Student Jan Hagnberger Awarded Wübben Foundation Student Grant

October 25, 2024

Jan Hagnberger, currently pursuing his Master’s degree in Artificial Intelligence and Data Science at the University of Stuttgart, has been awarded the Wübben Foundation Student Grant, following a nomination by SimTech spokesperson Steffen Staab and SimTech PR Mathias Niepert.

Hagnberger recently gained attention for his bachelor’s thesis, titled Vectorized Conditional Neural Fields: A Framework for Solving Time-dependent Parametric Partial Differential Equations. This work was published as a full peer-reviewed paper at the renowned International Conference on Machine Learning (ICML), one of the top conferences in the field of AI and machine learning.

Jan Hagnberger's research focuses on using machine learning to solve an important challenge: partial differential equations (PDEs). These equations are crucial for simulations in fields like physics, climate modeling, and engineering. Traditional methods for solving PDEs can be very computationally expensive and have limitations.

In his bachelor's thesis, Hagnberger introduced a new framework called "Vectorized Conditional Neural Fields" (VCNeFs). This framework aims to overcome the problems of existing machine learning models, like transformers, which are often slow and struggle to efficiently solve PDEs.

Hagnberger's VCNeFs framework offers several key advantages:

  1. It can generalize to new parameters of PDEs (so it’s flexible),
  2. It can compute solutions for both space and time in parallel (which makes it much faster),
  3. It supports solving PDEs in 1D, 2D, and 3D spaces,
  4. It has the ability to do "zero-shot super-resolution," meaning it can generate higher-resolution solutions in both space and time without needing extra training data.

These improvements make VCNeFs a powerful tool for advancing scientific simulations, offering more accurate and efficient solutions to complex, time-dependent problems.

Mathias Niepert, a SimTech researcher and one of the directors of the Institute for Artificial Intelligence, heads the Machine Learning for Simulation Science group. His team’s research focuses on core areas of machine learning, such as graph neural networks, geometric deep learning, and the integration of discrete algorithmic methods with continuous learning systems - all relevant to the intersection of machine learning and simulation sciences.

The Wübben Stiftung Wissenschaft, a private foundation dedicated to promoting scientific excellence in Germany, awards grants to outstanding students in all fields who demonstrate bold and innovative approaches to solving significant research challenges. Hagnberger is one of the recipients of the 2024 grant and will receive €550 per month for one year in recognition of his academic work.

Find more about the publication here.

To the top of the page