Speaker: Matthias Zezulka
Abstract
This master’s thesis builds on our previous papers on simulating thermal radiation by precomputing a radiative transport operator using GPU-accelerated photon tracing. We subsequently run a finite element method (FEM) solver to compute the resulting surface temperatures. However, this transport operator grows quadratically with the mesh size, significantly limiting its applicability to large scenes or intricate geometries.
To address these challenges, we envision a novel approach for handling the transport operator and efficiently simulating surface temperatures using an auxiliary grid data structure. This structure aggregates thermal transport data from detailed mesh geometry, enabling more efficient computations. By combining fast GPU-accelerated photon tracing with the robustness and controllability of regular grid data structures, we aim to simulate radiative heating and cooling for arbitrary input geometries at interactive rates.