Machine Learning Approaches to Quantum Simulators
| When |
Mar 08, 2024
from 11:00 to 12:00 |
|---|---|
| Where | M7-101 |
| Contact Name | Tiago MENDES SANTOS |
| Attendees |
Tiago MENDES SANTOS |
| Add event to calendar |
vCal
iCal
|
Titre: Machine Learning Approaches to Quantum Simulators
Abstract: Programmable quantum devices are now capable of simulating and probing quantum many-body states at unprecedented levels. A current challenge is the extraction of properties and the benchmarking of the large quantum states generated on these devices. In this seminar, we discuss how data mining and machine learning approaches can be utilized to characterize and simulate such systems. Specifically, in the first part, we explore the concept that many-body states, when stochastically sampled (for instance, through projective measurements in quantum simulators), can be considered as high-dimensional manifolds. These manifolds can then be characterized using data set features, such as the intrinsic dimension, or be mapped onto a complex network. By examining physical data sets generated near phase transitions, we demonstrate how typical features of these data sets reveal signatures of the universal behavior of quantum many-body systems. In the second part, we discuss recent advances in simulating the dynamics of many-body systems realized in quantum devices using Neural Quantum States (NQSs). In particular, we show that NQSs allow us to access spectral properties near quantum critical points for system sizes and time scales that pose challenges for other numerical approaches.
Website: https://scholar.google.com/citations?user=jssnMbIAAAAJ&hl=en
Room : M7 101 (ENSL, site Monod)