BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260903T040937EDT-5426VnzuUM@132.216.98.100
DTSTAMP:20260903T080937Z
DESCRIPTION:Partial Benders Decomposition Strategies for Two-Stage Stochast
 ic Integer Programs\n\nBenders decomposition is a broadly used used exact 
 solution method for stochastic programming\, enabling such programs to be 
 decomposed according to the realizations of the random events that set the
  values of their stochastic parameters. This strategy also comes with impo
 rtant drawbacks\, however\, such as a weak master problem following the re
 laxation step that confines the dual cuts to the scenario sub-problems. We
  propose the first comprehensive Partial Benders Decomposition methodology
  for two- stage integer stochastic program\, based on the idea of includin
 g explicit information from the scenario sub-problems in the master. We pr
 opose two scenario-retention strategies that include a subset of the secon
 d stage scenario variables in the master\, aiming to significantly reduce 
 the number of feasibility cuts generated. We also propose a scenario- crea
 tion strategy to improve the lower-bound provided by the master problem\, 
 as well as three hybrids obtained by combining these pure strategies. We r
 eport the results of an extensive experimental campaign on two-stage stoch
 astic multicommodity network design problems. The analysis clearly shows t
 he proposed methodology yields significant benefits in computation efficie
 ncy\, solution quality\, and stability of the solution process.\n
DTSTART:20181115T203000Z
DTEND:20181115T213000Z
LOCATION:Room PK-5115 \, CA\, Pavillon President-Kennedy
SUMMARY:Walter Rei\, ESG\, UQAM
URL:https://www.mcgill.ca/mathstat/channels/event/walter-rei-esg-uqam-29149
 3
END:VEVENT
END:VCALENDAR