For some phenomena of multi-body quantum physics, several competing theories exist. But which one of them best describes a quantum phenomenon? A team of researchers from the Technical University of Munich (TUM) and Harvard University in the United States has now successfully deployed artificial neural networks for image analysis of quantum systems.
Is it a dog or a cat? Such a classification is an excellent example of automatic learning: Artificial neural networks can be trained in image analysis by searching for characteristic patterns of specific objects. Provided the system has learned such models, it is able to recognize dogs or cats on all images.
Using the same principle, neural networks can detect tissue changes on radiological images. Physicists are now using this method to analyze images, called snapshots, of multiple-body quantum systems and determine which theory best describes the phenomena observed.
The quantum world of probabilities
Several phenomena of the physics of condensed matter, which study solids and liquids, remain mysterious. For example, until now, it is difficult to understand why the electrical resistance of high temperature superconductors drops to zero at temperatures of about -200 degrees Celsius.
Understanding such extraordinary states of matter is a challenge: quantum simulators based on ultra-cold lithium atoms have been developed to study the physics of superconductors at high temperatures. They take snapshots of the quantum system, which exists simultaneously in different configurations – the physicists speak of superposition. Each snapshot of the quantum system gives a specific configuration according to its quantum mechanical probability.
In order to understand such quantum systems, various theoretical models have been developed. But to what extent do they reflect reality? You can answer the question by analyzing the image data.
Neural networks study the quantum world
To this end, a team of researchers from the Technical University of Munich and Harvard University successfully used machine learning: The researchers formed a network of artificial neurons to distinguish two competing theories.
"Similar to the detection of images of cats or dogs, images of configurations of all quantum theories are introduced into the neural network," says Annabelle Bohrdt, PhD student at TUM. "The network parameters are then optimized to give each image the appropriate label, in which case it is only theory A or theory B instead of cat or dog."
After the training phase with theoretical data, the neural network had to apply what it had learned and assign quantum simulator snapshots to theory A or B. The network thus selected the most predictive theory.
In the future, researchers plan to use this new method to evaluate the accuracy of several theoretical descriptions. The goal is to understand the main physical effects of high temperature superconductivity, which has many important applications, such as lossless electrical power transmission and efficient magnetic resonance imaging.
Simulation of quantum systems with neural networks
Annabelle Bohrdt et al, Snapshot classification of the Hubbard model doped with machine learning, Physical Nature (2019). DOI: 10.1038 / s41567-019-0565-x
What is the perfect quantum theory? (July 12, 2019)
recovered on July 12, 2019
from https://phys.org/news/2019-07- quantum-theory.html
This document is subject to copyright. Apart from any fair use for study or private research purposes, no
part may be reproduced without written permission. Content is provided for information only.