Guest lectures and seminars - Page 53

C*-algebra seminar talk by Magnus Goffeng

Abstract: In computational mechanics, high fidelity simulations of a parameterized partial differential equation (PDE) are often computational expensive, which make them impractical for real-time predictions. Non-intrusive reduced order modelling aims to address this problem with a fast low rank approximation. This is usually done in two phases: the model is built in the offline phase and the prediction is done in the online phase. In the offline phase, data points, or so-called snapshots, are collected from simulations or measurements. The reduced basis space can then be obtained from the dataset using Proper Orthogonal Decomposition. In the online phase, the solution for a new set of parameters is obtained by first recovering the expansion coefficients for the reduced basis and then projecting them back into the uncompressed real-life space. The non-intrusive approach relies on a statistical mapping between the coefficients and the parameters. Various methods have been proposed to do so, this seminar will discuss radial basis function interpolations and dynamic mode decompositions.

This talk is part of the Mechanics Lunch Seminar series. That means 20min talks plus discussion in an informal setting.

Zoom: To obtain the Zoom meeting details please contact Timo Koch (timokoch at math.uio.no).

C*-algebra seminar talk by Ulrich Pennig

Abstract: Graphics processing units, or GPUs, offer significantly increased performance for some scientific computing workloads. But in the case of finite element simulations on unstructured meshes, the benefits of using GPUs are still the subject of an ongoing discussion for which there is no clear conclusion. We describe our work on improving the GPU acceleration of a finite element solver framework called FEniCS, where code is automatically generated for the user from a high-level description of their finite element problem. We use automated code generation to offload the assembly of linear systems to a GPU, while taking care that data transfers between CPU and GPU do not become a performance bottleneck. We provide examples to show that GPUs and automated code generation can be used to accelerate finite element solvers. Even though more work is needed to find efficient GPU-based linear solvers, our improvements to FEniCS can be used as a starting point for exploring the potential of GPU acceleration for finite element simulations.

This talk is part of the Mechanics Lunch Seminar series. That means 20min talks plus discussion in an informal setting.

Zoom: To obtain the Zoom meeting details please contact Timo Koch (timokoch at math.uio.no).