arXiv:2609.39962
Quantum Algorithms for Computational Fluid Dynamics
Mario Guillaume Cecile, Nis-Luca van Hülst, Tomohiro Hashizume, Pia Siegl, Abhishek Setty, José Diogo da Costa Jesus, Paul Over, Sergio Bengoechea, Muhammad Umer, Spyros Tserkis, Eleftherios Mastorakis, Tristan Kraft, Francisco Cárdenas-López, Leonardo Scandurra, Thomas Rung, Felix Motzoi, Belda Yesil, Barbara Kraus, Martin Kiffner, Dimitris G. Angelakis, Eugene de Villiers, Dieter Jaksch
Abstract
We present a comprehensive review of quantum approaches for solving partial differential equations (PDEs) arising in computational fluid dynamics (CFD). We examine fully quantum approaches, including quantum linear system algorithms (QLSAs), ranging from the Harrow--Hassidim--Lloyd (HHL) algorithm to quantum singular value transformation (QSVT), Hamiltonian simulation, and quantum lattice Boltzmann methods (QLBMs), while emphasizing hybrid quantum--classical approaches, including quantum physics-informed neural networks (QPINNs) and amplitude-encoded variational PDE solvers. We focus on hardware-agnostic algorithms compatible with present noisy processors and emerging fault-tolerant architectures. For each framework, we analyze the mathematical formulation, algorithmic structure, and principal limitations. We also examine tensor-network (TN) representations, since CFD fields, differential operators, and geometrical information can often be encoded efficiently in low-rank form. The TN formalism bridges CFD discretizations and quantum states, operators, and circuits, enabling compact representations to be translated into tensor-programmable variational quantum algorithms (TP-VQAs). We further review benchmark problems, including Poisson, reaction, diffusion, and nonlinear model equations, and assess how well quantum algorithms capture key features of fluid dynamics. Our analysis highlights that potential quantum advantage is highly problem dependent and governed by condition number, representational complexity, state preparation, and measurement constraints. We outline capabilities, limitations, and challenges toward scalable quantum algorithms for CFD.
AI Summary
This review asks which quantum algorithms could support computational fluid dynamics (CFD), and under what assumptions compact quantum representations could yield an end-to-end advantage. It surveys quantum linear-system algorithms (HHL and QSVT), Hamiltonian-based evolution, quantum lattice Boltzmann methods, quantum physics-informed neural networks, and amplitude-encoded variational PDE solvers. Its organizing emphasis is tensor networks as an intermediate representation connecting CFD discretizations to quantum circuits. It describes matrix-product-state preparation, matrix-product-operator block encoding, and tensor-programmable variational algorithms, including nonlinear terms and boundary treatments. The analysis synthesizes published complexity bounds, numerical benchmarks, hardware demonstrations, and resource estimates rather than presenting a new solver or experimental result. Several quantitative examples frame the limitations. Amplitude encoding uses $\lceil\log_2N\rceil$ qubits, but unstructured input loading and complete classical readout can each cost $\mathcal{O}(N)$. Standard elliptic discretizations have condition number $\kappa=\mathcal{O}(N^{2/d})$, potentially offsetting linear-solver speedups. Reviewed tensor-programmable benchmarks report operator success probabilities around $0.35$–$0.50$ for selected linear stencils; nonlinear diagonal encoding improves success probability but requires classical reconstruction and compilation. A cited specialized fault-tolerant Navier–Stokes estimate projects $8.71\times10^6$ physical qubits and 42.6 days for approximately $2^{80}$ cells, whereas a current variational prototype estimates a runtime $1.3\times10^{12}$ times its classical baseline. The review identifies unresolved costs in state preparation, measurement, optimization, nonlinear processing, and error correction. Low-rank compressibility is problem dependent; arbitrary unstructured meshes, realistic turbulent evolution, and an early-fault-tolerant crossover for variational CFD remain open.
Error found: The review's expectation of 'at least polynomial advantages' from transferring tensor-network representations to quantum circuits is not established generally by its resource comparisons, which include matching O(χ³) classical overhead for diagonal-encoding TP-VQA and leading classical Hadamard-product methods. The review incorrectly defines the condition number of a general diagonalizable operator through eigenvalue magnitudes rather than singular values, which can substantially underestimate conditioning for nonnormal CFD matrices. The review incorrectly states that an MPS has Schmidt rank at most χ and entropy at most log χ across any bipartition; these bounds apply to single-bond cuts, while multiple-bond cuts can exceed them.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 41 pages, 14 figures, 2 tables, Review article