Variational Neural Annealing
Presented by Juan Felipe Carrasquilla (Vector Institute and University of Waterloo)
Many important challenges in science and technology can be cast as optimization problems. When viewed in a statistical physics framework, these can be tackled by simulated annealing, where a gradual cooling procedure helps search for ground state solutions of a target Hamiltonian. While powerful, simulated annealing is known to have prohibitively slow sampling dynamics when the optimization landscape is rough or glassy. In this talk I will show that by generalizing the target distribution with a parameterized model, an analogous annealing framework based on the variational principle can be used to search for ground state solutions. Autoregressive models such as recurrent neural networks provide ideal parameterizations since they can be exactly sampled without slow dynamics even when the model encodes a rough landscape. We implement this procedure in the classical and quantum settings on several prototypical spin glass Hamiltonians, and find that it significantly outperforms traditional simulated annealing in the asymptotic limit, illustrating the potential power of this yet unexplored route to optimization.
3:30 pm, Thursday, November 4, 2021
Individuals with disabilities who need an auxiliary aid or service to attend or participate in P&A events should contact the Physics Department (phone: 505-277-2616, email: email@example.com) well in advance to ensure your needs are accomodated. Event handouts can be provided in alternative accessible formats upon request. Please contact the Physics front office if you need written information in an alternative format.
A schedule of talks within the Department of Physics and Astronomy is available on the P&A web site at http://physics.unm.edu/pandaweb/events/index.php