End-to-End Quantum Estimation of Non-Hermitian Pseudospectra

Quantum Scientific Computation and Quantum Artificial Intelligence

Organizer：

Jin-Peng Liu 刘锦鹏

Speaker：

Weiliang Wang 王纬梁

Time：

Wed., 17:15-18:15, May 27, 2026

Venue：

Ningzhai 204

Title：

End-to-End Quantum Estimation of Non-Hermitian Pseudospectra

Abstract:

Non-Hermitian operators appear naturally in open quantum systems, dissipative dynamics, and non-reciprocal quantum matter. Unlike the Hermitian case, their spectra are often highly unstable under small perturbations, motivating the study of pseudospectra as a more robust and physically meaningful object.

In this talk, we introduce the recent work “Towards End-to-End Quantum Estimation of Non-Hermitian Pseudospectra.” We will explain how pseudospectrum estimation can be reformulated as a smallest-singular-value problem, discuss the complexity-theoretic perspective behind the paper, and clarify the distinction between spectral instability and computational hardness.

The main focus will be the paper’s end-to-end quantum algorithmic framework: Gaussian-filtered singular-value estimation, dissipative preparation of approximate singular-vector states, and their integration into a practical pipeline for estimating non-Hermitian pseudospectra.

References:

[1] G. Yang, J. Leng, X. Wu, and L. Lin, “Towards End-to-End Quantum Estimation of Non-Hermitian Pseudospectra,” Mar. 17, 2026, arXiv: arXiv:2603.16214. doi: 10.48550/arXiv.2603.16214.

Bio: Weiliang Wang (王纬梁), undergraduate student of Institute for Interdisciplinary Information Science, Tsinghua University