Learning models of quantum systems from experiments

As Hamiltonian models underpin the study and analysis of physical and chemical processes, it is crucial that they are faithful to the system they represent. However, formulating and testing candidate Hamiltonians for quantum systems from experimental data is difficult, because one cannot directly ob...

Full description

Saved in:
Bibliographic Details
Published in:Nature physics 2021-07, Vol.17 (7), p.837-843
Main Authors: Gentile, Antonio A., Flynn, Brian, Knauer, Sebastian, Wiebe, Nathan, Paesani, Stefano, Granade, Christopher E., Rarity, John G., Santagati, Raffaele, Laing, Anthony
Format: Article
Language:English
Subjects:
639/705/1042
639/766/483
639/766/483/3926
639/766/483/481
639/925/927/481
Atomic
Automation
Chemical reactions
Classical and Continuum Physics
Complex Systems
Condensed Matter Physics
Experimental data
Genetic algorithms
Machine learning
Mathematical and Computational Physics
Molecular
Nitrogen
Optical and Plasma Physics
Physics
Physics and Astronomy
Target recognition
Theoretical
Vacancies
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:As Hamiltonian models underpin the study and analysis of physical and chemical processes, it is crucial that they are faithful to the system they represent. However, formulating and testing candidate Hamiltonians for quantum systems from experimental data is difficult, because one cannot directly observe which interactions are present. Here we propose and demonstrate an automated protocol to overcome this challenge by designing an agent that exploits unsupervised machine learning. We first show the capabilities of our approach to infer the correct Hamiltonian when studying a nitrogen-vacancy centre set-up. In preliminary simulations, the exact model is known and is correctly inferred with success rates up to 59%. When using experimental data, 74% of protocol instances retrieve models that are deemed plausible. Simulated multi-spin systems, characterized by a space of 10 10 possible models, are also investigated by incorporating a genetic algorithm in our protocol, which identifies the target model in 85% of instances. The development of automated agents, capable of formulating and testing modelling hypotheses from limited prior assumptions, represents a fundamental step towards the characterization of large quantum systems. Quantum systems make it challenging to determine candidate Hamiltonians from experimental data. An automated protocol is presented and its capabilities to infer the correct Hamiltonian are demonstrated in a nitrogen-vacancy centre set-up.
ISSN:1745-2473
1745-2481
DOI:10.1038/s41567-021-01201-7