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:20250110T023309Z
LOCATION:Hall B5 (1)\, B Block\, Level 5
DTSTART;TZID=Asia/Tokyo:20241203T150800
DTEND;TZID=Asia/Tokyo:20241203T151900
UID:siggraphasia_SIGGRAPH Asia 2024_sess106_papers_1139@linklings.com
SUMMARY:Solving Inverse PDE Problems using Grid-Free Monte Carlo Estimator
 s
DESCRIPTION:Technical Papers\n\nEkrem Fatih Yilmazer (EPFL), Delio Vicini 
 (Google Inc.), and Wenzel Jakob (EPFL)\n\nPartial differential equations c
 an model diverse physical phenomena including heat diffusion, incompressib
 le flows, and electrostatic potentials. Given a description of an object's
  boundary and interior, traditional methods solve such PDEs by densely mes
 hing the interior and then solving a large and sparse linear system derive
 d from this mesh. Recent grid-free solvers take an alternative approach an
 d avoid this complexity in exchange for randomness: they compute stochasti
 c solution estimates and generally bear a striking resemblance to physical
 ly-based rendering algorithms.\n\nIn this article, we develop algorithms t
 argeting the inverse form of this problem: given an already existing solut
 ion of a PDE, we infer parameters characterizing the boundary and interior
 . In the grid-free setting, there are again significant connections to ren
 dering, and we show how insights from both fields can be combined to compu
 te unbiased derivative estimates that enable gradient-based optimization. 
 In this process, we encounter new challenges that must be addressed to obt
 ain practical solutions. We introduce acceleration and variance reduction 
 strategies and show how to differentiate branching random walks in reverse
  mode.\n\nWe finally demonstrate our approach on both simulated data and a
  real-world electrical impedance tomography experiment, where we reconstru
 ct the position of a conducting object from voltage measurements taken in 
 a saline-filled tank.\n\nRegistration Category: Full Access, Full Access S
 upporter\n\nLanguage Format: English Language\n\nSession Chair: Yonghao Yu
 e (Aoyama Gakuin University)
URL:https://asia.siggraph.org/2024/program/?id=papers_1139&sess=sess106
END:VEVENT
END:VCALENDAR