Hamiltonian Simulation for Advection-Diffusion Equation with arbitrary transport field
Niladri Gomes, Gautam Sharma, Jay Pathak · 2025
Read Hamiltonian Simulation for Advection-Diffusion Equation with arbitrary transport field by Niladri Gomes, Gautam Sharma, Jay Pathak (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-2508-16728.txt.
Introduction
Niladri Gomes, Gautam Sharma, and Jay Pathak solve the advection diffusion equation on quantum style hardware in 2025. The equation models how a substance spreads while being carried by flow. Their method uses a trick that recasts the problem as a Hamiltonian system. It handles transport fields that vary across space, in two and three dimensions.
Stability comes from pairing an upwind scheme for flow with a centered scheme for spread. Benchmarks cover swirling, shearing, and spreading flows. The team also ran the two dimensional case on real quantum hardware with sixteen qubits. Results held up well. The work shows the approach is sound and robust for long term quantum machines.
Catalog record
| CREATOR | Niladri Gomes, Gautam Sharma, Jay Pathak |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 293 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-16728
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Hamiltonian Simulation for Advection-Diffusion Equation with arbitrary transport field Niladri Gomes, Gautam Sharma, Jay Pathak Abstract We present a novel approach to solve the advection-diffusion equation under arbitrary transporting fields using a quantum-inspired ‘Schrödingerisation’ technique for Hamiltonian simulation. Although numerous methods exist for solving partial differential equations (PDEs), Hamiltonian simulation remains a relatively underexplored yet promising direction—particularly in the context of long-term, fault-tolerant quantum computing. Building on this potential, our quantum algorithm is designed to accommodate non-trivial, spatially varying transport fields and is applicable to both 2D and 3D advection-diffusion problems. To ensure numerical stability and accuracy, the algorithm combines an upwinding discretization scheme for the advective component and the central differencing for diffusion, adapted for quantum implementation through a tailored mix of approximation and optimization techniques. We demonstrate the algorithm’s effectiveness on benchmark scenarios involving coupled rotational, shear, and diffusive transport in two and three dimensions. Additionally, we implement the 2D advection-diffusion equation using 16 qubits on IBM Quantum hardware, validating our method and highlighting its practical applicability and robustness. I Introduction Achieving quantum speedup for the numerical solution of ordinary and partial differential equations (ODEs/PDEs), beyond those specific to the Schrödinger equation, holds considerable promise for advancing a wide range of scientific and engineering disciplines [ 1 , 2 , 3 ] . PDEs play a central role in modeling the dynamic behavior of physical systems and arise in numerous applications, including reservoir simulation, climate and ocean circulation, and wave propagation. In particular, computational fluid dynamics (CFD) relies fundamentally on the numerical solutions of PDEs that describe fluid flow phenomena [ 4 ] . While high-performance computing (HPC) has enabled significant strides in solving such complex models, the computational cost remains a persistent challenge, especially in problems involving large-scale domains, high dimensionality, or multiscale dynamics. The scalar advection-diffusion equation provides a canonical building block in conventional numerical methods development for computational fluid dynamics [ 5 ] . It encapsulates two fundamental mechanisms of transport, advection, associated with bulk motion, and diffusion, representing molecular or turbulent mixing [ 6 ] . Despite not completely covering all complexities such as pressure-velocity coupling and nonlinearities expressible with the complete Navier-Stökes set of governing equations, the scalar advection-diffusion equation presents a model analytical framework to test and benchmark early algorithms. It captures both the hyperbolic and parabolic behaviours in a single framework, providing for a robust and amenable prototype problem class. Quantum algorithms for time-dependent problems governed by PDEs can generally be categorized into two categories: a) the variational quantum simulation methods [ 7 , 8 ] and b) Hamiltonian simulation techniques [ 9 ] . The latter represents a more promising direction for long-term, fault-tolerant quantum computing [ 9 , 10 ] . Hamiltonian simulation refers to the construction of a quantum circuit that models the time evolution of a quantum system, typically represented by the unitary operator e^{-i\mathbf{H}t} , where \mathbf{H} denotes the system’s Hamiltonian and t is the time increment. Studies have explored the use of Hamiltonian simulation by reformulating the governing equations of classical systems into an equivalent Schrödinger equation [ 11 , 12 , 13 ] . Owing to inherent algorithmic and hardware constraints, computational fluid dynamics (CFD) presents a compelling opportunity for exploring quantum acceleration in classical engineering applications and remains an active area of investigation [ 2 , 14 , 15 , 16 , 17 , 18 ] . Steijl et al. [ 19 ] demonstrated that a Poisson solver used in a vortex-in-cell approach could be reformulated as a hybrid quantum algorithm through the application of the quantum Fourier transform. Gaitan [ 20 ] proposed a quantum algorithm for solving the steady-state, inviscid, one-dimensional, compressible Navier–Stokes equations (NSE) in a converging-diverging nozzle, leveraging techniques from quantum ODE solvers [ 21 ] . Extending this work, Oz et al. [ 22 ] implemented a quantum algorithm to address the nonlinear Burgers equation. In parallel, Suau et al. [ 23 ] showcased a quantum approach for solving the wave equation based on the Hamiltonian simulation framework introduced by Costa et al. [ 11 ] . A recent study by Lee et al. [ 24 ] proposed a multiple-circuit approach to reduce quantum resource requirements in the quantum lattice Boltzmann method, offering a practical direction for quantum computational fluid dynamics. Brearley et al. used an explicit time marching strategy for solving the advection equation by embedding numerical integrators into a series of Hamiltonian simulations [ 25 ] . While these findings indicate that quantum algorithms have the potential to achieve speedup in simulating classical systems, they either typically assume oracle access to the Hamiltonian, or simulate idealistic models [ 26 ] or may not apply to a general class of PDEs [ 25 ] . With a goal of building a long term general purpose PDE solver, here we adopt the Hamiltonian simulation approach with Schrödingerisation. In order to move forward with realistic calculations, we consider vortex-dominated flows, characterized by circular or nearly circular streamlines and are governed by the dynamics of vorticity, a fundamental quantity in fluid mechanics [ 4 , 27 ] . Vorticity, defined as the curl of the spatially varying transport field, quantifies the local rotational behavior of fluid elements and provides insight into the structure and evolution of complex flow fields. In particular, we will consider a transport field with non-zero vorticity, where the direction of the vorticity vector denotes the axis of rotation, while its magnitude reflects the intensity of local fluid spin. Vorticity plays a critical role in the analysis and modeling of key fluid dynamic phenomena, including boundary layers, shear layers, wakes, and mixing regions [ 27 ] . Our key contributions in this work can be summarized as follows: • Algorithmic scaling upto 30 qubits of Hamiltonian simulation of non-trivial transport fields : We present a new algorithm for solving the 2D/3D advection-diffusion equation with spatially varying transport fields using upwinding discretization scheme for the advective component and central difference scheme for the diffusive component that scales to 30 qubits on emulator. • First hardware implementation up to 16 qubits : Using the same discretization scheme, we implemented evolution of advection-diffusion equation in 2D for a (256x256 grid) with a spatially varying transport field for 40 time steps. • Efficient circuits for upwinding scheme : We introduce practical approximations to the upwinding scheme for the advective component that eliminates the need for ancilla qubits, thereby removing mid-circuit measurements and reducing the overall circuit depth by at least 30%. II Theory and Methods In this section, we will describe the form of the advection-diffusion equation we want to solve and the Schrödingerisation technique applied in order to convert it into a Schrödinger-like equation. II-A Advection-Diffusion Equation and Schrödingerisation A d -dimensional advection-diffusion equation is written as, \displaystyle\partialderivative{u(t,\mathbf{x})}{t} \displaystyle=-v\cdot\gradient{u(t,\mathbf{x})}+D\gradient^{2}{u(t,\mathbf{x})} \displaystyle\equiv-\sum_{\alpha=1}^{d}v_{\alpha}\partial_{\alpha}u(t,\mathbf{x})+D\gradient^{2}{u(t,\mathbf{x})}, (1) where u is the scalar field moving with transport v in a medium with diffusivity D\geq 0 . \gradient and \gradient^{2} are the spatial gradient and diffusion operators respectively. Applying the warped phase transformation \mathbf{v}(t,p)=e^{-p}u(t,\mathbf{x}) , and subsequently applying a Fourier transform ( \mathcal{F} ) over p , ( \hat{\mathbf{v}(t,\eta)}=\mathcal{F}(\mathbf{v}(t,p)) ), a spatially discretized version of Eq. ( 1 ) can be written as \displaystyle\partialderivative{\hat{\mathbf{v}}(t,\mathbf{x},\eta)}{t}=\mathbf{H_{D}}\hat{\mathbf{v}}(t,\mathbf{x},\eta)+\mathbf{H}\hat{\mathbf{v}}(t,\mathbf{x},\eta), (2) where we refer