On Performance and Limitations of NISQ Hardware for Simulations of Quantum Wave Packet Dynamics

This paper presents a grid-based digital quantum simulation method for one-dimensional wave packet dynamics that reduces operator scaling to O(2^n) and demonstrates that while small-scale implementations on IBM and IonQ hardware qualitatively reproduce benchmarked dynamics, IonQ maintains higher accuracy than IBM as the qubit count increases.

The Gist

Imagine you are trying to predict how a drop of ink spreads in a glass of water, or how a ball bounces inside a box. In the world of physics, these movements are described by something called a "wave packet." To simulate this on a regular computer, you have to break the space into a grid of tiny dots. The problem is, as soon as you add more dimensions or make the grid finer, the number of dots explodes, and even the world's fastest supercomputers get stuck. This is known as the "curse of dimensionality."

This paper explores a different tool: Quantum Computers. Instead of using bits (0s and 1s) like a normal computer, quantum computers use "qubits." Because qubits can exist in many states at once, they can naturally represent these complex wave patterns without needing an impossible number of grid points.

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

1. The New "Recipe" for Simulation

The researchers wanted to simulate how a particle moves over time. They used a method called the Split-Operator Approach. Think of this like a dance routine that is split into two distinct moves:

• Move A (Kinetic Energy): This is how the particle moves on its own. The researchers used a mathematical trick called the Quantum Fourier Transform (QFT) to handle this. Imagine this as a special lens that instantly shifts your view from "where the particle is" to "how fast it is going," making the calculation much faster.
• Move B (Potential Energy): This is how the environment (like a wall or a hill) affects the particle. In the past, researchers had to build a custom circuit for every specific type of wall. In this paper, they developed a universal "Lego set" using simple building blocks called Pauli-Z gates. This allows them to plug in any shape of potential energy (flat, bumpy, or wavy) without redesigning the whole machine.

The Big Win: Usually, breaking down a complex problem into these Lego blocks gets exponentially harder as you add more qubits (like trying to build a skyscraper with Legos where the number of pieces doubles every time you add a floor). The authors' new method cuts that difficulty in half, making it much more manageable for current technology.

2. The Race: Who Can Dance Best?

To test their new recipe, the team ran simulations on two types of quantum hardware:

• IBM's Superconducting Processors: Think of these as high-speed race cars that are very fast but sensitive to bumps in the road (noise). They tested three different models: Torino, Miami, and Boston.
• IonQ's Trapped-Ion Device: Think of this as a precision gymnast. It moves slightly slower but is incredibly stable and accurate, with the ability to connect any part of its body to any other part (all-to-all connectivity).

They tested three scenarios:

1. The Free Runner: A particle moving on a flat surface (like a skater on ice).
2. The Tunneling Hiker: A particle trying to pass through a barrier it shouldn't be able to cross (quantum tunneling).
3. The Bouncing Ball: A particle trapped in a bowl, bouncing back and forth (a harmonic oscillator).

3. The Results: Small Steps vs. Big Leaps

The researchers tested these scenarios using 2, 3, 4, and 5 "qubits" (which corresponds to grids of 4, 8, 16, and 32 points).

• The Small Scale (2 & 3 Qubits): Both the IBM race cars and the IonQ gymnast performed well. They could all qualitatively reproduce the correct movement, though the newer IBM models (Boston and Miami) were slightly better than the older ones.
• The Medium Scale (4 Qubits): The gap started to widen. The IBM race cars began to wobble and lose their way, while the IonQ gymnast stayed steady and accurate.
• The Large Scale (5 Qubits): This is where the difference became dramatic.
  • IBM: The superconducting processors got so "noisy" that the wave packet (the ink drop) lost its shape entirely. It collapsed into a uniform blur, like a spilled cup of coffee that has been stirred too much. The simulation failed to show any meaningful physics.
  • IonQ: The trapped-ion device continued to track the simulation accurately, staying very close to the perfect "ideal" result.

The Bottom Line

The paper concludes that while quantum computers are promising for simulating how particles move, current hardware is still very fragile.

• Noise is the enemy: As the simulation gets more complex (more qubits), the errors in the hardware pile up quickly.
• Hardware matters: The type of quantum computer makes a huge difference. The IonQ trapped-ion device, with its superior stability and connectivity, handled the noise much better than the IBM superconducting chips.
• Design matters: The new method the authors developed (using the specific Pauli-Z gates and QFT) is more efficient than older methods, but even the best design hits a wall if the hardware is too noisy.

In short, the researchers successfully built a better "map" for quantum simulations, but they found that the "terrain" (the current hardware) is still too rough for long, complex journeys. Only the most stable machines (like IonQ) could complete the longer trips without getting lost.

Technical Summary

Technical Summary: On Performance and Limitations of NISQ Hardware for Simulations of Quantum Wave Packet Dynamics

Problem Statement
The accurate numerical simulation of quantum dynamical processes, such as molecular vibrations, scattering, and tunneling, is governed by the time-dependent Schrödinger equation. Classical approaches face the "curse of dimensionality," where the computational cost grows exponentially with the number of degrees of freedom, rendering long-time propagation intractable for systems beyond a few dimensions. While quantum computers offer a natural platform for simulating quantum systems due to their exponential state space scaling, the explicit real-time propagation of wave packets using digital quantum circuits remains underexplored compared to electronic structure problems. Furthermore, existing quantum simulation methods often rely on analytic potential energy functions or suffer from exponential scaling in operator decomposition (typically $\mathcal{O}(4^n)$ for $n$ qubits), limiting their applicability to near-term, noisy intermediate-scale quantum (NISQ) hardware.

Methodology
The authors propose a grid-based encoding of the wave function onto an $n$-qubit register, where $M = 2^n$ grid points represent the discretized spatial coordinate. The time evolution is implemented using the split-operator approach, which separates the kinetic ($\hat{T}$) and potential ($\hat{V}$) energy operators.

1. Kinetic Energy: The kinetic operator is applied using the Quantum Fourier Transform (QFT). This maps the wave function to the momentum basis where $\hat{T}$ is diagonal, allowing for efficient application with polynomial circuit depth scaling of $\mathcal{O}(n^2)$.
2. Potential Energy: Unlike previous work restricted to analytic potentials, this study employs a general representation of the potential energy operator as a linear combination of commuting Pauli-$Z$ strings. Because the potential is diagonal in the position basis, its decomposition requires only $2^n$ terms (involving only $I$ and $Z$ operators) rather than the full $\mathcal{O}(4^n)$ Pauli decomposition. This reduces the operator scaling to $\mathcal{O}(2^n)$ and allows for the direct encoding of arbitrary discretized potential energy landscapes.
3. Implementation: The global propagator is approximated via a first-order Trotter decomposition. Simulations were conducted on classical noise-free emulators (Qiskit SDK) and executed on NISQ hardware, specifically IBM superconducting processors (Torino, Boston, Miami) and the IonQ Forte trapped-ion device.

Key Contributions

• Algorithmic Efficiency: The paper demonstrates a method that reduces the scaling of the potential energy operator decomposition from $\mathcal{O}(4^n)$ to $\mathcal{O}(2^n)$ by exploiting the diagonal nature of the potential in the position basis.
• Generalizability: The approach enables the simulation of arbitrary discretized potentials, moving beyond systems limited to known analytic functional forms.
• Hardware Benchmarking: The study provides a comprehensive benchmark of wave packet dynamics across qubit counts of 2 to 5 (corresponding to 4 to 32 grid points) on multiple current hardware platforms.

Results
The methodology was tested on three representative one-dimensional problems: free particle propagation, tunneling through a barrier, and harmonic oscillator vibrations.

• Small-Scale Performance (2–3 Qubits): All tested platforms (IBM and IonQ) qualitatively reproduced the benchmark dynamics. For 2-qubit cases, newer IBM devices (Boston, Miami) showed near-quantitative agreement with the emulator.
• Scalability and Noise Sensitivity: As qubit counts increased to 4 and 5, performance diverged significantly between platforms.
  • IBM Superconducting Devices: Results showed rapid degradation with increasing qubit count. In 5-qubit simulations, the wave function on IBM hardware collapsed to a uniform distribution by the third time step, indicating a loss of wave-packet structure due to accumulated gate errors and circuit depth limitations.
  • IonQ Trapped-Ion Device: The IonQ Forte processor maintained significantly better agreement with the classical benchmark across all qubit counts, including the 5-qubit free-particle case. This performance is attributed to higher two-qubit gate fidelities and all-to-all connectivity inherent to trapped-ion architectures.
• Problem Specifics: In the tunneling and harmonic oscillator scenarios, IonQ consistently outperformed superconducting devices, particularly in the 5-qubit regime where IBM devices exhibited substantial deviations.

Significance and Claims
The paper concludes that while current NISQ hardware can qualitatively reproduce quantum dynamics for small system sizes, the fidelity of time-dependent simulations is highly sensitive to circuit depth and hardware noise. The results highlight that trapped-ion architectures currently offer superior performance for these specific dynamical simulations compared to superconducting processors. The authors assert that scalable implementations of quantum wave packet dynamics will require circuit designs that minimize depth and are robust against hardware noise, emphasizing that the proposed $\mathcal{O}(2^n)$ potential encoding is a necessary step toward utilizing arbitrary potentials on near-term devices.