pith. sign in

arxiv: 2604.16179 · v1 · submitted 2026-04-17 · ⚛️ physics.flu-dyn · physics.comp-ph· quant-ph

Quantum-Inspired Simulation of 2D Turbulent Rayleigh-B\'enard Convection

Nis-Luca van H\"ulst , Mario Guillaume Cecile , Hai-Yen Van , Tomohiro Hashizume , Eugene de Villiers , Dieter Jaksch This is my paper

Pith reviewed 2026-05-10 07:03 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn physics.comp-phquant-ph
keywords Rayleigh-Bénard convectionmatrix product statesturbulent heat transportNusselt numberbond dimension2D turbulenceBoussinesq equationscomputational fluid dynamics
0
0 comments X

The pith

Dynamical MPS simulations of 2D Rayleigh-Bénard convection recover accurate Nusselt numbers at Ra=10^10 with far fewer degrees of freedom than snapshot complexity predicts.

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

The paper applies Matrix Product State compression to two-dimensional turbulent thermal convection governed by the Boussinesq equations. An a priori decomposition of high-resolution snapshots indicates that the bond dimension needed to represent the fields grows steadily with Rayleigh number. However, when the governing equations are evolved directly inside the truncated MPS manifold at fixed bond dimension, statistical observables such as the mean Nusselt number are recovered with substantially lower cost. At Ra=10^10 this yields a relative error of only 1.8 percent while using a bond dimension comparable to that required at Ra=10^9 and cutting the number of degrees of freedom by nearly a factor of nine.

Core claim

Evolving the Boussinesq equations inside the compressed MPS representation preserves the essential statistical properties of the turbulence, including boundary-layer structure and plume dynamics that control heat transport, so that the bond dimension required for accurate Nusselt numbers scales far more favorably with Rayleigh number than the a priori representational complexity would suggest.

What carries the argument

Matrix Product State (MPS) representation of the velocity and temperature fields, with fixed bond dimension χ, inside which the governing equations are integrated dynamically rather than post-processed.

If this is right

  • Spectral content of the flow is recovered progressively with rising χ, confirming that the compression selectively retains the energetically relevant modes.
  • The computational savings demonstrated at Ra=10^10 suggest the approach remains practical for exploring the ultimate regime at still higher Rayleigh numbers.
  • Statistical quantities tied to large-scale circulation and heat transport converge faster than pointwise field accuracy under MPS truncation.

Where Pith is reading between the lines

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

  • Similar compression benefits might appear in other buoyancy-driven flows once the same dynamical evolution inside the MPS manifold is tested.
  • The favorable scaling could reduce the barrier to parameter sweeps that map the transition between soft and ultimate turbulence regimes.
  • Tensor-network methods of this type may eventually allow direct simulation of three-dimensional convection at Rayleigh numbers currently accessible only through heavily parameterized models.

Load-bearing premise

That truncating the MPS manifold and evolving the equations inside it does not systematically distort the boundary layers or plume statistics that determine the Nusselt number.

What would settle it

A high-resolution DNS at the same Ra showing that the time-averaged Nusselt number from the fixed-χ MPS run remains offset by more than the reported error even after χ is increased enough to capture all observed spatial and temporal scales.

Figures

Figures reproduced from arXiv: 2604.16179 by Dieter Jaksch, Eugene de Villiers, Hai-Yen Van, Mario Guillaume Cecile, Nis-Luca van H\"ulst, Tomohiro Hashizume.

Figure 1
Figure 1. Figure 1: Schematic representation of the staggered grid, [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A priori MPS compression and entanglement structure of turbulent RBC fields at Ra = 1010 . All panels display data from a reference DNS on a 2 11 × 2 10 grid (full simulation parameters are detailed in Appendix A). (a) Instantaneous temperature field θ at Ra = 1010 from a DNS simulation on a 2 11×2 10 grid. (b) Compressed MPS reconstruction of the same field (focusing on the green square in panel (a)), ret… view at source ↗
Figure 3
Figure 3. Figure 3: MPS compression accuracy and bond di￾mension scaling for turbulent RBC fields. (a) Relative ℓ2-error versus χ for various Ra, shown for u and θ. Results for v are nearly identical to u (see Appendix B). (b) Scaling of the minimum χ required to achieve ϵℓ2 < 10−2 with Ra, interpolated from the convergence data in (a) [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Dynamical validation at Ra = 108 . (a)– (c) Time evolution of ϵℓ2 (Eq. 9) for (a) velocity u, (b) tem￾perature θ, and (c) ϵNu. (d) Convergence of ϵ at tf = 1.56 tFF as a function of χ. (e) Temperature power spectral den￾sity Pθ(k) at t = 1.56 tFF. The dashed line indicates the classical Bolgiano–Obukhov scaling k −7/5 for the tempera￾ture spectrum [62, 63]. (f) Pointwise relative spectral error ϵPθ (k) = |… view at source ↗
Figure 6
Figure 6. Figure 6: Scaling analysis. (a) Relative error in the ensemble-averaged Nusselt number, ϵNu, shown as a function of χ for Ra = 108 , 109 , 1010. For each χ, Nu is computed from M realizations at the maximum sampling window T , for Ra = 109 and Ra = 1010 the same M, T values as in the corresponding [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 5
Figure 5. Figure 5: Convergence of turbulent heat trans￾port. (a–b) Evolution of the ensemble-averaged relative error ϵNu(T ) with increasing sampling window T for (a) Ra = 109 (M = 19) and (b) Ra = 1010 (M = 24). Horizontal dashed lines represent 10%, 5%, and 1% error thresholds. Shaded bands indicate the standard error of the mean (SEM, cf. Ap￾pendix F). (c–d) Probability density functions (PDFs) of the spatio-temporally av… view at source ↗
Figure 7
Figure 7. Figure 7: Temporal spectral analysis of the Nus￾selt number. (a, b) Power spectral density PNu(f) of the Nuinst(t) time series, computed via Welch’s method [66], com￾paring the DNS reference (solid gray) with MPS solutions (dashed) at varying bond dimension χ for (a) Ra = 109 and (b) Ra = 1010. (c) Relative error in the integrated spectral power ϵPNu = |EMPS Nu − EDNS Nu | / E DNS Nu (cf. Eq. 11) as a function of χ.… view at source ↗
Figure 8
Figure 8. Figure 8: Nusselt statistics. Probability density functions (PDF) of the instantaneous Nusselt number Nuinst (Eq. (4)) for (a–d) Ra = 109 (T = 24.3 tFF, M = 19) and (e–h) Ra = 1010 (T = 10.0 tFF, M = 24). Histograms are constructed with bin widths of 6.1 and 10.6, respectively, comparing the reference DNS (grey shaded area) with the MPS predictions (solid black lines). Subplots are arranged from left to right by inc… view at source ↗
Figure 9
Figure 9. Figure 9: Relative ℓ2-error as a function of χ for u, v, and θ at various Ra. Compare with [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Wall-clock time per timestep tstep (left) and peak memory usage in GB (right) as a function of Ra for DNS (solid, stars) and MPS (dashed, squares). MPS results are shown for χ ∈ {60, 90, . . . , 210}. Appendix F: Statistical Estimation and Error Propagation To characterize the convergence of the heat transport statistics, we define the time-averaged Nusselt number for each independent realization m ∈ {1, … view at source ↗
read the original abstract

Turbulent thermal convection governs heat transport in systems ranging from stellar interiors to industrial heat exchangers. Two-dimensional Rayleigh-B\'enard convection serves as a paradigm for these flows, reproducing key features such as thin boundary layers, large-scale circulation, and sustained plume dynamics. While Matrix Product State (MPS) methods have demonstrated significant compression of isothermal turbulent fields, their application to buoyancy-driven flows with active thermal coupling has remained unexplored. We apply MPS to two-dimensional Rayleigh-B\'enard convection with dynamical simulations up to $\mathrm{Ra} = 10^{10}$. An a priori decomposition of DNS snapshots up to $\mathrm{Ra} = 10^{11}$ shows that the bond dimension $\chi$ required to represent the flow fields grows without saturation, in contrast to the plateauing of $\chi$ reported for velocity fields in isothermal 2D turbulence. Crucially, however, dynamical simulations solving the governing equations directly in the compressed MPS format at fixed $\chi$ show that the $\chi$ required to recover statistical observables, such as the Nusselt number, scales significantly more favorably with $\mathrm{Ra}$ than the a priori complexity suggests. At $\mathrm{Ra} = 10^{10}$, a relative error of $1.8\%$ in the mean Nusselt number is achieved with a nearly 9-fold reduction in degrees of freedom, using a $\chi$ comparable to that required at $\mathrm{Ra} = 10^{9}$. Spectral analysis confirms the progressive recovery of spatial and temporal scales with increasing $\chi$. These findings establish MPS as a scalable tool for simulating thermally driven turbulence, suggesting the method may remain viable for investigations of the ultimate regime at substantially higher $\mathrm{Ra}$.

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 applies Matrix Product State (MPS) compression to two-dimensional Rayleigh-Bénard convection. An a priori analysis of DNS snapshots up to Ra=10^11 shows that the required bond dimension χ grows without saturation. Dynamical simulations evolved directly inside the χ-truncated MPS manifold, however, recover statistical observables such as the mean Nusselt number with substantially milder χ scaling; at Ra=10^10 a 1.8% relative error is reported together with a roughly nine-fold reduction in degrees of freedom using a χ comparable to that needed at Ra=10^9. Spectral recovery of spatial and temporal scales is also shown, and the authors conclude that MPS may remain viable for ultimate-regime studies at higher Ra.

Significance. If the central claim holds, the work would demonstrate that tensor-network compression can exploit the structure of thermally driven turbulence more efficiently than snapshot-based estimates suggest, offering a potential route to simulations at Rayleigh numbers inaccessible to conventional DNS. The explicit comparison between a priori and dynamical χ requirements, together with the concrete error metric on Nu, is a clear strength.

major comments (2)
  1. [Abstract and §4] Abstract and §4 (dynamical results): the 1.8% relative error on the mean Nusselt number at Ra=10^10 is the primary quantitative support for the claim of favorable scaling. However, no direct comparison of boundary-layer profiles, plume detachment statistics, or velocity-temperature cross-correlations is reported. Because the Nusselt number is an integral quantity controlled by precisely these structures, the absence of such diagnostics leaves open the possibility that truncation-induced bias in local dissipation or coherence is masked by the global metric.
  2. [§3 and §4] §3 (a priori decomposition) and §4 (dynamical runs): χ is chosen as the value that recovers the target Nusselt number to within the stated tolerance. While the a priori versus dynamical comparison itself is not tautological, this selection procedure means the performance metric used to judge success is also used to fix the compression level; an independent, a priori criterion for acceptable χ (e.g., based on energy or enstrophy spectra alone) would strengthen the claim that the manifold preserves essential dynamics without post-hoc tuning.
minor comments (2)
  1. [Abstract] The abstract states that error bars are absent from the reported metrics; inclusion of statistical uncertainty estimates on Nu and on the spectral quantities would improve reproducibility.
  2. [Abstract] Notation for the bond dimension is introduced as χ but the precise truncation scheme (e.g., singular-value cutoff or fixed-rank) is not stated in the abstract; a short clarifying sentence would help readers unfamiliar with MPS.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and the positive assessment of the potential impact of our work. We address each major comment below.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (dynamical results): the 1.8% relative error on the mean Nusselt number at Ra=10^10 is the primary quantitative support for the claim of favorable scaling. However, no direct comparison of boundary-layer profiles, plume detachment statistics, or velocity-temperature cross-correlations is reported. Because the Nusselt number is an integral quantity controlled by precisely these structures, the absence of such diagnostics leaves open the possibility that truncation-induced bias in local dissipation or coherence is masked by the global metric.

    Authors: We agree that the mean Nusselt number is a global integral and that direct local diagnostics would strengthen validation against possible truncation bias. The spectral analysis already presented in §4 shows progressive recovery of spatial and temporal scales, which provides indirect support for structure preservation. To address the concern explicitly, we will add boundary-layer profiles and velocity-temperature cross-correlations to the revised manuscript. revision: yes

  2. Referee: [§3 and §4] §3 (a priori decomposition) and §4 (dynamical runs): χ is chosen as the value that recovers the target Nusselt number to within the stated tolerance. While the a priori versus dynamical comparison itself is not tautological, this selection procedure means the performance metric used to judge success is also used to fix the compression level; an independent, a priori criterion for acceptable χ (e.g., based on energy or enstrophy spectra alone) would strengthen the claim that the manifold preserves essential dynamics without post-hoc tuning.

    Authors: The χ values employed in the dynamical runs are taken directly from the a priori analysis in §3, which measures the truncation error required to represent the full DNS fields. The dynamical simulations then test whether these χ suffice for statistical observables. We concur that an independent a priori metric would clarify the procedure. In the revision we will state explicitly that χ is fixed by the a priori field reconstruction error and will report the corresponding energy and enstrophy spectra to furnish an independent check. revision: yes

Circularity Check

0 steps flagged

No significant circularity; a priori field compression and dynamical MPS evolution remain distinct.

full rationale

The paper separates an a priori singular-value decomposition of DNS snapshots (determining χ needed for field reconstruction error) from dynamical evolution of the Boussinesq equations inside the fixed-χ MPS manifold, with the latter validated by comparing resulting Nusselt number and spectra against independent full DNS runs. No equation or procedure reduces the reported scaling advantage to a fit of the same observable used for truncation selection; the 1.8% Nu error at Ra=10^10 is an external benchmark, not a self-referential definition. No self-citations are load-bearing for the central claim, and no ansatz or uniqueness theorem is imported from prior author work.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the standard Boussinesq equations for incompressible flow with buoyancy and the tensor-network structure of MPS; the only tunable element is the bond dimension χ chosen per run to meet accuracy targets on observables.

free parameters (1)
  • bond dimension χ
    Varied across simulations to achieve target accuracy on statistical observables such as the Nusselt number; not a single global fit but a per-case compression parameter.
axioms (1)
  • domain assumption The 2D Rayleigh-Bénard equations can be evolved accurately inside the truncated MPS representation for the chosen χ values.
    Invoked when performing dynamical simulations directly in compressed format.

pith-pipeline@v0.9.0 · 5650 in / 1468 out tokens · 67429 ms · 2026-05-10T07:03:16.964327+00:00 · methodology

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

76 extracted references · 12 canonical work pages · 2 internal anchors

  1. [1]

    Quantencomputing mit neu- tralen Atomen

    [50], significantly improving efficiency for high-Ra sim- ulations. C. Nusselt Number The Nusselt number, denoted asNuin the following, serves as our most important system response quantify- ing the efficiency of convective heat transport relative to pure conduction [51]. Physically, it is defined as the ratio of total vertical heat flux to the conductive...

    2021

  2. [2]

    Fonds of the Hamburg Ministry of Science, Research, Equalities and Districts (BWFGB)

    and the Hamburg Quantum Computing Initiative (HQIC) project EFRE. The EFRE project is co-financed by ERDF of the European Union and by “Fonds of the Hamburg Ministry of Science, Research, Equalities and Districts (BWFGB)”. TH and DJ are partly funded by the Cluster of Excellence ‘Advanced Imaging of Mat- ter’oftheDeutscheForschungsgemeinschaft(DFG)|EXC 20...

    2056

  3. [3]

    C. A. Jones, Planetary Magnetic Fields and Fluid Dy- namos, Annual Review of Fluid Mechanics43, 583 (2011)

    2011

  4. [4]

    Kuhlbrodt, A

    T. Kuhlbrodt, A. Griesel, M. Montoya, A. Levermann, M. Hofmann, and S. Rahmstorf, On the driving processes of the Atlantic meridional overturning circulation, Re- views of Geophysics45, RG2001 (2007)

    2007

  5. [5]

    Vallis,Atmospheric and Oceanic Fluid Dynamics (2006) p

    G. Vallis,Atmospheric and Oceanic Fluid Dynamics (2006) p. 745

    2006

  6. [6]

    Rayleigh, On convection currents in a horizontal layer of fluid, when the higher temperature is on the under side, The Philosophical Magazine32, 529 (1916)

    L. Rayleigh, On convection currents in a horizontal layer of fluid, when the higher temperature is on the under side, The Philosophical Magazine32, 529 (1916)

    1916

  7. [7]

    Bénard, Les tourbillons cellulaires dans une nappe liq- uide transportant de la chaleur par convection en régime permanent, Annales de Chimie et de Physique23, 62 (1900)

    H. Bénard, Les tourbillons cellulaires dans une nappe liq- uide transportant de la chaleur par convection en régime permanent, Annales de Chimie et de Physique23, 62 (1900)

    1900

  8. [8]

    Oberbeck, Über die wärmeleitung der flüssigkeiten bei berücksichtigungderströmungeninfolgevontemperatur- differenzen, Annalen der Physik243, 271 (1879)

    A. Oberbeck, Über die wärmeleitung der flüssigkeiten bei berücksichtigungderströmungeninfolgevontemperatur- differenzen, Annalen der Physik243, 271 (1879)

  9. [9]

    Boussinesq,Théorie Analytique De La Chaleur: Mise En Harmonie Avec La Thermodynamique Et Avec La Théorie Mécanique De La Lumière(Gauthier-Villars, 1903)

    J. Boussinesq,Théorie Analytique De La Chaleur: Mise En Harmonie Avec La Thermodynamique Et Avec La Théorie Mécanique De La Lumière(Gauthier-Villars, 1903)

    1903

  10. [10]

    W. P. E. Bodenschatz and G. Ahlers, Recent develop- ments in rayleigh-bénard convection, Annual Review of Fluid Mechanics32, 709 (2000)

    2000

  11. [11]

    M. C. Cross and P. C. Hohenberg, Pattern formation outside of equilibrium, Rev. Mod. Phys.65, 851 (1993)

    1993

  12. [12]

    Ahlers, S

    G. Ahlers, S. Grossmann, and D. Lohse, Heat transfer and large scale dynamics in turbulent rayleigh-bénard convection, Rev. Mod. Phys.81, 503 (2009)

    2009

  13. [13]

    E. P. van der Poel, R. Ostilla-Mónico, J. Donners, and R. Verzicco, A pencil distributed finite difference code for strongly turbulent wall-bounded flows, Computers & Fluids116, 10 (2015)

    2015

  14. [14]

    X. Zhu, V. Mathai, R. J. A. M. Stevens, R. Verzicco, and D. Lohse, Transition to the ultimate regime in two- dimensional rayleigh-bénard convection, Phys. Rev. Lett. 120, 144502 (2018)

    2018

  15. [15]

    Zhang, Q

    Y. Zhang, Q. Zhou, and C. Sun, Statistics of kinetic and thermal energy dissipation rates in two-dimensional tur- bulent rayleigh–bénard convection, Journal of Fluid Me- chanics814, 165–184 (2017)

    2017

  16. [16]

    R. H. Kraichnan, Turbulent thermal convection at ar- bitrary prandtl number, The Physics of Fluids5, 1374 (1962)

    1962

  17. [17]

    Grossmann and D

    S. Grossmann and D. Lohse, Scaling in thermal convec- tion: a unifying theory, Journal of Fluid Mechanics407, 27–56 (2000)

    2000

  18. [18]

    Grossmann and D

    S. Grossmann and D. Lohse, Multiple scaling in the ulti- mate regime of thermal convection, Physics of Fluids23, 045108 (2011)

    2011

  19. [19]

    Shishkina and D

    O. Shishkina and D. Lohse, Ultimate regime of rayleigh- bénard turbulence: Subregimes and their scaling rela- tions for the nusselt vs rayleigh and prandtl numbers, Phys. Rev. Lett.133, 144001 (2024)

    2024

  20. [20]

    D.LohseandO.Shishkina,Ultimaterayleigh-bénardtur- bulence, Rev. Mod. Phys.96, 035001 (2024)

    2024

  21. [21]

    Chavanne, F

    X. Chavanne, F. Chillà, B. Castaing, B. Hébral, B. Chabaud, and J. Chaussy, Observation of the ulti- mate regime in rayleigh-bénard convection, Phys. Rev. Lett.79, 3648 (1997)

    1997

  22. [22]

    J. D. Scheel and J. Schumacher, Local boundary layer scales in turbulent rayleigh–bénard convection, Journal of Fluid Mechanics758, 344 (2014)

    2014

  23. [23]

    Moin and K

    P. Moin and K. Mahesh, Direct numerical simulation: A tool in turbulence research, Annual Review of Fluid Mechanics30, 539 (1998)

    1998

  24. [24]

    S. B. Pope,Turbulent Flows(Cambridge University Press, 2000)

    2000

  25. [25]

    Chandrasekhar,Hydrodynamic and Hydromagnetic Stability(Oxford University Press, 1961)

    S. Chandrasekhar,Hydrodynamic and Hydromagnetic Stability(Oxford University Press, 1961)

    1961

  26. [26]

    F. H. Busse, Non-linear properties of thermal convection, Reports on Progress in Physics41, 1929 (1978)

    1929

  27. [27]

    M. M. Rai and P. Moin, Direct simulations of turbulent flow using finite-difference schemes, Journal of Compu- tational Physics96, 15 (1991)

    1991

  28. [28]

    I. V. Oseledets, Tensor-train decomposition, SIAM Jour- nal on Scientific Computing33, 2295 (2011)

    2011

  29. [29]

    V. A. Kazeev and B. N. Khoromskij, Low-rank explicit qtt representation of the laplace operator and its inverse, SIAM Journal on Matrix Analysis and Applications33, 742 (2012)

    2012

  30. [30]

    Arenstein , author M

    L. Arenstein, M. Mikkelsen, and M. Kastoryano, Fast and flexible quantum-inspired differential equation solvers with data integration (2025), arXiv:2505.17046 [math.NA]

    work page arXiv 2025

  31. [31]

    S. R. White, Density matrix formulation for quantum renormalization groups, Physical Review Letters69, 2863 (1992)

    1992

  32. [32]

    V. M. F. Verstraete and J. I. Cirac, Matrix product states, projected entangled pair states, and variational renormalization group methods for quantum spin sys- tems, Advances in Physics57, 143 (2008)

    2008

  33. [33]

    U.Schollwöck,Thedensity-matrixrenormalizationgroup in the age of matrix product states, Annals of Physics 326, 96–192 (2011)

    2011

  34. [34]

    Orús, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Annals of Physics349, 117 (2014)

    R. Orús, A practical introduction to tensor networks: Matrix product states and projected entangled pair states, Annals of Physics349, 117 (2014)

    2014

  35. [35]

    Gourianov, M

    N. Gourianov, M. Lubasch, S. Dolgov, Q. Y. van den Berg, H. Babaee, P. Givi, M. Kiffner, and D. Jaksch, A quantum-inspired approach to exploit turbulence struc- tures, Nature Computational Science2, 30 (2022)

    2022

  36. [36]

    Gourianov,Exploiting the Structure of Turbulence with Tensor Networks, Phd thesis, University of Oxford, Oxford, UK (2022)

    N. Gourianov,Exploiting the Structure of Turbulence with Tensor Networks, Phd thesis, University of Oxford, Oxford, UK (2022)

    2022

  37. [37]

    Gourianov, P

    N. Gourianov, P. Givi, D. Jaksch, and S. B. Pope, Tensor networks enable the calculation of turbulence probability distributions, Science Advances11, eads5990 (2025)

    2025

  38. [38]

    Hölscher, P

    L. Hölscher, P. Rao, L. Müller, J. Klepsch, A. Luckow, T. Stollenwerk, and F. K. Wilhelm, Quantum-inspired fluid simulation of two-dimensional turbulence with GPU acceleration, Phys. Rev. Res.7, 013112 (2025)

    2025

  39. [39]

    Pisoni, R

    S. Pisoni, R. D. Peddinti, E. Tiunov, S. E. Guzman, and L. Aolita, Compression, simulation, and synthesis of tur- bulent flows with tensor trains (2025), arXiv:2506.05477 [physics.flu-dyn]

    work page arXiv 2025

  40. [40]

    Pinkston, N

    R. Pinkston, N. Gourianov, H. Alipanah, P. Givi, D. Jaksch, and J. J. Mendoza-Arenas, Matrix prod- uct state simulation of reacting shear flows (2025), arXiv:2512.13661 [physics.flu-dyn]. 12

    work page arXiv 2025

  41. [41]

    R. J. J. Connor, C. W. Duncan, and A. J. Daley, Tensor network methods for the gross-pitaevskii equation on fine grids (2025), arXiv:2507.01149 [cond-mat.quant-gas]

    work page arXiv 2025

  42. [42]

    G´ omez-Lozada, N

    F. Gómez-Lozada, N. Perico-García, N. Gourianov, H. Salman, and J. J. Mendoza-Arenas, Simulating quan- tum turbulence with matrix product states (2025), arXiv:2508.12191 [quant-ph]

    work page arXiv 2025

  43. [43]

    Ye and N

    E. Ye and N. F. G. Loureiro, Quantum-inspired method for solving the vlasov-poisson equations, Phys. Rev. E 106, 035208 (2022)

    2022

  44. [44]

    Ameri, E

    A. Ameri, E. Ye, P. Cappellaro, H. Krovi, and N. F. Loureiro, Quantum algorithm for the linear vlasov equa- tion with collisions, Phys. Rev. A107, 062412 (2023)

    2023

  45. [45]

    Quantized tensor networks for solving the Vlasov-Maxwell equations

    E. Ye and N. Loureiro, Quantized tensor net- works for solving the vlasov-maxwell equations (2024), arXiv:2311.07756 [physics.comp-ph]

    work page internal anchor Pith review arXiv 2024

  46. [46]

    R. J. J. Connor, P. Soin, C. W. Duncan, and A. J. Daley, Tensor network approaches for plasma dynamics (2025), arXiv:2512.15924 [physics.plasm-ph]

    work page arXiv 2025

  47. [47]

    F. H. Harlow and J. E. Welch, Numerical Calculation of Time-Dependent Viscous Incompressible Flow of Fluid with Free Surface, The Physics of Fluids8, 2182 (1965)

    1965

  48. [48]

    J. H. Ferziger and M. Perić,Computational Methods for Fluid Dynamics, 3rd ed. (Springer, Berlin, 2002)

    2002

  49. [49]

    R. J. LeVeque,Finite Difference Methods for Ordinary and Partial Differential Equations: Steady-State and Time-Dependent Problems(SIAM, 2007)

    2007

  50. [50]

    U. M. A. S. J. Ruuth and R. J. Spiteri, Implicit-explicit runge-kutta methods for time-dependent partial differen- tial equations, Applied Numerical Mathematics25, 151 (1997)

    1997

  51. [51]

    A. J. Chorin, Numerical solution of the navier-stokes equations, Mathematics of Computation22, 745 (1968)

    1968

  52. [52]

    Courant, K

    R. Courant, K. Friedrichs, and H. Lewy, Über die partiellen differenzengleichungen der mathematischen physik, Mathematische Annalen100, 32 (1928)

    1928

  53. [53]

    Y. A. Çengel, J. M. Cimbala, and R. H. Turner,Funda- mentals of Thermal-Fluid Sciences,5thed.(McGraw-Hill Education, 2017)

    2017

  54. [54]

    S. S. Vallender, Calculation of the wasserstein dis- tance between probability distributions on the line, The- ory of Probability & Its Applications18, 784 (1974), https://doi.org/10.1137/1118101

    work page doi:10.1137/1118101 1974

  55. [55]

    C. R. Givens and R. M. Shortt, A class of wasserstein metrics for probability distributions., Michigan Mathe- matical Journal31, 231 (1984)

    1984

  56. [56]

    Vidal, Efficient classical simulation of slightly entan- gled quantum computations, Physical Review Letters91, 147902 (2003)

    G. Vidal, Efficient classical simulation of slightly entan- gled quantum computations, Physical Review Letters91, 147902 (2003)

    2003

  57. [57]

    Quantum-Inspired Tensor-Network Fractional-Step Method for Incompressible Flow in Curvilinear Coordinates

    N.-L. van Hülst, P. Siegl, P. Over, S. Ben- goechea, T. Hashizume, M. G. Cecile, T. Rung, and D. Jaksch, Quantum-inspired tensor-network fractional- step method for incompressible flow in curvilinear coor- dinates (2025), arXiv:2507.05222 [physics.flu-dyn]

    work page internal anchor Pith review arXiv 2025

  58. [58]

    van Hülst, T

    N.-L. van Hülst, T. Panagos, G. S. Reese, S. Panahiyan, and T. Hashizume, Quantum solvers: Predic- tive aeroacoustic & aerodynamic modeling (2025), arXiv:2507.21683 [quant-ph]

    work page arXiv 2025

  59. [59]

    Holtz, T

    S. Holtz, T. Rohwedder, and R. Schneider, The alternat- ing linear scheme for tensor optimization in the tensor train format, SIAM Journal on Scientific Computing34, A683 (2012)

    2012

  60. [60]

    S. V. Dolgov, B. N. Khoromskij, and I. V. Oseledets, Fast solution of parabolic problems in the tensor train/quan- tized tensor train format with initial application to the Fokker–Planck equation, SIAM Journal on Scientific Computing34, A3016 (2012)

    2012

  61. [61]

    S.V.DolgovandD.V.Savostyanov,Alternatingminimal energy methods for linear systems in higher dimensions, SIAM J. Sci. Comput.36(2013)

    2013

  62. [62]

    J. Chen, E. Stoudenmire, and S. R. White, Quantum fouriertransformhassmallentanglement,PRXQuantum 4, 040318 (2023)

    2023

  63. [63]

    Y. N. Fernández, M. K. Ritter, M. Jeannin, J.-W. Li, T. Kloss, T. Louvet, S. Terasaki, O. Parcollet, J. von Delft, H. Shinaoka, and X. Waintal, Learning tensor net- works with tensor cross interpolation: New algorithms and libraries, SciPost Phys.18, 104 (2025)

    2025

  64. [64]

    Bolgiano Jr., Turbulent spectra in a stably strati- fied atmosphere, Journal of Geophysical Research (1896- 1977)64, 2226 (1959)

    R. Bolgiano Jr., Turbulent spectra in a stably strati- fied atmosphere, Journal of Geophysical Research (1896- 1977)64, 2226 (1959)

    1977

  65. [65]

    R.SamuelandM.K.Verma,Bolgiano-obukhovscalingin two-dimensional rayleigh-bénard convection at extreme rayleigh numbers, Phys. Rev. Fluids9, 023502 (2024)

    2024

  66. [66]

    J. A. Vastano and R. D. Moser, Short-time lyapunov exponent analysis and the transition to chaos in tay- lor–couette flow, Journal of Fluid Mechanics233, 83–118 (1991)

    1991

  67. [67]

    A. Xu, X. Chen, F. Wang, and H.-D. Xi, Correlation of internal flow structure with heat transfer efficiency in turbulent Rayleigh-Bénard convection, Physics of Fluids 32, 105112 (2020)

    2020

  68. [68]

    P. Welch, The use of fast fourier transform for the estima- tion of power spectra: A method based on time averaging overshort, modifiedperiodograms,IEEETransactionson Audio and Electroacoustics15, 70 (1967)

    1967

  69. [69]

    J. P. Boyd,Chebyshev and Fourier Spectral Methods, 2nd ed. (Dover Publications, Mineola, NY, 2001)

    2001

  70. [70]

    A. A. Michailidis, C. Fenton, and M. Kiffner, Element-wise multiplication of tensor trains, SIAM Journal on Scientific Computing47, B1158 (2025), https://doi.org/10.1137/24M1714149

    work page doi:10.1137/24m1714149 2025

  71. [71]

    Z. Meng, Y. Khoo, J. Li, and E. M. Stoudenmire, Recur- sive sketched interpolation: Efficient hadamard products of tensor trains (2026), arXiv:2602.17974 [quant-ph]

    work page arXiv 2026

  72. [72]

    E. M. Stoudenmire and S. R. White, Real-space parallel density matrix renormalization group, Phys. Rev. B87, 155137 (2013)

    2013

  73. [73]

    Secular, N

    P. Secular, N. Gourianov, M. Lubasch, S. Dolgov, S. R. Clark,andD.Jaksch,Paralleltime-dependentvariational principle algorithm for matrix product states, Phys. Rev. B101, 235123 (2020)

    2020

  74. [74]

    Lubasch, J

    M. Lubasch, J. Joo, P. Moinier, M. Kiffner, and D. Jaksch, Variational quantum algorithms for nonlinear problems, Phys. Rev. A101, 010301 (2020)

    2020

  75. [75]

    Jaksch, P

    D. Jaksch, P. Givi, A. J. Daley, and T. Rung, Variational quantum algorithms for computational fluid dynamics, AIAA Journal61, 1885 (2023)

    2023

  76. [76]

    Siegl, G

    P. Siegl, G. S. Reese, T. Hashizume, N.-L. van Hülst, and D. Jaksch, Tensor-programmable quantum circuits for solving differential equations, Phys. Rev. Res.8, 013052 (2026). 13 101 102 103 χ 10−5 10−4 10−3 10−2 10−1 100 Rel. Error ϵℓ2 (a) Ra = 10 8 Ra = 10 9 Ra = 10 10 Ra = 10 11 θ u v θ u v 108 109 1010 1011 Ra 20 100 400 χ (b) θ u v Figure 9. Relativeℓ...

    2026