Aug 8 – 13, 2022
Hörsaalzentrum Poppelsdorf
Europe/Berlin timezone

Generative models for scalar field theories: how to deal with poor scaling?

Aug 10, 2022, 5:50 PM
CP1-HSZ/1.004 (CP1-HSZ) - HS7 (CP1-HSZ)

CP1-HSZ/1.004 (CP1-HSZ) - HS7


Show room on map
Oral Presentation Algorithms (including Machine Learning, Quantum Computing, Tensor Networks) Algorithms


Javad Komijani (ETH Zurich)


Deep generative models such as normalizing flows are suggested as alternatives to standard methods for generating lattice gauge field configurations. Previous studies on normalizing flows demonstrate proof of principle for simple models in two dimensions. However, further studies indicate that the training cost can be, in general, very high for large lattices. The poor scaling traits of current models indicate that moderate-size networks cannot efficiently handle the inherently multi-scale aspects of the problem, especially around critical points. In this talk, we explore current models that lead to poor acceptance rates for large lattices and explain how to use effective field theories as a guide to design models with improved scaling costs. Finally, we discuss alternative ways of handling poor acceptance rates for large lattices.

Primary authors

Javad Komijani (ETH Zurich) Marina Krstic Marinkovic (ETH Zurich)

Presentation materials