BEGIN:VCALENDAR CALSCALE:GREGORIAN VERSION:2.0 X-WR-CALNAME:CIGMO Seminar Series - Ioannis P. A. Papadopoulos METHOD:PUBLISH PRODID:-//Apple Inc.//macOS 26.6.2//EN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT TZOFFSETFROM:-0500 DTSTART:20070311T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU TZNAME:EDT TZOFFSETTO:-0400 END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0400 DTSTART:20071104T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU TZNAME:EST TZOFFSETTO:-0500 END:STANDARD END:VTIMEZONE BEGIN:VEVENT CREATED:20260924T210352Z DTEND;TZID=America/New_York:20261005T120000 STATUS:CONFIRMED TRANSP:OPAQUE UID:706fe9qm5odrda2jtn48erdt9u@google.com DTSTART;TZID=America/New_York:20261005T110000 LAST-MODIFIED:20260925T021128Z DTSTAMP:20260925T021216Z SUMMARY:CIGMO Seminar Series - Ioannis P. A. Papadopoulos SEQUENCE:0 DESCRIPTION:Speaker: Ioannis P. A. Papadopoulos\nTitle: Mesh-dependent i teration count growth in primal-dual active set strategies\n\nAbstract: Quadratic programming problems often arise in mathematical models\, and primal-dual active set (PDAS) strategies provide a highly effective algo rithm for their solution. For finite-dimensional problems\, an equivalen ce between PDAS and semismooth Newton methods was established in 2002\, explaining the fast local superlinear convergence observed for these met hods. This connection has also been shown to hold for certain infinite-d imensional problems\, such as optimal control problems with control cons traints.\n\nOn the other hand\, PDAS is not a semismooth Newton method w hen applied to the infinite-dimensional obstacle problem. It is neverthe less a semismooth Newton method for any finite-dimensional discretizatio n\, and local superlinear convergence is therefore observed in practice. But there is a catch: with each uniform refinement of the mesh\, the nu mber of iterations required for convergence grows dramatically\, eventua lly doubling with each refinement.\n\nThis phenomenon has been occasiona lly noted in the literature but never fully investigated. In this talk\, we provide a detailed explanation of why this mesh-dependent behaviour occurs. We also investigate PDAS applied to the Signorini problem\, wher e the convergence behaviour is also mesh-dependent but fundamentally dif ferent from that of the obstacle problem.\n\nPreprint: https://arxiv.org /abs/2607.26622.\n\n-::~:~::~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~: ~:~:~:~:~:~:~:~:~:~:~:~:~:~::~:~::-\nJoin Zoom Meeting\nhttps://brown.zo om.us/j/9011251684?pwd=5mZNPYbleyTlsQb6hxqD3oPdqcehaS.1&omn=98211733179& jst=2 (ID: 9011251684\, passcode: 171236)\n\nJoin by phone\n(US) +1 877- 853-5247 (passcode: 171236)\n\nJoin using SIP\n9011251684@zoomcrc.com (p asscode: 171236)\n\nJoining instructions: https://www.google.com/url?q=h ttps://applications.zoom.us/addon/invitation/detail?meetingUuid%3DHVuIHI p8RN%252Baxtk9qoE9GA%253D%253D%26signature%3D9871784a0641ed0b6f670a41aa5 a9e499f4272cf8a5e54017fa493e06dcfc525%26v%3D1&sa=D&source=calendar&usg=A OvVaw029aER9K7_td4hnVEJDozi\n\nMeeting host: dohyun_kim@brown.edu

Join Zoom Meeting:
https://brown.zoom.us/j/9011251684?pwd=5mZNP YbleyTlsQb6hxqD3oPdqcehaS.1&\;omn=98211733179&\;jst=2\n\nPleas e do not edit this section.\n-::~:~::~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~ :~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~::~:~::- END:VEVENT END:VCALENDAR