Estimation of deuteron binding energy with renormalization group-based effective interactions using the variational quantum eigensolver

This paper demonstrates the calculation of the deuteron binding energy on a quantum simulator using the variational quantum eigensolver with renormalization group-based effective interactions, showing that decreasing the RG parameter $\lambda$ reduces the required qubit count while enabling noise-mitigated results that closely match experimental values.

The Gist

The Big Picture: Solving a Nuclear Puzzle on a New Computer

Imagine you are trying to solve a very complex jigsaw puzzle. The picture on the box is a deuteron, which is the simplest "nucleus" in the universe, made of just two particles (a proton and a neutron) stuck together.

For a long time, scientists have used powerful classical supercomputers to figure out exactly how tightly these two particles are holding hands. This "tightness" is called the binding energy. If you know this number, you understand the fundamental glue of the universe.

However, these puzzles are incredibly hard. The pieces (particles) interact in messy, complicated ways, especially when they get very close to each other.

The New Twist:
This paper describes an experiment where the researchers tried to solve this specific puzzle using a quantum computer (or more accurately, a simulator that acts like one). They wanted to see if these new machines could handle nuclear physics problems better than old ones, and how to make the job easier.

The Problem: Too Many Pieces, Too Much Noise

Think of the classical way of solving this puzzle as trying to fit the pieces into a giant, rigid box.

1. The Box Size: To get an accurate answer, you need a huge box with millions of tiny slots (mathematical states) to represent where the particles could be. This requires a massive amount of computing power.
2. The Noise: Real quantum computers are like trying to solve a puzzle while someone is shaking the table and blowing wind on the pieces. The machines are "noisy," meaning they make mistakes easily.

The Solution: Smoothing the Rough Edges (Renormalization)

The researchers used a clever trick called Renormalization Group (RG) evolution.

The Analogy:
Imagine the interaction between the proton and neutron is like a very rough, jagged rock. If you try to fit this jagged rock into a smooth box, it's a nightmare. You need a huge box to accommodate all the jagged edges.

The researchers used a mathematical "sander" (the RG method) to smooth out that jagged rock. They didn't change the weight of the rock (the physics remains the same), but they made the surface smooth.

• Before sanding: You needed a massive box (many qubits) to fit the jagged rock.
• After sanding: The rock is smooth. It fits into a much smaller box.

The Result:
By using this "sanded" version of the physics, they found that they needed far fewer qubits (the basic units of a quantum computer) to get an accurate answer. As they smoothed the interaction more (lowering a parameter called $\lambda$), the puzzle became easier to solve, requiring fewer resources.

The Experiment: Testing in the Real World

The team used a tool called VQE (Variational Quantum Eigensolver). Think of VQE as a smart robot that tries different ways to arrange the puzzle pieces, checks how well they fit, and then tweaks the arrangement to get closer to the perfect solution.

They ran this experiment in two ways:

1. Perfect World (Noise-free): Using a simulator that acts like a perfect quantum computer.
2. Real World (Noisy): Using a simulator that mimics the actual, imperfect IBM quantum hardware (specifically the "Brisbane" machine).

The "Zero-Noise" Magic Trick:
Since the real machines make mistakes, the researchers used a technique called Zero-Noise Extrapolation.

• The Analogy: Imagine you are trying to measure the height of a building, but your ruler is slightly bent. You measure the building three times: once with the ruler bent a little, once bent a lot, and once bent even more. By looking at the pattern of your errors, you can mathematically guess what the height would be if the ruler were perfectly straight.
• The Outcome: Even with the "bent ruler" (noise), they were able to mathematically predict the correct answer. Their final result was within 1% of the actual experimental value found in nature.

The Hidden Discovery: Entanglement

The paper also looked at entanglement. In quantum physics, this is like a magical connection where two particles know what the other is doing instantly, no matter how far apart they are.

The researchers analyzed how "connected" the different parts of their puzzle were. They found that as they used their "sander" (RG method) to smooth the interaction, the particles became less entangled with the high-energy, complex parts of the system.

• Why this matters: Less entanglement means the quantum computer doesn't have to work as hard to keep track of the connections. It's like moving from a chaotic, noisy party where everyone is shouting to a quiet library where everyone is whispering. The quieter the room, the easier it is to have a conversation (or in this case, a calculation).

Summary of Findings

1. Smoothing helps: Using renormalized (smoothed) interactions makes nuclear physics problems much easier for quantum computers to solve.
2. Fewer resources needed: The smoother the interaction, the fewer qubits are required to get an accurate answer.
3. Noise is manageable: Even with the errors inherent in current quantum hardware, they could use mathematical tricks to get a result that matches real-world experiments to within 1%.
4. Proof of concept: This is a successful first step in using quantum computers to solve real, complex nuclear structure problems using realistic physics models, rather than just simplified toy models.

In short, the researchers showed that by "smoothing out" the physics first, they could teach a noisy, early-stage quantum computer to solve a difficult nuclear puzzle with high accuracy.

Technical Summary

Technical Summary: Estimation of Deuteron Binding Energy with RG-Based Effective Interactions Using VQE

Problem Statement
Ab-initio descriptions of nuclear structure face significant conceptual and computational challenges due to the strong correlations and complex interactions within few-body systems. Traditional phenomenological interactions often possess strong short-range repulsion, necessitating numerically expensive resummations. While effective field theories (EFTs) and renormalization group (RG) methods have improved classical perturbative calculations by generating effective interactions with better convergence properties, the application of these methods to quantum computing platforms remains an active area of investigation. Specifically, there is a need to understand how the numerical advantages of RG-evolved interactions translate to the quantum context, particularly regarding qubit requirements and error mitigation in the Noisy Intermediate-Scale Quantum (NISQ) era.

Methodology
The authors investigate the binding energy (BE) of the deuteron using the Variational Quantum Eigensolver (VQE) algorithm. The study employs two realistic two-body interactions: the chiral $N^4LO$ interaction and the Argonne V18 (AV18) potential. These interactions are evolved to low momenta using the Similarity Renormalization Group (SRG) approach to yield effective interactions ($V_{SRG}$) parameterized by a cutoff scale $\lambda$.

The deuteron Hamiltonian is formulated in a truncated 3D isotropic harmonic oscillator (HO) basis, characterized by the principal quantum number $n$ (truncated at $n_{max}$) and orbital angular momentum $l$ (taking values 0 and 2 due to tensor forces). The basis states are mapped to qubits using the Jordan-Wigner transformation, where each oscillator mode corresponds to a qubit occupation number. The VQE ansatz is constructed using a Unitary Coupled-Cluster (UCC) approach with single excitations, acting on a Hartree-Fock reference state where the lowest $0s$ mode is occupied.

Simulations were performed using the Qiskit-Aer simulator in two configurations:

1. Noise-free: Utilizing the statevector simulator and probabilistic simulations with $\sim 10^6$ shots to assess sampling errors.
2. Noisy: Incorporating noise models derived from actual IBM quantum hardware (specifically ibm_brisbane and ibm_kyiv).

To mitigate hardware noise, the authors employed Zero-Noise Extrapolation (ZNE) using global circuit folding and polynomial extrapolation (up to cubic degree). The study systematically varies the SRG cutoff $\lambda$, the oscillator frequency $\hbar\omega$, and the basis size $N$ ($N = n_{max} + 1$) to analyze convergence and entanglement.

Key Contributions and Results

• Qubit Efficiency via RG Evolution: The primary finding is that the number of HO basis states (and consequently qubits) required to compute the deuteron BE within 1% of the experimental value decreases as the RG parameter $\lambda$ decreases. For bare interactions, convergence requires larger basis sizes (e.g., $N \sim 6$), whereas for evolved interactions with small $\lambda$ (e.g., $\lambda \sim 0.1$ fm$^{-1}$), convergence is achieved even with $N=1$ or $N=2$. This demonstrates that RG-evolved interactions significantly reduce the qubit resource requirements for quantum simulations.
• Noise Mitigation and Accuracy: The study successfully extrapolated noisy simulation results to the zero-noise limit. The cubic polynomial extrapolation proved effective. The final results, obtained via ZNE, agree with exact diagonalization results and noiseless simulations. The calculated BE matches experimental values to within 1%, comparable to previous results using pionless EFT.
• Mode Entanglement Analysis: The authors analyzed the bipartite mode entanglement (quantified by concurrence) between oscillator modes as a function of $\lambda$. For $N \ge 3$, the average concurrence generally decreases as $\lambda$ decreases (down to $\lambda \sim 1.0$ fm$^{-1}$), indicating a reduction in correlations within the HO basis. This suppression of correlations, particularly the $s$-$d$ coupling, aligns with the weakening of the repulsive tensor force under SRG flow. However, for very small $\lambda$ and $N > 1$, the behavior becomes non-monotonic.
• Robustness of Optimization: The use of the COBYLA optimizer (gradient-free) combined with ZNE techniques was shown to be robust, yielding consistent results regardless of whether initial parameters were randomly generated or derived from statevector simulations.

Significance and Claims
The paper positions itself as a "first step" in the program of performing ab-initio nuclear structure computations on quantum platforms using realistic, RG-based interactions. The authors claim that their work establishes a direct link between the classical numerical advantages of RG methods (improved convergence) and quantum resource optimization (reduced qubit count).

The study demonstrates that the RG approach, traditionally used to aid classical many-body calculations, is equally beneficial for quantum computing platforms by mitigating errors and reducing the hardware footprint required for accurate simulations. The results validate the reliability of VQE combined with ZNE for nuclear physics problems, achieving experimental accuracy with minimal qubits when appropriate effective interactions are used. The authors note that while the overall trend of decreasing entanglement with lower $\lambda$ is observed, the detailed behavior at very small $\lambda$ requires further understanding. Future extensions to many-body systems (e.g., triton, $\alpha$-nucleus) are identified as necessary next steps.