Speaker: Christopher Brandt, Computer Graphics and Visualization, TU Delft


Thursday November 1st, 10:30am, in BC 229
Title: Model Reduction Approaches for Geometry Processing

Abstract:
We will look at two applications in Geometry Processing: tangential (n-)vector-field design and the simulation of elastic deformable objects. On these examples, I will demonstrate how model reduction methods can be applied to achieve interactive rates, independently of the underlying mesh resolution. For tangent field design, a dimensionality reduction that uses the spectrum of a discrete Hodge-Laplace operator will be enough to achieve this goal. For elasticity simulations, an approximation for the non-linearities in the system is required in addition to the subspace.

Bio:
Christopher Brandt studied Mathematics at FU Berlin, with a focus on Dynamical System and Discrete Mathematics. He started his PhD studies at the Max-Planck-Institut for Informatics in Saarbrücken and is now finishing them at the Computer Graphics Group of the Delft University of Technology, under supervision of Dr. Klaus Hildebrandt and Prof. Elmar Eisemann. There, he studies various problems in Geometry Processing, from optimal splines, to tangential vector fields or elasticity problems. A focus therein is the application of model order reduction approaches to obtain interactive tools for simulation, modelling and design.