High-Performance Mesh Generation for Scientific Computing and Graphics

This workshop introduces core methods for generating high-quality meshes used in large-scale simulation, scientific computing, and computer graphics. We will explore how Voronoi tessellation and Delaunay triangulation form the foundation of modern meshing algorithms, how parallel computing enables scalable geometry processing, and how emerging learning-based approaches can learn to generate meshes directly from data.

The session will demonstrate several open-source tools, including:

• Voro++ — a multi-threaded library for scalable Voronoi diagram computation,
• TriMe++ — a high-performance library for fast mesh generation, and
• VoroLight — a lightweight learning-based framework for producing Voronoi meshes from general inputs.

Participants will gain both theoretical insight and practical experience with modern meshing pipelines.