Quantum Lattice Boltzmann Solutions for Transport under 3D Spatially Varying Advection on Trapped Ion Hardware

This paper presents the first demonstration of Quantum Lattice Boltzmann Method simulations for transport under non-uniform 3D advection on trapped-ion hardware, introducing novel wall boundary conditions, identifying density readout as a key bottleneck, and proposing MPS shadow tomography as a scalable solution for complex fluid dynamics problems.

The Gist

Imagine you are trying to predict how a drop of ink spreads through a swirling river. In the real world, this is a messy, complex problem involving fluid dynamics. Scientists usually solve this using supercomputers that break the river into a giant 3D grid of tiny boxes, calculating how the ink moves from one box to the next. This is called the Lattice Boltzmann Method (LBM).

This paper describes a new attempt to do this calculation using a Quantum Computer instead of a classical one. Specifically, the researchers used a special type of quantum computer that traps individual atoms (ions) in a vacuum to act as the "processors."

Here is a breakdown of what they did, using simple analogies:

1. The Goal: Simulating a Swirling River in 3D

The researchers wanted to simulate a specific type of fluid flow: a 3D swirl where the speed and direction of the water change depending on where you are in the grid.

• The Challenge: Previous quantum experiments could only handle simple, flat (2D) flows or flows where the water moved at a constant speed everywhere. Real rivers are 3D and twisty.
• The Achievement: They successfully ran a simulation of this complex 3D swirling flow on actual quantum hardware (IonQ's trapped-ion systems). They managed to track the "ink" (fluid density) as it moved and diffused over time.

2. The "Readout" Problem: Taking a Snapshot of a Ghost

In a quantum computer, the information exists as a "superposition" (a cloud of possibilities). To see the result, you have to "measure" it, which collapses the cloud into a single picture.

• The Bottleneck: The researchers found that trying to take a perfect picture of the fluid's position after every step was like trying to photograph a ghost with a slow camera. The "noise" from the hardware and the sheer number of measurements needed made it hard to get a clear picture, especially as the grid got bigger.
• The Solution (The "Shadow" Trick): To fix this, they invented a new way to read the data. Instead of trying to take one perfect photo, they took many "shadow" snapshots from different angles (randomized measurements).
  • Analogy: Imagine trying to figure out the shape of a complex sculpture in a dark room. Instead of turning on a blinding light that ruins the view, you shine a flashlight from many different random angles and use a computer to piece together the shadows to reconstruct the 3D shape.
  • Result: This "Shadow Tomography" method allowed them to reconstruct the fluid's shape much more accurately and with fewer measurements than before.

3. The "Reload" Problem: Keeping the Story Going

To simulate time passing, the computer needs to finish one step, read the result, and then "reload" that result to start the next step.

• The Innovation: They used a mathematical compression technique called MPS (Matrix Product States). Think of this like compressing a high-definition video into a smaller file size without losing the important details.
• Why it matters: Because the fluid density in their simulation is "smooth" (it doesn't have jagged, random noise), it can be compressed efficiently. This allowed them to read the data, compress it, and reload it back into the quantum computer to continue the simulation for many more steps than was previously possible.

4. Adding Walls and Obstacles

Real rivers have banks, rocks, and pipes. The researchers also showed how to program the quantum computer to respect "walls."

• The Method: They created a digital "oracle" (a rulebook) that tells the quantum computer: "If the fluid hits this coordinate, stop it from moving forward."
• The Result: They successfully simulated fluid flowing around a solid cube suspended inside a pipe, ensuring the fluid didn't magically pass through the solid object.

5. The Hardware: Trapped Ions

They ran these experiments on IonQ's quantum computers.

• The Setup: These computers use individual Barium or Ytterbium atoms held in place by magnetic fields (like a cage).
• The Performance: Despite the hardware being "noisy" (prone to errors), their method was surprisingly robust. Even though the computer made mistakes, the way they structured the math meant that many errors were naturally filtered out or didn't ruin the final picture. They achieved high accuracy (over 88% fidelity) even after six steps of simulation.

Summary

In short, this paper is a proof-of-concept that says: "We can use current quantum computers to simulate complex, 3D fluid flows that change over time."

They didn't just run a simple test; they solved three major headaches that usually stop these simulations:

1. Complexity: They handled 3D, twisting flows (not just flat ones).
2. Measurement: They found a smarter way to "read" the quantum data using "shadows" so they didn't need millions of measurements.
3. Continuity: They figured out how to compress the data and reload it to keep the simulation running for longer.

This is a stepping stone toward eventually using quantum computers to help engineers design better airplanes, cars, or weather models, but for now, it is a successful demonstration of the method working on real hardware.

Technical Summary

1. Problem Statement

The paper addresses the challenge of simulating Computational Fluid Dynamics (CFD) problems, specifically advection-diffusion, on near-term quantum hardware.

• Context: Classical CFD solvers are computationally expensive. Quantum algorithms, particularly the Quantum Lattice Boltzmann Method (QLBM), offer a promising alternative by encoding fluid dynamics into mesoscopic particle distribution functions on discrete lattices.
• Specific Challenges:
  • Complexity: Previous QLBM demonstrations were limited to 2D simulations with constant velocity fields. Real-world CFD requires 3D spatially varying velocity fields (e.g., swirls) and complex boundary conditions.
  • Readout Bottleneck: Extracting the fluid density distribution from a quantum state is a major bottleneck. Full state tomography is exponentially expensive, and direct histogram reconstruction suffers from statistical noise (shot noise) and post-selection losses, especially as system size and time steps increase.
  • Hardware Limitations: Current Noisy Intermediate-Scale Quantum (NISQ) devices struggle with deep circuits and high gate counts required for multi-step simulations.

2. Methodology

The authors propose a comprehensive framework to simulate 3D transport on trapped-ion hardware, focusing on efficient state preparation, evolution, and readout.

A. Algorithmic Framework (QLBM)

• Model: Uses the D3Q7 (3D, 7 discrete velocities) and D2Q5 models with a relaxation time $\tau=1$ under the Bhatnagar-Gross-Krook (BGK) approximation.
• Operators:
  • Streaming ($U_S$): Propagates particle distributions along discrete directions.
  • Collision ($U_P, U_Q$): Relaxation toward equilibrium. The authors implement these using Linear Combination of Unitaries (LCU) and PREP/UNPREP operators.
  • Velocity Fields: They introduce a method to implement non-uniform, divergence-free velocity fields (specifically a 3D swirl) using quantum oracles and controlled rotations.
  • Boundary Conditions: A novel method is introduced to handle wall boundaries. Instead of modifying the velocity field globally, they use a "wall register" and controlled RBS (Reconfigurable Beam Splitter) gates to zero out streaming amplitudes into solid walls while conserving probability mass.

B. Readout and State Reconstruction Strategies

To overcome the readout bottleneck, the paper evaluates and compares three classical post-processing techniques:

1. Kernel Density Estimation (KDE): Smoothes statistical fluctuations in measured histograms.
2. MPS Smoothing: Fits a Matrix Product State (MPS) with a fixed bond dimension to the histogram data, leveraging the spatial smoothness of fluid density.
3. MPS Shadow Tomography: A novel approach using Classical Shadows. Instead of measuring in the computational basis only, the circuit applies random single-qubit rotations (SU(2)) before measurement. The resulting data is used to train an MPS model directly via stochastic gradient descent (minimizing Hellinger loss).

C. Hardware Implementation

• Platform: Experiments were conducted on IonQ's trapped-ion systems, including:
  • Forte: Ytterbium ($Yb^+$) based systems.
  • Tempo-line Prototype: A 64-qubit Barium ($Ba^+$) development system with a long-chain, steered-beam configuration.
• Circuit Depth: The simulations involve deep circuits (~300 two-qubit gates) with error detection gadgets (ancilla qubits) and post-selection.

3. Key Contributions

1. First 3D Spatially Varying QLBM on Hardware: The team successfully demonstrated advection-diffusion under a 3D swirl velocity field on a trapped-ion quantum computer, moving beyond previous 2D constant-field limitations.
2. Efficient Readout & Reloading: They demonstrated that KDE-assisted MPS reconstruction enables accurate iterative readout and state reloading, a critical requirement for multi-step time evolution on NISQ devices.
3. Scalable Shadow-MPS Tomography: They introduced a Classical Shadow-based MPS tomography protocol. This method outperforms direct histogram reconstruction, particularly for larger system sizes and lower shot counts, by aggregating information from randomized bases to suppress noise.
4. Generalized Boundary Conditions: The paper presents a scalable simulation framework for implementing wall boundaries and general non-trivial velocity fields without requiring a full re-derivation of the collision operator, improving efficiency over prior methods.
5. Optimized Classical Pre-processing: They developed an Interpolated Fast Walsh-Hadamard Transform (FWHT) algorithm (Algorithm 2) to reduce the classical complexity of preparing velocity field angles from $O(N \log N)$ to $O(K^3 \log K \log(N/K))$ for smooth fields.

4. Experimental Results

• Simulation Scope:
  • Grid Size: $8 \times 8 \times 8$ (Barium system) and $16 \times 16 \times 16$ (Forte/Simulator).
  • Time Steps: Up to 10 steps.
  • Shots: 20,000 to 50,000 shots per time step.
• Performance Metrics:
  • Fidelity: The simulation achieved a fidelity ($|\langle \phi | \psi \rangle|^2$) exceeding 88% after six time steps on the Barium system.
  • Noise Resilience: Despite hardware noise, the one-hot encoding and post-selection structure naturally filtered out certain error types (bit-flips in the direction register), maintaining high fidelity (~97% after one step).
  • Readout Comparison:
    • Shadow-MPS vs. Direct: On a $16^3$ grid with 20,000 shots, Shadow-MPS maintained a fidelity of 0.83 at $t=10$, whereas direct tomography with MPS smoothing dropped to 0.57.
    • Low-Shot Regime: With only 1,000 shots, Shadow-MPS achieved 0.75 fidelity at $t=10$, while the direct approach degraded to 0.1.
  • Scaling: The Shadow-MPS method showed superior robustness as grid size increased to $32^3$, whereas direct reconstruction errors compounded rapidly.

5. Significance and Outlook

• Practicality for NISQ: This work proves that QLBM is viable for complex, realistic fluid dynamics problems on current hardware, provided that efficient readout strategies (like Shadow Tomography) are employed to mitigate noise and shot limitations.
• Scalability: The introduction of Shadow-MPS tomography provides a pathway to scale QLBM to larger grids where standard readout methods fail due to the "curse of dimensionality" in measurement requirements.
• Industrial Relevance: By solving 3D advection-diffusion with spatially varying fields and wall boundaries, the work bridges the gap between theoretical quantum algorithms and practical engineering applications in CFD.
• Future Directions: The authors plan to extend this to more complex geometries, fully treat boundary conditions, and develop cohesive hybrid quantum-classical workflows to integrate these solvers into industrial-scale physics software.

In summary, this paper represents a significant leap in quantum fluid dynamics, moving from toy models to realistic 3D simulations on hardware, while solving the critical "readout bottleneck" through novel classical shadow techniques.