Transfer learning for neural-network quantum states

Remmy Zen

Neural network quantum states are neural networks that are used to represent the states or wave functions of quantum systems. Following a variational approach, different kinds of neural networks can be trained to be surrogates of the wave functions that, for example, minimize the energy of the ground state for a given quantum many-body system. We proposed to use transfer learning techniques, using networks obtained for a given set of parameters for another set, in order to improve the effectiveness and efficiency of this method to describe large systems, to explore phase transitions by varying parameters or to calculate excited states and gaps.

Dr. Remmy Zen is a postdoctoral fellow at the Max Planck Institute for the Science of Light, Germany. His main research interest is in the intersection between machine learning and quantum physics, particularly how machine learning can help to solve quantum physics problems. He is currently working on reinforcement learning methods to make a better quantum computer under the Munich Quantum Valley project. He got his PhD degree from the National University of Singapore, Singapore. His PhD thesis was on transfer learning protocols for neural-network quantum states.