Extrapolation for observable estimation

Quantum Scientific Computation and Quantum Artificial Intelligence量子科学计算与量子人工智能

Organizer：

Jin-Peng Liu 刘锦鹏

Speaker：

Chang Liu (China Academy of Engineering Physics)

Time：

Wednesday, 10:30-11:30 am

Dec 17, 2025

Venue：

Ningzhai 204

Voov: 963-193-309

Title：

Extrapolation for observable estimation

Abstract:

Estimating expectation values of observables under quantum dynamics is a central primitive in quantum algorithms, often requiring substantially less information than full state or unitary reconstruction. This distinction exposes additional mathematical structure that can be exploited algorithmically. In this talk, I discuss extrapolation-based approaches to observable estimation, using Hamiltonian simulation as a primary example. I show that expectation values obtained from Trotterized dynamics exhibit controlled analytic and polynomial dependence on simulation parameters such as the step size, enabling systematic accuracy improvement via extrapolation and interpolation without increasing circuit depth. I will further explore how similar ideas extend to a broader class of observables and dynamical settings.