BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Gradient Flows of potential energies in the geometry of Sinkhorn d
 ivergences
DTSTART:20250512T080000Z
DTEND:20250512T090000Z
DTSTAMP:20260418T142200Z
UID:indico-event-14164@indico.math.cnrs.fr
DESCRIPTION:Speakers: Hugo Lavenant\n\nWasserstein gradient flows define e
 volution equations in the space of measures which play a fundamental role 
 in PDEs\, probability theory\, and machine learning. But what happens when
  entropic optimal transport is used\, instead of classical optimal transpo
 rt defining the Wasserstein geometry? I will explain why it may be relevan
 t to use Sinkhorn divergences\, built on entropic optimal transport\, as t
 hey allow the regularization parameter to remain fixed. This approach lead
 s to studying the Riemannian geometry induced by Sinkhorn divergences\, wh
 ich retains some characteristics of optimal transport geometry while being
  smoother. The gradient flows of potential energies in this geometry revea
 l intriguing features that I will discuss. This is joint work with Mathis 
 Hardion\, Jonas Luckhardt\, Gilles Mordant\, Bernhard Schmitzer\, and Luca
  Tamanini.\n\nhttps://indico.math.cnrs.fr/event/14164/
LOCATION:0010 (Université Paris-Cité (campus Grands Moulins))
URL:https://indico.math.cnrs.fr/event/14164/
END:VEVENT
END:VCALENDAR