A hybrid quantum-classical lattice kinetic scheme with tensor-train feedback for incompressible flows

arXiv:2611.04500 · math-ph, math.MP · Submitted 2026-11-03 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematical Physics. The day in mathematical physics: rigorous models of physical systems.

Kai: We're looking at the paper: "A hybrid quantum-classical lattice kinetic scheme with tensor-train feedback for incompressible flows".

Mira: Here's the gist of the paper. A hybrid quantum–classical lattice kinetic scheme with tensor-train feedback for incompressible flows constructs an interface that connects quantum propagation to nonlinear feedback and assesses the separate demands of compression, velocity accuracy, and readout precision The Hybrid Interface Construction The interface retains a classical tensor-train (TT) representation of the physical distribution after each accepted step to supply local moments and gradients for coefficient evaluation The stored cores provide the next coefficients and prepare the next quantum input, allowing the system to proceed without replaying previous quantum evolution The cycle involves initialization, classical coefficient evaluation, quantum propagation and coordinate recovery, then core reconstruction and acceptance Feedback Control Conditions The analysis identifies conditions under which this interface controls feedback error and the resources needed to meet those conditions Under positive, unit-sum reference coefficients and the stated density and Lipschitz sensitivity assumptions, Proposition 4.1 propagates local coefficient and update defects The single-step circuit retains the actual nodewise coefficient sum, including the correction caused by discrete velocity divergence Resource Requirements Analysis The resource model counts classical TT operations, circuit preparation and execution, sampling, and verification at the required precision The total hybrid work is bounded by Ctotal = Wclassical + Cmeas + Ccheck + CI/O The conditional total complexity bound is O(Wclassical + Ccheck + CI/O + NtSpCmaxdmax max

1, H2 dmax: log 2dmax δmin!

Kai: Numerical Diagnostics and Findings The numerical tests examine whether the information retained by the hybrid interface is sufficient for accurate flow feedback and what precision its readout demands Classical Taylor–Green refinement shows nearly second-order velocity convergence at fixed Reynolds number and decreasing Mach number, verifying the single-step scaled-update algebra The periodic 80-step feedback test examines repeated reconstruction, where rank four nearly eliminates the vortex even though the final distribution error is below 1% Tighter sampling improves both distribution and velocity errors, but both metrics approach a trajectory that still contains initial compression and finite-sweep fitting errors Readout Precision vs.

Mira: Accuracy The four-step statistical readout test shows that tighter overlap estimates improve both distribution and velocity errors, but the trajectory approaches an exact-overlap baseline that retains initial compression and finite-sweep fitting errors The prescribed sufficient sampling budget quantifies the precision cost of this approach, rather than a necessary lower bound on readout The mean total budget is reported as 5.37 × 1013 shots for four steps The analysis distinguishes the conditions for encoding one step from those for controlling feedback over many steps The resource model counts classical TT operations, circuit preparation and execution, sampling, and verification at the required precision The numerical diagnostics identify why compression, feedback accuracy, and readout precision require separate control A full implementation can then be compared with purely classical TT propagation at matched output tolerances The present classical and statistical diagnostics provide the interface tests and error accounting for that comparison References

1: S.

Kai: Hou, Q.

Paper summary: Kai: So to wrap up this discussion on "A hybrid quantum–classical lattice kinetic scheme with tensor-train feedback for incompressible flows," the core finding is that you can separate the conditions for encoding one step from those for controlling feedback over many steps >

Mira: This means we're not just looking at a single successful simulation; we're looking at how to manage the whole process of getting that flow right through iteration >

Lev: For someone building real hardware, it confirms that you need a resource model that counts classical TT operations, circuit preparation and execution, sampling, and verification at the required precision >

Kai: And the numerical diagnostics clearly show why compression, feedback accuracy, and readout precision require separate control mechanisms in this hybrid approach >

Mira: It sets up a clear path for comparing this quantum-classical scheme against purely classical TT propagation when you can match the output tolerances between them >

Conclusion: Kai: So we’ve looked at how this hybrid setup works across all these steps, and now we're going to talk about what this whole paper is actually trying to say about incompressible flows and feedback >

Mira: It's called "A hybrid quantum–classical lattice kinetic scheme with tensor-train feedback for incompressible flows," which sounds complicated, but basically it’s a way to link quantum particle movement with classical math that learns from its mistakes >

Lev: From my side, what this paper really tackles is how to manage the complexity of getting accurate results when you have both quantum noise and classical approximation >

Kai: Right. So at its heart, the authors are showing how you can build a system where quantum propagation feeds information back into a classical tensor-train model that helps guide it >

Mira: They’re not just doing one thing; they’re breaking it down. They show how to separate the demands of getting the flow distribution right from making sure your feedback loop is stable >

Lev: I think the most important thing they establish is those clear conditions for when this interface actually controls the feedback error, which is a big deal when you're trying to build something that runs on real hardware >

Kai: It’s about finding those specific settings—like positive unit-sum reference coefficients—where the system doesn't just fail, but it actually manages the corrections caused by things like velocity divergence >

Mira: They spend a lot of time quantifying what resources you need to meet those conditions; they give you a total complexity bound that includes classical work and the measurement stuff >

Lev: That resource model is crucial because it tells you exactly how much classical tensor-train math and how many shots you need for sampling and verification, so we know the overhead when we try to run this on actual hardware >

Kai: So, for someone just listening, it means they’re proving that you can handle the trade-off between getting the right shape of a fluid flow and making sure your iterative correction process doesn't blow up >

Mira: And they’re setting up a comparison point, showing how this hybrid approach stacks up against just using purely classical tensor-train propagation when you have matched output tolerances >

Lev: It gives us a solid benchmark for understanding where the quantum part is actually adding value versus where the classical structure is just helping organize the calculation >

Kai: So we've seen the mechanics, now we need to figure out what this means for how we actually build these kinds of complex simulations in practice >

Yongwei Guo, Yuan Yu, Yue Yu, Yuting Zhou

School of Mathematics and Computational Science, Xiangtan University · Hunan Research Center of the Basic Discipline Fundamental Algorithmic Theory and Novel Computational Methods · National Center for Applied Mathematics in Hunan

math-ph, math.MP

Submitted: 2026-11-03

Updated: 2026-11-03

Comments: 42 pages, 4 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 83/100

The gist: A hybrid quantum–classical lattice kinetic scheme with tensor-train feedback for incompressible flows constructs an interface that connects quantum propagation to nonlinear feedback and assesses…

Key concepts

Hybrid Quantum–Classical Lattice Kinetic Scheme
This is a combined approach where quantum mechanics handles the complex particle movement (propagation), while classical methods manage the local information and feedback. It aims to leverage the strengths of both to solve fluid dynamics problems.
Tensor-Train Feedback
The system uses a tensor-train representation, which is an efficient way to store and manipulate large amounts of data related to the flow's distribution. This structure provides local moments and gradients needed for classical coefficient evaluation, which then feeds back into the quantum steps.
Resource Requirements Analysis
This section quantifies the computational cost of the hybrid method. It calculates total work by summing up classical operations (TT, circuit execution) and measurement costs. This analysis determines how much computing power is needed to achieve a specific level of accuracy.

Terminology

Summary

A hybrid quantum–classical lattice kinetic scheme with tensor-train feedback for incompressible flows constructs an interface that connects quantum propagation to nonlinear feedback and assesses the separate demands of compression, velocity accuracy, and readout precision

The Hybrid Interface Construction

The interface retains a classical tensor-train (TT) representation of the physical distribution after each accepted step to supply local moments and gradients for coefficient evaluation The stored cores provide the next coefficients and prepare the next quantum input, allowing the system to proceed without replaying previous quantum evolution The cycle involves initialization, classical coefficient evaluation, quantum propagation and coordinate recovery, then core reconstruction and acceptance

Feedback Control Conditions

The analysis identifies conditions under which this interface controls feedback error and the resources needed to meet those conditions Under positive, unit-sum reference coefficients and the stated density and Lipschitz sensitivity assumptions, Proposition 4.1 propagates local coefficient and update defects The single-step circuit retains the actual nodewise coefficient sum, including the correction caused by discrete velocity divergence

Resource Requirements Analysis

The resource model counts classical TT operations, circuit preparation and execution, sampling, and verification at the required precision The total hybrid work is bounded by Ctotal = Wclassical + Cmeas + Ccheck + CI/O The conditional total complexity bound is O(Wclassical + Ccheck + CI/O + NtSpCmaxdmax max[1, H2 dmax] log 2dmax δmin!

Numerical Diagnostics and Findings

The numerical tests examine whether the information retained by the hybrid interface is sufficient for accurate flow feedback and what precision its readout demands Classical Taylor–Green refinement shows nearly second-order velocity convergence at fixed Reynolds number and decreasing Mach number, verifying the single-step scaled-update algebra The periodic 80-step feedback test examines repeated reconstruction, where rank four nearly eliminates the vortex even though the final distribution error is below 1% Tighter sampling improves both distribution and velocity errors, but both metrics approach a trajectory that still contains initial compression and finite-sweep fitting errors

Readout Precision vs. Accuracy

The four-step statistical readout test shows that tighter overlap estimates improve both distribution and velocity errors, but the trajectory approaches an exact-overlap baseline that retains initial compression and finite-sweep fitting errors The prescribed sufficient sampling budget quantifies the precision cost of this approach, rather than a necessary lower bound on readout The mean total budget is reported as 5.37 × 1013 shots for four steps

Conclusion

The analysis distinguishes the conditions for encoding one step from those for controlling feedback over many steps The resource model counts classical TT operations, circuit preparation and execution, sampling, and verification at the required precision The numerical diagnostics identify why compression, feedback accuracy, and readout precision require separate control A full implementation can then be compared with purely classical TT propagation at matched output tolerances The present classical and statistical diagnostics provide the interface tests and error accounting for that comparison

References

[1] S. Hou, Q. Zou, S. Chen, G. D. Doolen, A. C. Cogley, Simulation of cavity flow by the lattice Boltzmann method

[2] J. D. Sterling, S. Chen, Stability analysis of lattice Boltzmann methods

[3] T. Kr¨uger, H. Kusumaatmaja, A. Kuzmin, O. Shardt, G. Silva, E. M. Viggen, The Lattice Boltzmann Method

[4] T. Astoul, G. Wissocq, J.-F. Boussuge, A Sengissen, P Sagaut, Analysis and reduction of spurious noise generated at grid refinement interfaces with the lattice Boltzmann method

[5] F. Renard, G. Wissocq, J.-F. Boussuge, P Sagaut, A linear stability analysis of compressible hybrid lattice Boltzmann methods

[6] Y. Xiao, L. Yang, C Shu, Y Du, H Dong, J Wu Quantum lattice kinetic scheme for solving two-dimensional and three-dimensional incompressible flows

[7] D. W Berry, High-order quantum algorithm for solving linear differential equations

[8] B. N Todorova, R Steijl Quantum algorithm for the collisionless Boltzmann equation

[9] L Budinski Quantum algorithm for the advection-diffusion equation simulated with the lattice Boltzmann method

[10] P C S Costa, D An, Y R Sanders, Y Su, R Babbush D W Berry Optimal scaling quantum linear-systems solver via discrete adiabatic theorem

[11] S Jin, N Liu, Y Yu Time complexity analysis of quantum difference methods for linear high dimensional and multiscale partial differential equations

[12] D Fang, L Lin, Y Tong Time-marching based quantum solvers for time-dependent linear differential equations

[13] M A Schalkers, M M¨oller Efficient and fail-safe quantum algorithm for the transport equation

[14] S Jin, N Liu, Y Yu Schrödingerization based quantum algorithms for the time-fractional heat equation

[15] I Novikau, I Joseph An efficient explicit implementation of a near-optimal quantum algorithm for simulating linear dissipative differential equations

[16] D Wawrzyniak J Winter S Schmidt T Indinger U Schramm C Janßen N A Adams Unitary quantum algorithm for the lattice-Boltzmann method

[17] L Xu, M Li L Zhang H Sun J Yao Improved quantum lattice Boltzmann method for advection-diffusion equations with a linear collision model

[18] D Wawrzyniak J Winter S Schmidt T Indinger C F Janßen U Schramm N A Adams Linearized quantum lattice-Boltzmann method for the advection-diffusion equation using dynamic circuits

[19] A Nagel, J Lowe Quantum lattice Boltzmann method for multiple time steps without reinitialization for linear advection-diffusion problems

[20] C A Georgescu M A Schalkers M M¨oller Fully quantum lattice gas automata building blocks for computational basis state encodings

[21] E D Kumar S H Frankel Quantum unitary matrix representation of the lattice Boltzmann model for low Reynolds fluid flow simulation

[22] Y He Y Yu Y Yu Y Zhou Energy-scaled quantum lattice Boltzmann methods for advection– diffusion equations

[23] J.-P Liu H Ø Kolden H K Krovi N F Loureiro K Trivisa A M Childs Efficient quantum algorithm for dissipative nonlinear differential equations

[24] S Jin, N Liu, Y Yu Time complexity analysis of quantum algorithms via linear representations for nonlinear ordinary and partial differential equations

[25] M A Schalkers M M¨oller On the importance of data encoding in quantum Boltzmann methods

[26] K Asztalos R Steijl R Maulik Reduced-order modeling on a near-term quantum computer

[27] B Wang Z Meng Y Zhao Y Yang Quantum lattice Boltzmann method for simulating nonlinear fluid dynamics

[28] T Duong M M¨oller N Hosters Quantum lattice Boltzmann with denoising collision operators

[29] S Kocherla Z Song F E Chrit B Gard E F Dumitrescu A Alexeev S H Bryngelson Fully quantum algorithm for mesoscale fluid simulations with application to partial differential equations

[30] A Tiwari J Iaconis J Jojo S Ray M Roetteler C Hill J Pathak Algorithmic advances towards a realizable quantum lattice Boltzmann method

[31] C Sch¨on E Solano F Verstraete J I Cirac M M Wolf Sequential generation of entangled multiqubit states

[32] I V Oseledets Tensor-train decomposition SIAM Journal on Scientific Computing 33 (5)

[33] P Gelß S Klus S Matera C Sch¨utte Nearest-neighbor interaction systems in the tensor-train format J Comput. Phys. 341 (2017)

[34] V Ehrlacher D Lombardi A dynamical adaptive tensor method for the Vlasov–Poisson system J Comput. Phys. 339 (2017)

[35] A M P Boelens D Venturi D M Tartakovsky Parallel tensor methods for high-dimensional linear PDEs J Comput. Phys. 375 (2018)

[36] A M P Boelens D Venturi D M Tartakovsky Tensor methods for the Boltzmann-BGK equation J Comput. Phys. 421 (2020)

[37] W Guo J Qiu A low rank tensor representation of linear transport and nonlinear Vlasov solutions and their associated flow maps J Comput. Phys. 458 (2022)

[38] L Einkemmer K Kormann J Kusch R G McClarren J M Qiu A review of low-rank methods for time-dependent kinetic simulations J Comput. Phys. 538 (2025)

[39] E M ˚Asgrim L Pennati M Pasquale S Markidis Tensor network compression for fully spectral Vlasov–Poisson simulation J Comput. Phys. 567 (2026)

[40] A Gorodetsky J Jakeman Gradient-based optimization for regression in the functional tensor-train format J Comput. Phys. 374 (2018)

[41] A Dektor D Venturi Dynamic tensor approximation of high-dimensional nonlinear PDEs J Comput. Phys. 437 (2021)

[42] M E Danis D Truong I Boureima O Korobkin K Ø Rasmussen B S Alexandrov Tensor-train WENO scheme for compressible flows J Comput. Phys. 529 (2025)

[43] J Gidi P García-Molina L Tagliacozzo J J García-Ripoll Pseudospectral method for solving PDEs using matrix product states J Comput. Phys. 539 (2025)

[44] J Coughlin J Hu U Shumlak Robust and conservative dynamical low-rank methods for the Vlasov equation via a novel macro-micro decomposition J Comput. Phys. 509 (2024)

[45] A Dektor L Einkemmer Interpolatory dynamical low-rank approximation for the 3+3d Boltzmann–BGK equation J Comput. Phys. 547 (2026)

[46] G Wang J Hu Dynamical tensor train approximation for kinetic equations J Comput. Phys. 558 (2026)

[47] G Brassard P Høyer M Mosca A Tapp Quantum amplitude amplification and estimation Contemp. Math. 305 (2002)

--- Page 39 ---

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this manuscript, the authors used OpenAI Codex to assist with manuscript structure, wording, and literature verification The authors reviewed and edited the content as needed and take full responsibility for the content of the published article

--- Page 42 ---

Data availability

Data and source code supporting the findings of this study are available from the corresponding author upon reasonable request

--- Page 39 ---

Acknowledgments

Yuan Yu was supported by the National Natural Science Foundation of China (No. 12101527) and the Natural Science Foundation of Hunan Province (No. p. 0 p. 0 26JJ60005) Yue Yu was supported by NSFC grant No. p. 0 p. 0 12341104, the Key Project of Scientific Research Project of Hunan Provincial Department of Education (No. p. 0 p. 0 24A0100), the Science and Technology Innovation Program of Hunan Province (No. p. 0 p. 0 26RC3150), the general program of Hunan Provincial Natural Science Foundation (No. p. 0 p. 0 26JJ50003) and partially supported by NSFC grant No. 12341104 Computational resources were provided by the High Performance Computing Platform of Xiangtan University

--- Page 39 ---

Acknowledgments

Yuan Yu was supported by the National Natural Science Foundation of China (No. 12101527) and the Natural Science Foundation of Hunan Province (No. 26JJ60005) Yue Yu was supported by NSFC grant No. 12341104, the Key Project of Scientific Research Project of Hunan Provincial Department of Education (No. 24A0100), the Science and Technology Innovation Program of Hunan Province (No. 26RC3150), the general program of Hunan Provincial Natural Science Foundation (No. 26JJ50003) and partially supported by NSFC grant No.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI systems can perform full-relaxation lattice kinetic scheme simulations by integrating a tensor train interface that retains reusable classical cores at physical amplitude scale for a full-relaxation lattice kinetic scheme. This allows for connecting quantum propagation to nonlinear feedback and assessing the separate demands of compression, velocity accuracy, and readout precision.

  2. The system can utilize a hybrid algorithm where the stored classical TT cores supply the next coefficients and prepare the next quantum input without replaying previous quantum evolution during subsequent readout. This enables a single-step construction that retains the actual nodewise coefficient sum, including its discrete velocity divergence.

  3. The AI system can perform error analysis by bounding local coefficient, operator, and core-fitting defects through Proposition 4.1, which relates these to distribution and macroscopic feedback errors under the assumption that the local coefficient and update defects propagate to distribution and macroscopic feedback errors.

Sources

Related papers