BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260114T163648Z
LOCATION:Meeting Room C4.9+C4.10\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231212T154500
DTEND;TZID=Australia/Melbourne:20231212T160000
UID:siggraphasia_SIGGRAPH Asia 2023_sess161_papers_238@linklings.com
SUMMARY:Neural Metamaterial Networks for Nonlinear Material Design
DESCRIPTION:Yue Li, Stelian Coros, and Bernhard Thomaszewski (ETH Zürich)\
 n\nNonlinear metamaterials with tailored mechanical properties have applic
 ations in engineering, medicine, robotics, and beyond. While modeling\nthe
 ir macromechanical behavior is challenging in itself, finding structure\np
 arameters that lead to an ideal approximation of high-level performance go
 als\nis a daunting task. In this work, we propose Neural Metamaterial Netw
 orks\n(NMN)—smooth neural representations that encode the nonlinear mechan
 ics of entire metamaterial families. Given structure parameters as input,\
 nNMN return continuously differentiable strain energy density functions,\n
 thus guaranteeing conservative forces by construction. Though trained on\n
 simulation data, NMN do not inherit the discontinuities resulting from top
 ological changes in finite element meshes. They instead provide a smooth\n
 map from parameter to performance space that is fully differentiable and\n
 thus well-suited for gradient-based optimization. On this basis, we formul
 ate\ninverse material design as a nonlinear programming problem that lever
 ages\nneural networks for both objective functions and constraints. We use
  this\napproach to automatically design materials with desired strain-stre
 ss curves,\nprescribed directional stiffness, and Poisson ratio profiles. 
 We furthermore\nconduct ablation studies on network nonlinearities and sho
 w the advantages\nof our approach compared to native-scale optimization.\n
 \nRegistration Category: Full Access\n\nSession Chair: J. Andreas Bærentze
 n (Technical University of Denmark)\n\n
URL:https://asia.siggraph.org/2023/full-program?id=papers_238&sess=sess161
END:VEVENT
END:VCALENDAR