LLNL Computer Scientist Kenneth Weiss and collaborators from the University of Colorado and Purdue University were recognized with one of five Technical Best Papers Awards at the international Association for Computing Machinery (ACM)'s Special Interest Group on Computer Graphics and Interactive Techniques (SIGGRAPH) conference for their innovative work implementing curved geometries into computing simulations.

Weiss has been studying the topic of initializing multimaterial simulations for the past eight years, where the main problem is to determine whether an arbitrary point is enclosed within a given geometric region of space. Existing approaches were fast and robust but did not support input with self-intersections or gaps in the geometric shapes bounding each material region. Weiss and his co-authors also found that the straight-edged segments produced by linearizing the curved shapes introduced geometric errors into the simulations.

These issues, along with the team’s idea to operate directly on curved geometry, were the driving force behind the development of “Robust Containment Queries Over Collections of Rational Parametric Curves via Generalized Winding Numbers.”

“We reframed the point containment problem over a collection of curves in terms of simpler winding number computations over the closure of each curve,” Weiss said. “So, that gave us a robust and computationally efficient way of determining the exact winding number for complicated curves.”

This approach allows multiple simulation runs to be set up over the same computational mesh, allowing users to more easily experiment with different input shapes. Weiss’ work is vital for LLNL’s MARBL project, a GPU-accelerated multiphysics application for simulating high-energy density physics experimental platforms.

In Computing’s Applications, Simulations & Quality Division, Weiss leads the Numerics, Modular Applications and Performance Group. He is also the computer science lead for the Multiphysics on Advanced Platforms Project (MAPP) and the deputy lead for Axom, a computer science infrastructure project for high performance computing.

In their own words, Weiss, Jacob Spainhour (Colorado), and David Gunderman (Purdue) write on the SIGGRAPH blog, “We extend the theory of generalized winding numbers to unstructured collections of rational parametric curves with a numerically stable algorithm, thereby allowing for robust and accurate containment classifications at arbitrary locations for non-watertight and self-intersecting shapes.”

Spainhour and Weiss will present their winning paper as part of the conference’s technical papers program. “I am very proud of the work we’ve done together, and it is really nice to see it recognized so publicly,” Weiss said.

—Deanna Willis