BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Optimizing Neural Quantum States with Decision Geometry
DTSTART:20260710T120000Z
DTEND:20260710T130000Z
DTSTAMP:20260717T110800Z
UID:indico-event-14066@indico.ijclab.in2p3.fr
CONTACT:pierre.arthuis@ijclab.in2p3.fr
DESCRIPTION:Speakers: Mehdi Drissi (TU Darmstadt)\n\nOver the past few yea
 rs\, the capabilities of Neural Quantum States (NQS) to solve the quantum 
 many-body problem have undergone a tremendous progression. The variety of 
 Hamiltonians being tackled now spans a wide range of fields of physics suc
 h as condensed matter\, quantum chemistry\, and nuclear physics. Much of t
 he progress have come from the development of new network architectures to
  efficiently capture the richness of quantum many-body states. At the same
  time\, there have been recently a renewed interest in revisiting the opti
 mization algorithms employed to adjust the parameters of NQS. Currently\, 
 the main approaches are based on variants of Adam and Stochastic Reconfigu
 ration (SR)\, two algorithms originally designed in the context of supervi
 sed machine learning (ML) and standard Variational Monte-Carlo (VMC)\, res
 pectively.\nIn my talk\, I will discuss our on-going efforts to develop a 
 new optimization algorithm based on Decision Geometry\, namely the Decisio
 nal Gradient Descent (DGD). The goal is to take into account the specifici
 ty of NQS optimization problems from the ground-up\, such as the particula
 r cost function and the large number of parameters which differ from stand
 ard ML and VMC problems. After introducing Decision Geometry and DGD for a
  general Hamiltonian\, I will mainly focus on NQS modelling spin lattices 
 with a J1-J2 interaction as a test bench. Preliminary results will be disc
 ussed comparing the current performance of DGD against the state-of-the-ar
 t SR optimizer in terms of stability and speed of convergence.\n----------
 ----Videoconference link: https://visio.numerique.gouv.fr/nwk-bmff-xny\n 
 \n\nhttps://indico.ijclab.in2p3.fr/event/14066/
LOCATION:100/-1-A900 - Auditorium Joliot Curie (IJCLab)
URL:https://indico.ijclab.in2p3.fr/event/14066/
END:VEVENT
END:VCALENDAR