A Randomized Approach to Structure-Preserving Hamiltonian Simulation
Date:
Abstract：
Accurate and efficient Hamiltonian simulation is a cornerstone application of quantum computing, yet existing digital methods often introduce errors that violate fundamental physical conservation laws. In this talk, I will present PhysDrift, a randomized Hamiltonian simulation technique that extends qDrift by incorporating physical constraints, such as particle number conservation. By leveraging structured term grouping, PhysDrift mitigates spectral errors and reduces state leakage into unphysical subspaces, improving both accuracy and feasibility for near-term quantum devices.
I will discuss the theoretical foundations of PhysDrift, its relationship to existing randomized simulation protocols, and its empirical performance on molecular systems. Additionally, I will examine how this approach interacts with noise models and error mitigation techniques, demonstrating its robustness against realistic hardware limitations. By preserving key symmetries and leveraging stochastic sampling, PhysDrift offers a promising pathway toward more reliable digital quantum simulations, bridging the gap between algorithmic efficiency and physical fidelity.