Cornell University
SOLVING INVERSE TRANSPORT PROBLEMS USING DIFFERENTIABLE SIMULATIONS
Abstract
dc:description.abstractThe reconstruction of real-world phenomena represents a crucial and challenging problem with broad relevance to both entertainment and scientific applications. This thesis addresses this by studying inverse problems within transport theory, which includes both radiative transfer and neutron transport, where the field's energy distribution is described by transport equations. A central challenge involves accurately determining the parameters of these equations from empirical or target radiation data, such as reflectance and transmittance. My thesis details three distinct works that contribute to the inversion of transport theory and its applications in reconstruction and design optimization across various disciplines. The initial work introduced a software pipeline designed to reconstruct the geometry and optical properties of semi-translucent objects from photographic input. Building upon this, the second work proposed the integration of hardware and software to facilitate the reconstruction of translucent thin-layered objects from photographs. The third work develops a theoretical framework and a gradient-based method for the shielding design of nuclear reactors using differentiable neutron transport simulation.
Degree
thesis:*- Name thesis:degree_name
- Ph. D., Computer Science
- Level thesis:degree_level
- Doctor of Philosophy
- Discipline thesis:degree_discipline
- Computer Science
- Grantor
- Cornell University
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Deng, Xi
- Chair dc:contributor.chair
-
- Marschner, Stephen
- Committee members dc:contributor.committeemember
-
- Snavely, Keith
- Arias, Tomas
Subjects
dc:subject × 6Rights
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Bibid
- 17423266
- Dc Identifier Other
-
ProQuest Submission ID: 15311
ProQuest Publication ID: 32278007