BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Decisional Gradient Descent: A New Optimizer for Variational Monte
  Carlo
DTSTART;VALUE=DATE-TIME:20241120T093000Z
DTEND;VALUE=DATE-TIME:20241120T094500Z
DTSTAMP;VALUE=DATE-TIME:20260421T165125Z
UID:indico-contribution-25464@indico.ific.uv.es
DESCRIPTION:Speakers: Javier Rozalén Sarmiento (Universitat de Barcelona)
 \nThe nuclear many-body problem is known to be computationally expensive t
 o solve. Recently\, with the advent of machine learning techniques in scie
 nce\, the method of Neural-Network Quantum States is being adopted by diff
 erent groups to tackle nuclear systems\, with the hope that it will be mor
 e efficient than the alternatives. Being a variational method\, one of the
  inherent difficulties is to optimize the energy. Even for simple systems\
 , the preferred optimization algorithm\, Stochastic Reconfiguration\, does
  not guarantee a smooth convergence towards the energy minimum. In this ta
 lk\, I present our latest optimizer\, Decisional Gradient Descent\, from t
 he point of view of second-order optimization theory. Not only does it con
 sistently outperform the state-of-the-art Stochastic Reconfiguration (for 
 our system of choice)\, but also the theoretical framework used to derive 
 it is very wide. We believe this will allow for the development of several
  powerful optimizers within this decisional framework.\n\nhttps://indico.i
 fic.uv.es/event/7664/contributions/25464/
LOCATION:
URL:https://indico.ific.uv.es/event/7664/contributions/25464/
END:VEVENT
END:VCALENDAR