Reproduce “VoroDuce: Massively Parallel 3D Voronoi Simplification with Volume-Aware Edge Collapse” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Felix Windisch, Lukas Radl, Thomas Köhler, Shrisudhan Govindarajan, Marcel Pur, Andrea Tagliasacchi, Markus Steinberger

VoroDuce simplifies volumetric fields represented by Voronoi cells by greedily collapsing neighboring sites according to estimated squared error. Unlike a midpoint edge-collapse baseline, it accounts for changes in neighboring cells’ regions and values; its Fast variant estimates swept volumes with truncated prismatoids. On the CT Head scan, reducing 250,000 cells to 37,500 yielded MSE 0.067 versus Ward’s 0.306 here.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.