BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Very persistent random walkers reveal transitions in landscape top
 ology
DTSTART;VALUE=DATE-TIME:20260413T162000Z
DTEND;VALUE=DATE-TIME:20260413T164000Z
DTSTAMP;VALUE=DATE-TIME:20260409T113213Z
UID:indico-contribution-329@fisindico.uniandes.edu.co
DESCRIPTION:Speakers: Jaron Kent-Dobias (ICTP-SAIFR & IFT-UNESP)\nIn large
  random systems\, certain behaviors are reliably predicted\, like the ener
 gy density of the ground state. The long-time behavior of many physical an
 d algorithmic dynamics is likewise predictable\, through DMFT and related 
 approaches. But can these behaviors be connected to static structures of t
 he problem at hand\, like its energy landscape? Recently\, development of 
 the Overlap Gap Property\, which depends on the existence of a system-span
 ning component of the energy level set\, suggests that static topological 
 properties can predict the performance of the best algorithms. Here\, I wi
 ll describe progress towards predicting the performance of the mediocre bu
 t simple algorithms we usually use. We use the ergodicity of a random walk
 er to probe whether typical configurations belong to a system-spanning com
 ponent of the energy level set. Passive random walkers lose ergodicity at 
 a depth associated with the glass transition\, but active random walkers r
 emain ergodic to greater depth. We argue that in the limit of infinite per
 sistence time\, the ergodicity-breaking transition coincides with the poin
 t at which system-spanning components become atypical\, and discuss connec
 tions with gradient descent dynamics.\n\nhttps://fisindico.uniandes.edu.co
 /event/23/contributions/329/
LOCATION:Universidad Nacional de Colombia Ed. 564
URL:https://fisindico.uniandes.edu.co/event/23/contributions/329/
END:VEVENT
END:VCALENDAR