⚛️ nuclear theory

Quantum Simulation of Nuclear Shell Model Using GCM-Based Methods on NISQ Devices

This paper demonstrates a robust and scalable approach for simulating low-lying nuclear eigenstates on noisy intermediate-scale quantum (NISQ) devices by employing a hybrid quantum-classical Generator Coordinate Method (QuGCM) enhanced with an adaptive selection strategy (ADAPT-GCIM) and optimized Gray code encoding, achieving accurate energy spectra for systems like the deuteron, $^6$ Li, and $^{38}$ Ar that align with classical results despite hardware limitations.

Original authors: Durgesh Pandey, Ashutosh Singh, Ankit Kumar Das, P. Arumugam

Published 2026-08-04

📖 3 min read🧠 Deep dive

CC BY 4.0

Original authors: Durgesh Pandey, Ashutosh Singh, Ankit Kumar Das, P. Arumugam

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe is built from tiny, invisible Lego bricks called atoms. Inside those atoms are even smaller particles called protons and neutrons, which huddle together in a nucleus. For decades, scientists have tried to figure out exactly how these particles dance and interact, but the math is so incredibly complicated that even the world's fastest supercomputers sometimes get stuck. It's like trying to predict the weather by tracking every single raindrop; there are just too many variables. This is where quantum physics steps in, offering a new kind of computer that doesn't just calculate numbers but actually acts like the tiny particles it's studying. The big question researchers are asking is: Can we use these new, experimental quantum computers to solve the mysteries of the atomic nucleus without getting lost in the noise and errors that currently plague these machines?

This paper is like a set of blueprints for building a better, more efficient quantum computer program specifically designed to solve nuclear puzzles. The authors, a team from India and France, are testing a new strategy called the "Generator Coordinate Method" (GCM) on these noisy, early-stage quantum devices. Think of the nucleus as a crowded dance floor. Traditional methods try to map every single dancer's move perfectly, which takes forever and requires a massive amount of space. The authors' new approach, called QuGCM, is smarter: instead of watching everyone, it picks a few key "dance moves" (called generator states) that capture the essence of the group's behavior. By mixing these specific moves together, they can figure out the energy levels of the nucleus without needing to simulate every single possibility.

To make this even faster and less prone to errors, they introduced a "self-learning" version called ADAPT-GCIM. Imagine a chef who doesn't just follow a fixed recipe but tastes the soup after every step and only adds the next ingredient if it actually improves the flavor. This method automatically picks the most important "dance moves" to add to the mix, skipping the ones that don't help. They tested this on three different nuclear systems: a simple pair of particles (deuteron), a medium-sized nucleus (Argon-38), and a slightly more complex one (Lithium-6). They also compared two different ways of translating the nuclear math into computer code: an old, clunky method (Jordan-Wigner) and a sleek, compressed new method (Gray Code).

The results are promising. In their simulations, which mimic real quantum computers but include the "noise" and errors you'd expect from current hardware, their new methods worked beautifully. They found the correct energy levels for the nuclei with high accuracy, often doing better than the standard methods used today. The "self-learning" version (ADAPT-GCIM) was particularly impressive because it needed far fewer steps and less computer power to get the right answer. Furthermore, using the sleek "Gray Code" translation made the circuits shorter and less likely to break, much like taking a shortcut through a maze instead of walking the long way around. While these results are currently based on simulations and not yet run on a physical quantum machine, the study suggests that this combination of smart, adaptive algorithms and efficient coding could be the key to unlocking the secrets of the atomic nucleus on the quantum computers we have today.

Technical Summary: Quantum Simulation of Nuclear Shell Model Using GCM-Based Methods on NISQ Devices

Problem Statement
Simulating correlated quantum systems, particularly nuclear many-body systems, presents significant challenges for classical computers due to the exponential growth of the Hilbert space with the number of collective coordinates. Traditional methods like the Generator Coordinate Method (GCM) require constructing large grids of generator states and solving generalized eigenvalue problems, which become computationally intractable as the system size increases. While Variational Quantum Eigensolvers (VQE) and their adaptive variants (ADAPT-VQE) have been proposed for Noisy Intermediate-Scale Quantum (NISQ) devices, they suffer from optimization difficulties, including barren plateaus, convergence to local minima, and the need for deep quantum circuits. Furthermore, standard fermion-to-qubit mappings, such as the Jordan-Wigner (JW) transformation with One-Hot (OH) encoding, often result in high qubit counts and deep circuits due to non-local parity strings, limiting their applicability on current hardware.

Methodology
The authors propose a hybrid quantum-classical framework utilizing two primary algorithms: Quantum Generator Coordinate Method (QuGCM) and its adaptive variant, Adaptive Generator Coordinate Inspired Method (ADAPT-GCIM).

Theoretical Framework:
- QuGCM: This approach constructs the total wavefunction as a superposition of non-orthogonal generator states. These states are generated by applying symmetry-adapted Unitary Coupled Cluster (UCC) operators (specifically UCCSD) to Hartree-Fock (HF) reference states. Unlike VQE, QuGCM does not optimize variational parameters; instead, it prepares these states as quantum circuits, measures pairwise overlaps and Hamiltonian kernels, and solves the resulting Hill-Wheeler generalized eigenvalue equation ( $H\psi = EN\psi$ ) on a classical computer.
- ADAPT-GCIM: To enhance efficiency, this method iteratively selects generator excitations based on energy gradients. It dynamically expands the Hilbert subspace by adding the operator that contributes most to energy reduction, avoiding the need to explore the full basis set. This reduces the number of required quantum circuits and basis states.
Fermion-to-Qubit Mappings:
The paper compares two encoding strategies:
- Jordan-Wigner (JW) with One-Hot (OH) encoding: The standard approach where each orbital maps to a qubit.
- Gray Code (GC) encoding: A binary-reflected ordering where consecutive basis states differ by a single bit. The authors demonstrate that GC encoding allows for a more compact representation of the Hilbert space and generates simpler, more local Pauli operators for excitations, thereby reducing circuit depth and gate counts compared to JW.
Implementation:
The methods were implemented for three nuclear systems: the deuteron (using the Reid68 potential), $^{38}$ Ar (shell-model Hamiltonian), and $^{6}$ Li (shell-model Hamiltonian). Simulations were performed using both noiseless state-vector simulators and noisy simulators (FakeBrisbane) to mimic IBM quantum hardware.

Key Results

Accuracy and Convergence: Both QuGCM and ADAPT-GCIM produced energy spectra in agreement with exact classical diagonalization results for all tested systems. ADAPT-GCIM demonstrated rapid convergence, often reaching the ground state energy within a few iterations (e.g., three iterations for the deuteron).
Noise Resilience: Under simulated noise conditions, the GCM-based methods exhibited greater stability than variational methods (VQE/ADAPT-VQE). Specifically, ADAPT-GCIM and QuGCM results remained closer to exact values with lower statistical uncertainty (Median Absolute Deviation) compared to their variational counterparts.
Efficiency of GC Encoding: The Gray Code encoding significantly outperformed JW encoding. For $^{38}$ Ar, while the JW mapping requires 6 qubits, the GC encoding reduced the number of required qubits for specific states (e.g., 3 qubits for the $0^+$ state, 2 for $1^+$ , and 1 for $2^+$ ) and drastically reduced the number of Pauli terms and quantum circuits required for QuGCM (approximately 27 times lower than JW). For $^{6}$ Li, GC encoding reduced the qubit requirement from 12 (in full m-scheme JW) to 4, enabling simulations that would otherwise be infeasible.
Resource Reduction: The adaptive nature of ADAPT-GCIM drastically reduced the number of basis states needed. For instance, in $^{38}$ Ar, ADAPT-GCIM required only 3–6 basis states compared to the full 15 required by standard QuGCM, significantly lowering the computational load.

Significance and Claims
The paper claims that the integration of QuGCM and ADAPT-GCIM with Gray Code encoding provides a practical and scalable path for simulating correlated nuclear systems on NISQ devices. The primary significance lies in:

Avoiding Optimization Pitfalls: By solving a generalized eigenvalue problem rather than minimizing a cost function, the methods circumvent barren plateaus and local minima issues inherent to variational algorithms.
Hardware Compatibility: The combination of non-variational, parameter-free (or fixed-parameter) circuits and the resource-efficient GC encoding results in shallower circuits with fewer gates, making the approach more robust against noise and better suited for current quantum hardware limitations.
Bridging Theory and Computation: The work establishes QuGCM as a quantum analog of the classical GCM, offering a pathway to study nuclear correlations and collective phenomena that are classically intractable, without the computational burden of large basis grids.

The authors conclude that while these methods show promise, future work involves benchmarking on real quantum hardware and extending the framework to compute additional observables like electromagnetic transition rates.

More like this