BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20181028T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20190331T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.18153.field_data.0@www.diag.uniroma1.it
DTSTAMP:20260415T191642Z
CREATED:20190322T095959Z
DESCRIPTION:In multiobjective optimization\, one considers optimization pro
 blems with several competing objective functions. For instance\, in engine
 ering problems a design often has to be stable and light weighted at the s
 ame time.  A classical approach to such optimization problems is to formul
 ate suitable parameter-dependent single-objective replacement problems\, c
 alled scalarizations\, such as considering a weighted sum of the objective
  functions. Then the parameters are varied and the scalarized problems are
  solved iteratively.  This talk is about numerical methods which do avoid 
 such scalarizations to improve the performance of the procedure.The first 
 algorithm which we present in this talk is for multiobjective optimization
  problems with non-convex objective functions.   Then\, methods of global 
 optimization are necessary to solve the replacement problems. Instead of t
 his detour via scalarization\, we presents a direct deterministic method f
 or finding a representation of all globally optimal solutions. This branch
 -and-bound method is based on a subdivision of the feasible set and the co
 nsideration of convex underestimators of the objective functions for the d
 etermination of lower bounds.The second algorithm is for so called heterog
 eneous multiobjective optimization problems\, i.e. problems\, where one of
  the functions is assumed to be an expensive black-box function while the 
 other objectives are given analytically. The proposed method uses the basi
 c trust region approach by restricting the computations in every iteration
  to a local area. The objective functions are replaced by suitable models 
 which reflect the heterogeneity of the functions. Convergence results as w
 ell as numerical experiments are presented.
DTSTART;TZID=Europe/Paris:20190325T100000
DTEND;TZID=Europe/Paris:20190325T100000
LAST-MODIFIED:20200515T073055Z
LOCATION:Aula A4 - DIAG
SUMMARY:Handling non-convex or expensive objectives: algorithms for multiob
 jective optimization without scalarization - Prof. Gabriele Eichfelder
URL;TYPE=URI:https://www.diag.uniroma1.it/node/18153
END:VEVENT
END:VCALENDAR