BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Drupal iCal API//EN
X-WR-CALNAME:Events items teaser
X-WR-TIMEZONE:America/Toronto
BEGIN:VTIMEZONE
TZID:America/Toronto
X-LIC-LOCATION:America/Toronto
BEGIN:DAYLIGHT
TZNAME:EDT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
DTSTART:20250309T070000
END:DAYLIGHT
BEGIN:STANDARD
TZNAME:EST
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
DTSTART:20241103T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:69d1feed78eb9
DTSTART;TZID=America/Toronto:20250328T153000
SEQUENCE:0
TRANSP:TRANSPARENT
DTEND;TZID=America/Toronto:20250328T163000
URL:https://uwaterloo.ca/combinatorics-and-optimization/events/tutte-colloq
 uium-arnesh-sujanani
SUMMARY:Tutte colloquium-Arnesh Sujanani
CLASS:PUBLIC
DESCRIPTION:TITLE:The Inexact Augmented Lagrangian Method: Optimal Complexi
 ty\nBounds and Applications to Solving Huge SDPs\n\nSPEAKER:\n Arnesh Suja
 nani\n\nAFFILIATION:\n University of Waterloo\n\nLOCATION:\n MC 5501\n\nAB
 STRACT:In the first part of this talk\, we present optimal and\nnearly-opt
 imal parameter-free augmented Lagrangian (AL) methods for\nconvex and stro
 ngly optimization with linear constraints. Our AL\nmethods employ tractabl
 e inexact criteria for solving their inner\nsubproblems\, which accelerate
 d methods can be shown to achieve in a\nfinite number of iterations that d
 epends on the target accuracy. \n\nIn the second part of this talk\, we pr
 esent a new inexact augmented\nLagrangian method\, namely\, HALLaR\, that 
 employs a Burer-Monteiro style\nlow-rank factorization for solving large-s
 cale semidefinite programs\n(SDPs). The AL subproblems are solved by a hyb
 rid low-rank method\,\nwhich is based on a combination of a low-rank Frank
 -Wolfe method and a\nnonconvex accelerated inexact proximal point method. 
 In contrast to\nthe classical low-rank method by Burer and Monteiro\, HALL
 aR finds a\nnear-optimal solution (with provable complexity bounds) of SDP
 \ninstances satisfying strong duality. Computational results comparing\nHA
 LLaR to state-of-the-art solvers on several large SDP instances show\nthat
  the former finds higher accurate solutions in substantially less\nCPU tim
 e than the latter ones. For example\, in less than 20 minutes\,\nHALLaR ca
 n solve (on a personal laptop) a maximum stable set SDP with\n1 million ve
 rtices and 10 million edges within 1e-5 relative accuracy.\n\nThis talk is
  based on joint work with Saeed Ghadimi and Henry\nWolkowicz from Universi
 ty of Waterloo and Diego Cifuentes and Renato\nMonteiro from Georgia Tech.
 \n\n \n\n 
DTSTAMP:20260405T061925Z
END:VEVENT
END:VCALENDAR