A Quantum-Inspired Algorithm for Wave Simulation Using Tensor Networks
Kevin Lively, Vittorio Pagni, Gonzalo Camacho · 2025
Read A Quantum-Inspired Algorithm for Wave Simulation Using Tensor Networks by Kevin Lively, Vittorio Pagni, Gonzalo Camacho (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-11181.txt.
Introduction
Kevin Lively, Vittorio Pagni, and Gonzalo Camacho presented a 2025 method for wave simulation that borrows ideas from quantum computing. Their target is the isotropic wave equation in one, two, or three dimensions. Direct grids fine enough for sharp waves are costly to run.
They rewrite the wave equation, using a link to the Dirac equation, so space and time steps become first order. A quantum-style circuit then drives the motion. Tensor networks store the state in compressed form, letting them reach a huge grid count on a laptop. A method for loading smooth start shapes into the state is also given. The authors suggest the same path for other wave systems, such as light.
Catalog record
| CREATOR | Kevin Lively, Vittorio Pagni, Gonzalo Camacho |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-11181
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Quantum-Inspired Algorithm for Wave Simulation Using Tensor Networks Kevin Lively, Vittorio Pagni, Gonzalo Camacho Abstract We present an efficient classical algorithm based on the construction of a unitary quantum circuit for simulating the IWE ( IWE ) in one, two, or three dimensions. Using an analogy with the massless Dirac equation, second order time and space derivatives in the IWE are reduced to first order, resulting in a Schrödinger equation of motion. Exact diagonalization of the unitary circuit in combination with Tensor Networks allows simulation of the wave equation with a resolution of 10 13 10^{13} grid points on a laptop. A method for encoding arbitrary analytical functions into diagonal MPOs ( MPOs ) is employed to prepare and evolve a MPS ( MPS ) encoding the solution. Since the method relies on the Quantum Fourier Transform, which has been shown to generate small entanglement when applied to arbitrary MPS s, simulating the evolution of initial conditions with sufficiently low bond dimensions to high accuracy becomes highly efficient, up to the cost of Trotterized propagation and sampling of the wavefunction. We conclude by discussing possible extensions of the approach for carrying out Tensor Network simulations of other partial differential equations such as Maxwell’s equations. I Introduction Due to their theoretical capacity to manipulate data of size 2^{n} within the Hilbert space of n qubits, QCs ( QCs ) have been advanced as promising platforms for large scale simulation of Partial Differential Equations (PDEs). However, the physics of quantum devices is ultimately governed by a single PDE ( PDE ), namely the Schrödinger equation. Thus, mapping other PDE s into forms which can be simulated by the evolution of a quantum mechanical wavefunction is the first challenge in designing any Quantum Algorithm for this task [ 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 ] . Subsequently, dozens of approaches to this problem have been put forward in the literature spanning both hybrid quantum-classical and fully-quantum algorithms. Many of these approaches are either based on linearizing the PDE over time and space derivatives along with a Carleman approach for sufficiently weak non-linear terms, then either inverting the resulting system of equations using the Harrow-Hassidim-Lloyed algorithm [ 19 , 20 , 21 , 1 , 22 , 5 , 23 ] , constructing a variational quantum algorithm [ 24 , 25 , 26 , 27 ] , or using QCs to (theoretically) speed up bottleneck integrations in integro-differential approaches [ 4 , 28 , 29 , 30 ] . A crucial aspect to this field of research is that while many of these algorithms appear to be impractical for near-term hardware, some of them have inspired highly efficient classical methods [ 31 , 24 , 32 , 33 , 26 , 34 , 35 ] . Given the centrality of PDE simulation across virtually every field of science, technology and engineering, fresh perspectives on this problem can have a substantial and immediate impact, as has been recently seen in novel computational fluid dynamics methods [ 33 , 34 , 35 , 36 ] . The wave equation is a PDE , whose simulation spans critical application areas in fluid dynamics, electromagnetism, seismology and medicine [ 37 , 38 , 39 , 40 ] . Numerical approaches for modeling wave propagation are generally grouped into real-space finite differencing methods or spectral approaches. Due to second-order derivatives in space and time, real space methods typically require high order stencils on a fine grid and tracking of multiple time steps simultaneously to be stable and accurate [ 41 ] . In contrast, spectral methods require conversion between momentum-space and real-space via the Fast Fourier Transform (FFT), or analogous polynomial decompositions, which for the usually large grid sizes involved in three-dimensional (3D) simulation can be difficult to parallelize, due to requiring an all-to-all operation [ 42 ] . In the limit of a uniform medium, the problem reduces to the Isotropic Wave Equation (IWE), an exactly solvable PDE which is often used as an introductory example to build up more complicated PDE simulation techniques. Thus this serves as a natural starting point when approaching the problem of simulating PDE s employing quantum algorithms. In a recent work by Wright et. al. a quantized representation of the IWE in one dimension (1D) was developed [ 12 ] , followed by implementing a concrete unitary circuit based on the abstract approach of Costa, Jordan and Ostrander [ 2 ] . By relying on the Quantum Fourier Transform (QFT) to map to reciprocal space, they found a unitary circuit capable of simulating the 1D IWE in the small wavenumber limit to arbitrary times in a fixed depth and demonstrated it on Quantinuum’s Ion Trap hardware. Two notable aspects of this work stand out for our purposes. One is that the central reliance on the QFT provides an appealing opportunity for quantum-inspired algorithms. Recently it was rigorously shown that the core circuit in the QFT, i.e. the combination of Hadamard and controlled phase gates without the bit reversal operation, can be very efficiently implemented using Tensor Networks [ 43 ] . Specifically using the Matrix Product State (MPS) or Tensor Train structure [ 44 , 45 , 46 , 47 , 48 ] , the resultant Matrix Product Operator (MPO) representation of the core of the QFT can be constructed to a high degree of accuracy with a fixed cost in the bond dimension, allowing it to outperform the FFT on data which can be efficiently represented as MPSs. The second notable aspect is that the method of Costa et. al relies on constructing a Hamiltonian operator whose square, i.e the second time derivative of the wavefunction, returns the Laplacian. This suggests a direct analogy with the Dirac equation in Quantum Mechanics, which we explicitly exploit in order to develop the associated quantum representation of the original problem. In this work, we combine these two insights to develop an MPS based algorithm capable of simulating the IAW in one, two or three dimensions. The paper is structured as follows. In Section II we lay out our derivation based on the Dirac Equation, whose square also returns the IAW in the massless limit. In section III, using a unitary circuit representation of the equation of motion in reciprocal space, we derive an exact solution and benchmark the results against direct integration of the real space equations of motion. We also describe how to approximately propagate the wave equation via Trotterization. In section IV we utilize an encoding method based on the sum of Pauli operators containing only \hat{I} and \hat{Z} operators in order to prepare simple analytical functions onto the diagonals of MPOs without having to actually hold them in memory, and use this to prepare such functions as MPSs. As with other methods for representing functions in a tensor network format [ 49 , 50 , 51 , 52 , 53 , 54 ] this allows us to push the effective grid size we can simulate well beyond classical storage limits, which