arXiv subjects

Robert Jendersie

Publications and source records attributed to Robert Jendersie.

5 recordsLinked to original sources

arXiv2026

A robust and stable hybrid neural network/finite element method for 2D flows that generalizes to different geometries

The deep neural network multigrid solver (DNN-MG) combines a coarse-grid finite element simulation with a deep neural network that corrects the solution on finer grid levels, thereby improving the computational efficiency. In this work, we discuss various design choices for the DNN-MG method and demonstrate significant improvements in accuracy and generalizability when applied to the solution of the nonstationary Navier-Stokes equations. We investigate the stability of the hybrid simulation and show how the neural networks can be made more robust with the help of replay buffers. After an initial single-step training, we run the hybrid simulation for extended periods and compute new reference solutions of the neural network perturbed state for each step. By retraining on this data, the error caused by the neural network over multiple time-steps due to distributional shift can be effectively reduced without the need for a differentiable numerical solver. Furthermore, we compare multiple neural network architectures, including recurrent neural networks and Transformers, and study their ability to utilize more information from an increased temporal and spatial receptive field. Transformers allow us to make use of information from cells outside the predicted patch even with unstructured meshes while maintaining the locality of our approach. This can further improve the accuracy of DNN-MG without a significant impact on performance.

math.NA↗

arXiv2025

An adaptive finite element multigrid solver using GPU acceleration

Adaptive finite elements combined with geometric multigrid solvers are one of the most efficient numerical methods for problems such as the instationary Navier-Stokes equations. Yet despite their efficiency, computations remain expensive and the simulation of, for example, complex flow problems can take many hours or days. GPUs provide an interesting avenue to speed up the calculations due to their very large theoretical performance. However, the large degree of parallelism and non-standard API make the use of GPUs in scientific computing challenging. In this work, we develop a GPU acceleration for the adaptive finite element library Gascoigne and study its effectiveness for different systems of partial differential equations. Our goal is thereby to integrate the GPU acceleration into the existing code with minimal changes, even when this requires a penalty in the GPU acceleration. Through the systematic formulation of all computations as linear algebra operations, we can employ GPU-accelerated linear algebra libraries, which simplifies the implementation and ensures the maintainability of the code while achieving very efficient GPU utilizations. Our results for a transport-diffusion equation, linear elasticity, and the instationary Navier-Stokes equations show substantial speedups of up to 20X compared to multi-core CPU implementations.

math.NA↗

arXiv2024

Towards a GPU-Parallelization of the neXtSIM-DG Dynamical Core

The cryosphere plays a significant role in Earth's climate system. Therefore, an accurate simulation of sea ice is of great importance to improve climate projections. To enable higher resolution simulations, graphics processing units (GPUs) have become increasingly attractive as they offer higher floating point peak performance and better energy efficiency compared to CPUs. However, making use of this theoretical peak performance, which is based on massive data parallelism, usually requires more care and effort in the implementation. In recent years, a number of frameworks have become available that promise to simplify general purpose GPU programming. In this work, we compare multiple such frameworks, including CUDA, SYCL, Kokkos and PyTorch, for the parallelization of \nextsim, a finite-element based dynamical core for sea ice. We evaluate the different approaches according to their usability and performance.

cs.DC↗

arXiv2023

DNN-MG: A Hybrid Neural Network/Finite Element Method with Applications to 3D Simulations of the Navier-Stokes Equations

We extend and analyze the deep neural network multigrid solver (DNN-MG) for the Navier-Stokes equations in three dimensions. The idea of the method is to augment a finite element simulation on coarse grids with fine scale information obtained using deep neural networks. The neural network operates locally on small patches of grid elements. The local approach proves to be highly efficient, since the network can be kept (relatively) small and since it can be applied in parallel on all grid patches. However, the main advantage of the local approach is the inherent generalizability of the method. Since the network only processes data of small sub-areas, it never ``sees'' the global problem and thus does not learn false biases. We describe the method with a focus on the interplay between the finite element method and deep neural networks. Further, we demonstrate with numerical examples the excellent efficiency of the hybrid approach, which allows us to achieve very high accuracy with a coarse grid and thus reduce the computation time by orders of magnitude.

math.NA↗

arXiv2021

Deep neural networks for geometric multigrid methods

We investigate scaling and efficiency of the deep neural network multigrid method (DNN-MG). DNN-MG is a novel neural network-based technique for the simulation of the Navier-Stokes equations that combines an adaptive geometric multigrid solver, i.e. a highly efficient classical solution scheme, with a recurrent neural network with memory. The neural network replaces in DNN-MG one or multiple finest multigrid layers and provides a correction for the classical solve in the next time step. This leads to little degradation in the solution quality while substantially reducing the overall computational costs. At the same time, the use of the multigrid solver at the coarse scales allows for a compact network that is easy to train, generalizes well, and allows for the incorporation of physical constraints. Previous work on DNN-MG focused on the overall scheme and how to enforce divergence freedom in the solution. In this work, we investigate how the network size affects training and solution quality and the overall runtime of the computations. Our results demonstrate that larger networks are able to capture the flow behavior better while requiring only little additional training time. At runtime, the use of the neural network correction can even reduce the computation time compared to a classical multigrid simulation through a faster convergence of the nonlinear solve that is required at every time step.

math.NA↗