3318 , primaryclass =

    Pith/arXiv arXiv 2015

  25. [26]

    , year =

    Bicknell, Geoffrey V. , year =. Relativistic. The Astrophysical Journal Supplement Series , volume =. doi:10.1086/192232 , url =

    doi:10.1086/192232

  26. [27]

    1984 , month = apr, journal =

    Theory of Extragalactic Radio Sources , author =. 1984 , month = apr, journal =

    1984

  27. [28]

    2012 , month = sep, volume =

    Observational Evidence of Active Galactic Nuclei Feedback , author =. 2012 , month = sep, volume =. doi:10.1146/annurev-astro-081811-125521 , url =. arXiv , date-added =:1204.4114 , pages =

    Pith/arXiv arXiv 2012

  28. [29]

    Seo, Jeongbhin and Kang, Hyesung and Ryu, Dongsu , year =. A. The Astrophysical Journal , volume =. doi:10.3847/1538-4357/ac19b4 , url =

    work page doi:10.3847/1538-4357/ac19b4

  29. [30]

    2006 , month = sep, journal =

    Equation of State in Numerical Relativistic Hydrodynamics , author =. 2006 , month = sep, journal =. doi:10.1086/505937 , url =. arXiv:astro-ph/0605550 , pages =

    Pith/arXiv arXiv 2006

  30. [31]

    The Astrophysical Journal Supplement Series , volume =

    The Piecewise Parabolic Method for Multidimensional Relativistic Fluid Dynamics , author =. The Astrophysical Journal Supplement Series , volume =. doi:10.1086/430905 , adsnote =. arXiv:astro-ph/0505200 , pages =

    Pith/arXiv arXiv

  31. [32]

    The Astrophysical Journal , volume =

    The Hydromagnetic Free Expansion of a Relativistic Gas , author =. The Astrophysical Journal , volume =. doi:10.1086/150883 , adsnote =

    doi:10.1086/150883

  32. [33]

    and Sauty, C

    Meliani, Z. and Sauty, C. and Tsinganos, K. and Vlahakis, N. , year = 2004, month = oct, volume =. Relativistic. doi:10.1051/0004-6361:20035653 , adsnote =. arXiv:astro-ph/0407100 , pages =

    Pith/arXiv arXiv 2004

  33. [34]

    , author =

    The Relativistic Gas. , author =

  34. [35]

    Advanced Magnetohydrodynamics , author =

  35. [36]

    and Keppens, R

    Bergmans, J. and Keppens, R. and. Simulations of. Adaptive. doi:10.1007/3-540-27039-6_15 , isbn =

    work page doi:10.1007/3-540-27039-6_15

  36. [37]

    and Asensio, I

    Modestov, M. and Asensio, I. Alonso and Khomenko, E. , year =. Astronomy & Astrophysics , volume =. doi:10.1051/0004-6361/202558286 , url =

    work page doi:10.1051/0004-6361/202558286

  37. [38]

    Teunissen, Jannis and Olivares S. Foap4:

  38. [39]

    Journal of Computational Physics , year = 2012, month = feb, volume =

    Parallel, grid-adaptive approaches for relativistic hydro and magnetohydrodynamics. Journal of Computational Physics , year = 2012, month = feb, volume =. doi:10.1016/j.jcp.2011.01.020 , adsurl =

    doi:10.1016/j.jcp.2011.01.020 2012

  39. [40]

    Computer Physics Communications , keywords =

    A geometric multigrid library for quadtree/octree AMR grids coupled to MPI-AMRVAC. Computer Physics Communications , keywords =. doi:10.1016/j.cpc.2019.106866 , archivePrefix =. 1901.11370 , primaryClass =

    arXiv 2019

  40. [41]

    Journal of Computational Physics , keywords =

    The Numerical Simulation of Two-Dimensional Fluid Flow with Strong Shocks. Journal of Computational Physics , keywords =. doi:10.1016/0021-9991(84)90142-6 , adsurl =

    doi:10.1016/0021-9991(84)90142-6

  41. [42]

    Journal of Computational Physics , keywords =

    Local Adaptive Mesh Refinement for Shock Hydrodynamics. Journal of Computational Physics , keywords =. doi:10.1016/0021-9991(89)90035-1 , adsurl =

    doi:10.1016/0021-9991(89)90035-1

  42. [43]

    Sembach, K. R. and Wakker, B. P. and Savage, B. D. and Richter, P. and Meade, M. and Shull, J. M. and Jenkins, E. B. and Sonneborn, G. and Moos, H. W. , year =. Highly. The Astrophysical Journal Supplement Series , volume =. doi:10.1086/346231 , url =

    doi:10.1086/346231

  43. [44]

    and Savage, Blair D

    Fox, Andrew J. and Savage, Blair D. and Wakker, Bart P. and Richter, Philipp and Sembach, Kenneth R. and Tripp, Todd M. , year =. Highly. The Astrophysical Journal , volume =. doi:10.1086/381024 , url =

    work page doi:10.1086/381024

  44. [45]

    Fabian, A. C. and Sanders, J. S. and Crawford, C. S. and Conselice, C. J. and Gallagher, J. S. and Wyse, R. F. G. , year =. The Relationship between the Optical. Monthly Notices of the Royal Astronomical Society , volume =. doi:10.1046/j.1365-8711.2003.06856.x , url =

    arXiv 2003

  45. [46]

    2020 , month = mar, journal =

    Properties of the Simulated Circumgalactic Medium , author =. 2020 , month = mar, journal =. doi:10.1093/mnras/staa358 , url =

    doi:10.1093/mnras/staa358 2020

  46. [48]

    2019 , month = jul, journal =

    Simulations of Radiative Turbulent Mixing Layers , author =. 2019 , month = jul, journal =. doi:10.1093/mnras/stz1248 , url =

    doi:10.1093/mnras/stz1248 2019

  47. [49]

    , year =

    Kwak, Kyujin and Shelton, Robin L. , year =. Numerical. The Astrophysical Journal , volume =. doi:10.1088/0004-637X/719/1/523 , url =

    doi:10.1088/0004-637x/719/1/523

  48. [50]

    and Hennebelle, P

    Audit, E. and Hennebelle, P. , year =. On the Structure of the Turbulent Interstellar Clouds. Astronomy and Astrophysics , volume =. doi:10.1051/0004-6361/200912695 , url =

    doi:10.1051/0004-6361/200912695

  49. [51]

    and Shull, J

    Slavin, Jonathan D. and Shull, J. M. and Begelman, Mitchell C. , year =. Turbulent. The Astrophysical Journal , volume =. doi:10.1086/172494 , url =

    doi:10.1086/172494

  50. [52]

    2021 , month = nov, journal =

    A Model for Line Absorption and Emission from Turbulent Mixing Layers , author =. 2021 , month = nov, journal =. doi:10.1093/mnrasl/slab100 , url =

    doi:10.1093/mnrasl/slab100 2021

  51. [53]

    Del Zanna, Luca and Landi, Simone and Serafini, Lorenzo and Bugli, Matteo and Papini, Emanuele , year = 2024, month = jan, journal =. A. doi:10.3390/fluids9010016 , urldate =

    work page doi:10.3390/fluids9010016 2024

  52. [54]

    Kestener, Pierre , year = 2017, month = oct, journal =. Ramses-

    2017

  53. [55]

    and Mullen, Patrick D

    Stone, James M. and Mullen, Patrick D. and Fielding, Drummond and Grete, Philipp and Guo, Minghao and Kempski, Philipp and Most, Elias R. and White, Christopher J. and Wong, George N. , year = 2026, month = feb, journal =. doi:10.3847/1538-4365/ae3717 , urldate =

    doi:10.3847/1538-4365/ae3717 2026

  54. [56]

    Classical and Quantum Gravity , volume =

    Shankar, Swapnil and M. Classical and Quantum Gravity , volume =. doi:10.1088/1361-6382/acf2d9 , urldate =

    doi:10.1088/1361-6382/acf2d9

  55. [57]

    Lesur, G. R. J. and Baghdadi, S. and. Astronomy and Astrophysics , volume =. doi:10.1051/0004-6361/202346005 , urldate =

    doi:10.1051/0004-6361/202346005

  56. [58]

    and Goldbaum, Nathan J

    Schive, Hsi-Yu and ZuHone, John A. and Goldbaum, Nathan J. and Turk, Matthew J. and Gaspari, Massimo and Cheng, Chin-Yu , year = 2018, month = dec, journal =. doi:10.1093/mnras/sty2586 , urldate =

    doi:10.1093/mnras/sty2586 2018

  57. [59]

    arXiv e-prints , doi =

    Parthenon -- a Performance Portable Block-Structured Adaptive Mesh Refinement Framework , author =. arXiv e-prints , doi =

  58. [60]

    Liska, M. T. P. and Chatterjee, K. and Issa, D. and Yoon, D. and Kaaz, N. and Tchekhovskoy, A. and. H-. The Astrophysical Journal Supplement Series , volume =. doi:10.3847/1538-4365/ac9966 , urldate =

    doi:10.3847/1538-4365/ac9966

  59. [61]

    , year = 2026, month = ?, journal =

    Teunissen, Jannis and et al. , year = 2026, month = ?, journal =. doi:? , langid =

    2026

  60. [62]

    doi:10.1016/j.camwa.2020.03.023 , urldate =

    Keppens, Rony and Teunissen, Jannis and Xia, Chun and Porth, Oliver , year = 2021, month = jan, journal =. doi:10.1016/j.camwa.2020.03.023 , urldate =

    doi:10.1016/j.camwa.2020.03.023 2021

  61. [63]

    , year = 2024, month = mar, journal =

    Mayer, Markus and Dash, Ashutosh and Inghirami, Gabriele and Elfner, Hannah and Rezzolla, Luciano and Rischke, Dirk H. , year = 2024, month = mar, journal =. doi:10.48550/arXiv.2403.08668 , urldate =

    Pith/arXiv arXiv doi:10.48550/arxiv.2403.08668 2024

  62. [64]

    Physical Review D , volume =

    Hybrid Approach to Long-Term Binary Neutron-Star Simulations , author =. Physical Review D , volume =. doi:10.1103/PhysRevD.109.064061 , urldate =

    doi:10.1103/physrevd.109.064061

  63. [65]

    Gmunu: Toward Multigrid Based

    Cheong, Patrick Chi-Kit and Lin, Lap-Ming and Li, Tjonnie Guang Feng , year = 2020, month = jul, journal =. Gmunu: Toward Multigrid Based. doi:10.1088/1361-6382/ab8e9c , urldate =

    doi:10.1088/1361-6382/ab8e9c 2020

  64. [66]

    and Varniere, P

    Casse, F. and Varniere, P. and Meliani, Z. , year = 2017, month = jan, journal =. Impact of the Gravity of a. doi:10.1093/mnras/stw2572 , urldate =

    work page doi:10.1093/mnras/stw2572 2017

  65. [67]

    Towards the ultimate conservative difference scheme

    Van Leer, Bram , journal =. Towards the ultimate conservative difference scheme. 1974 , doi =

    1974

  66. [68]

    An adaptive finite element scheme for transient problems in

    L. An adaptive finite element scheme for transient problems in. Computer Methods in Applied Mechanics and Engineering , year =

  67. [69]

    The Astrophysical Journal , author =

    Radiative. The Astrophysical Journal , author =. 2008 , pages =. doi:10.1086/592561 , language =

    work page doi:10.1086/592561 2008

  68. [70]

    Annual Review of Astronomy and Astrophysics , author =

    Heating and. Annual Review of Astronomy and Astrophysics , author =. 1972 , pages =. doi:10.1146/annurev.aa.10.090172.002111 , language =

    arXiv 1972

  69. [71]

    The in situ formation of molecular and warm ionized gas triggered by hot galactic outflows , volume =

    Girichidis, Philipp and Naab, Thorsten and Walch, Stefanie and Berlok, Thomas , month = jul, year =. The in situ formation of molecular and warm ionized gas triggered by hot galactic outflows , volume =. Monthly Notices of the Royal Astronomical Society , publisher =. doi:10.1093/mnras/stab1203 , urldate =

    work page doi:10.1093/mnras/stab1203

  70. [72]

    Monthly Notices of the Royal Astronomical Society: Letters , author =

    The growth and entrainment of cold gas in a hot wind , volume =. Monthly Notices of the Royal Astronomical Society: Letters , author =. 2018 , pages =. doi:10.1093/mnrasl/sly131 , language =

    doi:10.1093/mnrasl/sly131 2018

  71. [73]

    On the hydrodynamic interaction of shock waves with interstellar clouds. 1:. The Astrophysical Journal , author =. 1994 , pages =. doi:10.1086/173554 , language =

    doi:10.1086/173554 1994

  72. [74]

    Journal of Computational Physics , keywords =

    High-order conservative finite difference GLM-MHD schemes for cell-centered MHD. Journal of Computational Physics , keywords =. doi:10.1016/j.jcp.2010.04.013 , archivePrefix =. 1001.2832 , primaryClass =

    Pith/arXiv arXiv 2010

  73. [75]

    Journal of Computational Physics , year = 2002, month = jan, volume =

    Hyperbolic Divergence Cleaning for the MHD Equations. Journal of Computational Physics , year = 2002, month = jan, volume =. doi:10.1006/jcph.2001.6961 , adsurl =

    arXiv 2002

  74. [76]

    Frozen-field Modeling of Coronal Condensations with MPI-AMRVAC. II. Optimization and Application in 3D Models. , keywords =. doi:10.3847/1538-4357/ad96af , archivePrefix =. 2411.16415 , primaryClass =

    Pith/arXiv arXiv

  75. [77]

    Frozen-field Modeling of Coronal Condensations with MPI-AMRVAC. I. Demonstration in Two-dimensional Models. , keywords =. doi:10.3847/1538-4357/ad4466 , archivePrefix =. 2404.17056 , primaryClass =

    Pith/arXiv arXiv

  76. [78]

    , keywords =

    An Exact Integration Scheme for Radiative Cooling in Hydrodynamical Simulations. , keywords =. doi:10.1088/0067-0049/181/2/391 , archivePrefix =. 0901.3146 , primaryClass =

    Pith/arXiv arXiv

  77. [79]

    T. The. 1997 , month = dec, journal =. doi:10.1006/jcph.1997.5813 , url =

    arXiv 1997

  78. [80]

    Vincent, Jonathan and Gong, Jing and Karp, Martin and Peplinski, Adam and Jansson, Niclas and Podobas, Artur and Jocksch, Andreas and Yao, Jie and Hussain, Fazle and Markidis, Stefano and Karlsson, Matts and Pleiter, Dirk and Laure, Erwin and Schlatter, Philipp , year =. Strong. International. doi:10.1145/3492805.3492818 , url =

    arXiv

  79. [81]

    and Sivaraman, Pranav and Kitson, Joy and Parasyris, Konstantinos and Menon, Harshitha and Minn, Isaac and Georgakoudis, Giorgis and Bhatele, Abhinav , year =

    Davis, Joshua H. and Sivaraman, Pranav and Kitson, Joy and Parasyris, Konstantinos and Menon, Harshitha and Minn, Isaac and Georgakoudis, Giorgis and Bhatele, Abhinav , year =. Taking. doi:10.48550/arXiv.2402.08950 , url =

    Pith/arXiv arXiv doi:10.48550/arxiv.2402.08950

  80. [82]

    Rossazza, Marco and Mignone, Andrea and Bugli, Matteo and Truzzi, Stefano and Riha, Lubomir and Panoc, Tomas and Vysocky, Ondrej and Shukla, Nitin and Romeo, Alessandro and Berta, Vittoria , year = 2025, month = nov, publisher =. The. doi:10.48550/arXiv.2511.20337 , urldate =

    doi:10.48550/arxiv.2511.20337 2025

Showing first 80 references.