Neural Net Analysis of Phase Transitions in Geometrically Frustrated Classical Models. Analysis of Generalization Ability and Reliability of Predictions

Student: Bileva Anna

Supervisor: Evgeny Burovskiy

Faculty: HSE Tikhonov Moscow Institute of Electronics and Mathematics (MIEM HSE)

Educational Programme: Applied Mathematics (Bachelor)

Final Grade: 8

Year of Graduation: 2021

In this paper, we investigate the ability of neural networks to generalize information they obtain from training data and find out if they recognize the underlying physical structure of the model on which they are being trained. We use a novel simple approach to determine the critical temperature of the 2D Ising model using a neural network and discuss whether a network trained to distinguish states of a non-frustrated model can be applied to a frustrated one from the same universality class. We also try to interpret the output of the neural network and see what it takes into account to make a prediction.

Full text (added May 30, 2021)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses