BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Towards essentially decentralized interior point methods for distr
 ibuted non-convex optimization
DTSTART:20210924T100000
DTEND:20210924T110000
DTSTAMP:20260407T103306Z
UID:af05eb86b572b29250becf92023b8dff78e54a4a55d3db0fa5603e1b
CATEGORIES:Conferences - Seminars
DESCRIPTION:Alexander Engelmann\nTitle:\nTowards essentially decentralized
  interior point methods for distributed non-convex optimization\n\nSpeaker
 :\nAlexander Engelmann\n\nAbstract:\nDistributed and decentralized optimiz
 ation methods are key in distributed model predictive control\, in distrib
 uted sensing\, and estimation. Non-linear models\, however\, lead to probl
 ems with non-convex constraints for which established distributed and dece
 ntralized algorithms often lack convergence guarantees. Moreover\, decentr
 alized algorithms frequently exhibit rather slow linear convergence rates.
  In this talk we propose an essentially decentralized primal-dual interior
  point method with local convergence guarantees for non-convex problems at
  a superlinear rate. We draw upon different examples from power systems an
 d control illustrating its performance. The numerical results indicate tha
 t the proposed method is able to outperform ADMM in terms of computation t
 ime and it has the potential to overcome difficulties associated with acti
 ve-set detection in the context of distributed optimization.\n\nBio:\nAlex
 ander Engelmann (GSM'18) received the B.Sc. and M.Sc. degrees in electrica
 l engineering and computer science (with distinction) from the Karlsruhe I
 nstitute of Technology\, Karlsruhe\, Germany\, in 2014 and 2016\, respecti
 vely\, where since 2017\, he has been working toward the Ph.D. degree with
  the Optimization and Control group\, Institute for Automation and Applied
  Informatics\, where he is focusing on distributed optimization and optima
 l control for power and multi-energy systems.
LOCATION:MEB10 https://epfl.zoom.us/j/61755570164
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR