| Résumé | Variational Quantum Eigensolvers (VQEs) are a promising framework for estimating ground-state energies of molecular systems on near-term quantum hardware. A central challenge in VQE is the construction of ansätze that balance expressibility, trainability, and compatibility with noisy intermediate-scale quantum (NISQ) devices. While hardware-efficient ansätze (HEAs) are well-suited for current quantum hardware, they often lack the chemical structure required for achieving accurate and scalable quantum simulations. In this work, we investigate the performance and efficiency of HEAs in molecular simulations, focusing on the LiH and BeH molecules as examples. We explore different qubit mapping strategies and assess the accuracy and circuit resource requirements of various ansätze architectures. Our results indicate that achieving chemically accurate results with HEAs requires incorporating domain knowledge from physics and chemistry. Such informed modifications to standard HEAs can significantly enhance energy estimation accuracy and optimization robustness, while preserving hardware efficiency. Although the scope of our analysis is limited to small molecular systems and selected ansätze architectures and a fixed optimization method, this study highlights the importance of physically/chemically motivated Ansätze design in advancing quantum chemistry applications on NISQ devices. |
|---|
