SciML Frontiers: Progress, Evolution, and Future Directions — Workshop 2026

SciML Frontiers 2026 logo
22–23 September 2026 • Augsburg, Germany

Scientific Machine Learning (SciML), the fusion of scientific computing and machine learning algorithms, has proven to be a transformative force in academia and is rapidly establishing a firm foothold in industrial applications. The foundational 2018 SciML workshop outlined six priority research directions (PRDs) for the nascent field of SciML, anchoring future development around critical pillars such as domain awareness, robustness, and interpretability. The SciML landscape has evolved profoundly over the past seven years. Considering these advancements, this workshop, hosted by the Centre for Advanced Analytics and Predictive Sciences at the University of Augsburg, convenes researchers from diverse disciplines to re-evaluate the original priorities within the context of current technological frontiers. Through expert keynotes, technical presentations, and collaborative breakout sessions, participants will share practical insights and critically analyse current progress and emerging challenges within the field. Our primary objective is to assess the trajectory of the original PRDs and collaboratively shape the future research agenda for SciML. The insights generated during this two-day event will be synthesised into a comprehensive scientific article, targeting high-impact journals such as Nature Machine Intelligence or Neurocomputing.

We have synthesised the original six PRDs into four focused research directions, which participants will explore deeply during the breakout sessions.

A smooth continuous curve approximated by a discrete polyline
Numerical Foundation of SciML
Discretisation, Stability and Convergence
Gradient descent towards the minimum of an objective function
Beyond prediction
Control, Optimisation and Application
Two interlocking gears representing automated, scalable workflows
Automation and Scalability in SciML
From Prototype to Production Scale
A magnifying glass revealing hidden structure in the data
Domain Awareness and Interpretability
Physical Priors, Explainability and Trust

The event is organised by Prof. Lars Mikelsons, Prof. Michael Schlottke-Lakemper and Andreas Hofmann (see Organisers), and is hosted by the Centre for Advanced Analytics and Predictive Sciences at the University of Augsburg.

University of Augsburg logo