Resource-Efficient Simulation of Molecular Ground and Excited States Based on Contextual Subspace and Quantum...

Predicting spectra and photochemical pathways relies on the computation of molecular excited states. However, the practicality of near-term variational quantum algorithms is threatened by the prohibitive growth of measurement overheads. Integrating Quantum Subspace Expansion with the Contextual Subspace (CS) method, we propose a CS-QSE framework to resolve the scaling bottlenecks in excited-state energy estimation. By confining excitation operators to a compact CS, the framework removes redundant degrees of freedom and reduces Pauli string counts, thereby easing the measurement burden. The primary strength of CS-QSE lies in its substantial reduction of the operator-pool size. Benchmarking on LiH, HF, H2O, and HCl shows that CS-QSE achieves errors within the target tolerance of 1.6 × 10-3 Ha relative to full configuration interaction benchmarks in the same basis set while mitigating the prohibitive scaling inherent in the full-space method. Numerical simulations reveal that the operator pool size is consistently pruned by over 90% across all systems, with the reduction reaching as high as 99.7% for molecules such as HCl. This CS-QSE framework establishes a resource-efficient route for molecular simulations tailored to near-term noisy intermediate-scale quantum devices.