Many competing theories exist for certain phenomena in quantum many-body physics, but despite this fact, no one knows which of them best describes a quantum phenomenon.
Now, a research team from Harvard University in the United States and the Technical University of Munich (TUM) has effectively used artificial neural networks for image analysis of quantum systems.
Is that a cat or a dog? A classification like that is a major example of machine learning: it is possible to train artificial neural networks to examine images by searching for patterns that are typical of particular objects. If the system has learned such kinds of patterns, it would be able to identify cats or dogs on any kind of picture.
Neural networks—using the same principle—are capable of detecting differences in tissue on radiological images. Currently, physicists are applying this technique to study images—what is known as snapshots—of quantum many-body systems and determine which theory elucidates the observed phenomena most favorably.
The Quantum World of Probabilities
Condensed matter physics focuses on liquids and solids, but a number of phenomena in this field continue to be a mystery. For instance, to date, it is not known why the electrical resistance of high-temperature superconductors reduces to zero at temperatures of approximately –200 ºC.
It is difficult to understand such unusual states of matter. Therefore, to study the physics of high-temperature superconductors, quantum simulators based on ultra-cold Lithium atoms have been created. These simulators capture the images of the quantum system, which exists concurrently in varied configurations—researchers speak of a superposition. Each picture of the quantum system provides a particular configuration in accordance with its quantum mechanical probability.
Although various theoretical models have been created to interpret such quantum systems, it is not known how well they reflect reality. Here. the image data can be analyzed to address this concern.
Neural Networks Investigate the Quantum World
As such, a team of researchers at TUM and at Harvard University has used machine learning successfully: The group trained an artificial neural network to differentiate between two opposing theories.
Similar to the detection of cats or dogs in pictures, images of configurations from every quantum theory are fed into the neural network. The network parameters are then optimized to give each image the right label—in this case, they are just theory A or theory B instead of cat or dog.
Annabelle Bohrdt, Doctoral Student, Department of Physics, Technical University of Munich
Post the training phase with hypothetical data, the artificial neural network needs to apply what it had learned and eventually assign images from the quantum simulators to theory A or B. Thus, the artificial neural network chose the more predictive theory.
In the coming days, the scientists are planning to apply this latest technique to evaluate the precision of a number of theoretical descriptions. The objective is to figure out the key physical effects of high-temperature superconductivity, which can be used in a wide range of significant applications, with efficient magnetic resonance imaging and lossless electric power transmission being just a couple of examples.
The study was funded by the National Science Foundation (NSF), the US Air Force’s Office of Scientific Research (AFOSR), the National Defense Science and Engineering Graduate (NDSEG) Program of the US-Department of Defense, the Gordon and Betty Moore Foundation EPIQS program, the Studienstiftung des deutschen Volkes, the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) as part of the Cluster of Excellence Munich Center for Quantum Science and Technology (MCQST) and the Transregio TRR80 as well as the TUM Institute for Advanced Study, funded by the German Excellence Initiative and the European Union, where Professor Knap holds the Rudolf Mößbauer Tenure Track Professorship for Collective Quantum Dynamics.