BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Tokyo
X-LIC-LOCATION:Asia/Tokyo
BEGIN:STANDARD
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
TZNAME:JST
DTSTART:18871231T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260817T171516Z
LOCATION:Hall C\, C Block\, Level 4
DTSTART;TZID=Asia/Tokyo:20241203T090000
DTEND;TZID=Asia/Tokyo:20241203T120000
UID:siggraphasia_SIGGRAPH Asia 2024_sess223_papers_734@linklings.com
SUMMARY:Trust-Region Eigenvalue Filtering for Projected Newton
DESCRIPTION:Honglin Chen (Columbia University); Hsueh-Ti Derek Liu (Roblox
 , University of British Columbia); Alec Jacobson (University of Toronto, A
 dobe Research); David I.W. Levin (University of Toronto, NVIDIA); and Chan
 gxi Zheng (Columbia University)\n\nWe introduce a novel adaptive eigenvalu
 e filtering strategy to stabilize and accelerate the optimization of Neo-H
 ookean energy and its variants under the Projected Newton framework. For t
 he first time, we show that Newton’s method, Projected Newton with eigenva
 lue clamping and Projected Newton with absolute eigenvalue filtering can b
 e unified using ideas from the\ngeneralized trust region method. Based on 
 the trust-region fit, our model adaptively chooses the correct eigenvalue 
 filtering strategy to apply during the optimization. Our method is simple 
 but effective, requiring only two lines of code change in the existing Pro
 jected Newton framework. We validate our model outperforms stand-alone var
 iants across a number of experiments\non quasistatic simulation of deforma
 ble solids over a large dataset.\n\nRegistration Category: Enhanced Access
 , Full Access, Full Access Supporter\n\nLanguage Format: English Language\
 n\n
URL:https://asia.siggraph.org/2024/program/?id=papers_734&sess=sess141
END:VEVENT
END:VCALENDAR