Researchers Filter Spin to Improve Excited State Quantum Calculations

A new method improves the accuracy of excited state calculations on noisy intermediate-scale quantum (NISQ) computers. Young Kyun Ahn and Young Min Rhee at Korea Advanced Institute of Science and Technology present a spin-filtering variational quantum deflation (sfVQD) scheme that combines a symmetry-preserving ansatz with a shallow quantum phase estimation routine. The approach encodes spin information using a small ancilla register, effectively suppressing spin contamination without requiring computationally expensive evaluations of total spin. Demonstrations on LiH and BeH₂ reveal a sharply improved separation of singlet and triplet manifolds compared to conventional VQD, suggesting a modular and practical pathway towards reliable excited state calculations for physically relevant properties.

Spin contamination mitigation via ancilla-assisted variational quantum deflation enhances excited

A 50% increase in distinction between singlet and triplet manifolds has been achieved in LiH and BeH₂ calculations, a marked improvement over conventional variational quantum deflation (VQD) methods. Previously, spin contamination issues blurred the distinction between these electronic states, hindering accurate excited state determination. Spin contamination arises because the variational wavefunction used to approximate the true electronic state does not perfectly respect the symmetry of the system, specifically the total spin angular momentum. This leads to an unwanted mixing of different spin states, complicating the interpretation of results and reducing accuracy, particularly when calculating properties sensitive to spin. The new spin-filtering variational quantum deflation (sfVQD) scheme employs a symmetry-preserving ansatz alongside a shallow quantum phase estimation routine to encode spin information within an ancilla register, effectively suppressing these errors. The symmetry-preserving ansatz ensures that the initial trial wavefunction has the correct spin symmetry, while the shallow quantum phase estimation (QPE) routine, despite its limited depth, provides sufficient information to filter out unwanted spin components.

The ancilla-assisted screening module operates independently, providing a flexible addition to existing excited-state calculation schemes based on variational quantum eigensolvers (VQEs). Lithium hydride (LiH) and beryllium hydride (BeH₂) were subjected to calculations with varied molecular shapes, revealing a consistent improvement in distinguishing between singlet and triplet electronic states. Geometries were altered to thoroughly test the scheme’s durability. The choice of LiH and BeH₂ as test cases is significant as they represent relatively simple molecules where accurate electronic structure calculations are well-established, allowing for a clear assessment of the sfVQD scheme’s performance. The variation of molecular geometries ensures that the method is not simply optimised for a specific structure but is robust across a range of configurations. Utilising a small ancilla register, the screening module proved flexible enough to integrate with other variational quantum eigensolver-based excited-state calculations, broadening its potential application. The use of a small ancilla register is crucial for minimising the overhead in terms of required qubits, making the scheme more practical for implementation on NISQ devices. Encoding spin information using controlled rotations with modest circuit overhead suggests a pathway towards more efficient quantum computations, though scaling this approach to larger, more complex systems remains a key challenge and will require substantial algorithmic and hardware advancements. The controlled rotations are implemented using the operator $\mathrm{exp} (iθ\hat{S}{x})$, where $\hat{S}{x}$ is the spin operator along the x-axis and θ is a parameter optimised during the variational procedure.

Mitigating spin contamination enhances accuracy in variational quantum eigensolver calculations

This work offers a promising route to more accurate excited state calculations, while also highlighting a fundamental tension within the field of variational quantum eigensolvers. The authors acknowledge that current demonstrations are limited to small molecules, and scaling to larger systems presents a significant hurdle. However, the broader challenge lies in balancing the need for expressive, accurate wavefunctions with the constraints of available quantum hardware. Unitary coupled-cluster theory offers high accuracy but demands substantial computational resources, even on classical computers. Traditional methods like configuration interaction (CI) can achieve high accuracy but suffer from exponential scaling with system size, rendering them impractical for all but the smallest molecules. VQEs offer a potential solution by leveraging the principles of quantum mechanics to perform calculations more efficiently, but they are susceptible to errors like spin contamination and require careful selection of the ansatz and optimisation strategy.

Despite the limitations of applying it to complex systems, this development remains important as it offers a practical, modular improvement to existing excited state calculations. ‘Spin contamination’, an error arising from inaccuracies in representing electron behaviour, poses a challenge for variational quantum eigensolvers, a core technique in quantum computing for modelling molecules. A small additional quantum register filters out these errors within the new scheme, improving accuracy without drastically increasing the demands on current quantum hardware. The demonstrated improvement in distinguishing between singlet and triplet electronic states in lithium and beryllium hydride suggests a viable path towards more reliable molecular simulations. Accurate excited state calculations are crucial for understanding a wide range of chemical and physical phenomena, including photochemistry, spectroscopy, and materials science.

Errors caused by ‘spin contamination’ have been demonstrably reduced by combining a symmetry-preserving approach with a shallow quantum phase estimation routine. This filtering method operates independently of the core calculation, offering a flexible upgrade to existing quantum algorithms. A new technique has been developed to refine calculations of excited states in molecules, a vital step for modelling chemical reactions and understanding the underlying principles of molecular behaviour. The sfVQD scheme’s modularity allows it to be incorporated into existing VQE workflows without requiring a complete overhaul of the underlying algorithm. Future research will likely focus on extending this approach to larger molecules and exploring its performance on different quantum hardware platforms, ultimately paving the way for more accurate and efficient molecular simulations.

The researchers developed a new scheme, sfVQD, which improves the accuracy of excited state calculations for molecules like lithium hydride and beryllium hydride. This method reduces errors arising from spin contamination by using a small additional quantum register to filter out inaccuracies, without substantially increasing computational demands. The filtering module functions independently, meaning it can be added to existing quantum algorithms. These findings suggest a practical and adaptable route to more reliable molecular simulations, crucial for understanding chemical and physical phenomena.