Quantum Simulation Cuts Error in Complex Materials Modelling to below 1 Per Cent

Scientists are developing new methods to simulate the complex behaviour of electrons in materials, a crucial step towards designing novel electronic devices. Aadi Singh, Chakradhar Rangi, and Ka-Ming Tam, from the Center for Computation and Technology and Department of Physics and Astronomy at Louisiana State University, present a symmetry-adapted variational quantum eigensolver to tackle the Anderson Impurity Model, a key component of Dynamical Mean-Field Theory. Their research addresses limitations in current approximations by exploring larger, more accurate models within the constraints of existing quantum hardware. This work is significant because it demonstrates the potential to accurately calculate the single-particle Green’s function, a vital quantity for understanding material properties, using a quantum-classical hybrid approach, effectively extending beyond simplified two-site models and offering a pathway to simulate strongly correlated materials on near-term quantum computers.

Recognizing the limitations of simplified two-site approximations in accurately capturing essential spectral features, this research investigates the effectiveness of VQE for larger bath discretizations while remaining compatible with the constraints of current quantum hardware.

The team implemented a symmetry-adapted ansatz, rigorously enforcing conservation of particle number, spin projection, and total spin symmetry, and meticulously benchmarked its performance against exact diagonalization across a range of interaction strengths using parameters derived from the DMFT self-consistency loop. For a four-site model, the relative error in calculating the ground state energy remained consistently below 0.01% using a compact parameter set containing no more than 30 parameters.
Critically, the single-particle Green’s function, a central quantity in DMFT, was successfully extracted from the VQE-prepared ground states through real-time evolution, demonstrating accuracy in the intermediate to strong interaction regimes. However, noticeable deviations emerged in the weak interaction regime, particularly in resolving fine details of low-energy spectral features, despite the excellent agreement in ground state energy calculations.

These findings demonstrate that the combination of VQE with real-time evolution effectively extends quantum-classical hybrid DMFT beyond the limitations of the two-site approximation, offering a promising route for describing insulating phases of materials. While this approach presents a viable pathway for simulating strongly correlated materials on near-term quantum devices, the observation that accurate ground state energy does not automatically guarantee accurate dynamical properties underscores a significant challenge for applying these methods to correlated metals. Recognizing limitations of the minimal two-site approximation, researchers investigated the efficacy of VQE with larger bath discretizations while considering constraints imposed by near-term quantum hardware.

The study employed a symmetry-adapted ansatz designed to enforce conservation of particle number, spin projection along the z-axis, and total spin symmetry, facilitating benchmarking against exact diagonalization across a range of interaction strengths. For a four-site model, the relative error in the calculated ground state energy remained consistently below 0.01% using a compact parameter set.

Crucially, the single-particle Green’s function, a central quantity in DMFT, was extracted from VQE-prepared ground states through real-time evolution utilizing the Suzuki-Trotter decomposition. This real-time evolution enabled the probing of dynamical properties, particularly in the intermediate to strong interaction regimes where accurate results were obtained.

However, analysis revealed noticeable deviations in the Green’s function from the exact benchmark in the weak interaction regime, specifically in resolving low-energy spectral features, despite excellent agreement in ground state energies. This observation highlights a key challenge for applying variational approaches to correlated metals, demonstrating that accurate ground state energy does not necessarily guarantee accurate dynamical properties. The work demonstrates that VQE, combined with real-time evolution, can effectively extend quantum-classical hybrid DMFT beyond the two-site approximation, offering a viable pathway for simulating strongly correlated materials on near-term quantum devices and describing insulating phases.

Ground state energy and single-particle Green’s function accuracy across interaction strengths are demonstrably linked

For a four-site model, relative error in the ground state energy remained well below 1% with a compact parameter set. This compact parameter set facilitated accurate ground state energy calculations despite the complexity of the Anderson Impurity Model. However, in the weak interaction regime, the Green’s function exhibited noticeable deviations from the exact benchmark, particularly in resolving low-energy spectral features, despite excellent agreement in ground state energy.

These deviations, while limited to the weak interaction regime, highlight a key challenge in applying this approach to correlated metals. The research utilized a symmetry-adapted ansatz enforcing conservation of particle number, spin projection, and total spin symmetry, benchmarking performance against exact diagonalization across different interaction strengths.

Bath parameters were extracted from the DMFT self-consistency loop to ensure accurate representation of the system. The study employed a four-site model to assess the efficacy of the variational quantum eigensolver for larger bath discretizations while adhering to near-term hardware constraints. Real-time evolution was performed on the optimized wavefunctions to compute the single-particle Green’s function, providing insights into the system’s dynamical properties.

The findings demonstrate that the combination of variational quantum eigensolver with real-time evolution can effectively extend quantum-classical hybrid DMFT beyond the two-site approximation, particularly for describing insulating phases. This extension is significant for simulating strongly correlated materials on near-term devices.

The work focused on a discretized Anderson Impurity Model with a linear chain geometry, mapping fermionic operators to qubits via the Jordan-Wigner transformation. This transformation minimized long-range strings in the Pauli operators, simplifying both classical exact diagonalization and quantum circuit implementation. This method addresses limitations found in simpler two-site approximations by investigating larger bath discretizations suitable for current quantum hardware.

The research employs a symmetry-adapted ansatz, conserving particle number, spin projection, and total spin, and benchmarks its performance against exact diagonalization across varying interaction strengths. For a four-site model, the VQE method achieves ground state energies with relative errors below one percent using a concise parameter set.

Critically, the single-particle Green’s function, essential for DMFT calculations, can be accurately determined from VQE-prepared ground states through real-time evolution, particularly in intermediate to strong interaction regimes. However, the accuracy of the Green’s function diminishes in the weak interaction regime, where resolving low-energy spectral features proves challenging despite excellent agreement in ground state energy.

This work extends quantum-classical hybrid DMFT beyond the two-site approximation, enabling the description of insulating phases and offering a pathway for simulating strongly correlated materials on near-term quantum devices. A key limitation identified is that achieving accurate ground state energies does not automatically guarantee accurate dynamical properties, posing a challenge for applying this approach to correlated metals.

Future research may focus on improving the resolution of low-lying excitations and refining the method for application to systems with stronger correlations, potentially enabling more detailed spectral analysis. The demonstrated viability of this VQE approach with realistic DMFT parameters signifies a practical advancement for simulating genuine impurity problems within DMFT workflows.