SKM 2023 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

DY: Fachverband Dynamik und Statistische Physik

DY 46: Poster: Machine Learning and Data Analytics

DY 46.3: Poster

Donnerstag, 30. März 2023, 13:00–16:00, P1

Machine learning categorization of the Anderson model — •Quangminh Bui-Le and Rudolf Römer — Department of Physics, University of Warwick, Coventry, CV4 7AL

Machine learning (ML) methods have been used to identify phase transitions of physical systems by categorizing systems based on the Ψ² values of their wave-functions into extended and localized states, which a model is then trained on in order to identify between the extended and localized states. Here we want to see if ML is powerful enough to categorize systems into even more specific groups by attempting to categorize Anderson model data into categories based on the disorder of the wave-function. We are using a PyTorch model to create a convolutional neural network using a ResNet18 model. This model will be trained on 3D Anderson model Ψ² values from 17 disorder values spanning a range of 15 to 18.