Machine learning applications in physics is a rapidly expanding and very broad field. We have been working on two different machine learning projects in quantum physics. One is using deep reinforcement learning for topological error correcting codes that are being developed for the purpose of quantum computing. The other project relates to classifying topological phases in condensed matter physics using deep learning. Both problems are quite closely related in terms of machine learning methodology, using various neural network architectures.
We are also using HPC resources on Vera, within project C3SE 2021/1-2. The storage will also be used for our needs within that project. In addition to the co-applicants, people that have been involved in these projects (and the previous C3SE 2020/1-18) are (CID) hamkarl, adaolss, gablinde, orokhov, erijoel, hybelius, gablinde, annawis.
Relevant publications are:
Unsupervised interpretable learning of topological indices invariant under permutations of atomic bands, O Balabanov, M Granath, arXiv preprint arXiv:2008.07268 (2020)
Deep Q-learning decoder for depolarizing noise on the toric code
D Fitzek, M Eliasson, AF Kockum, M Granath
Physical Review Research 2 (2), 023230 (2020)
Unsupervised learning using topological data augmentation
O Balabanov, M Granath
Physical Review Research 2 (1), 013354 (2020)
Quantum error correction for the toric code using deep reinforcement learning
P Andreasson, J Johansson, S Liljestrand, M Granath
Quantum 3, 183 (2019)