This workshop will be about scientific machine learning surrogates — including neural operator architectures based on transformers, DeepONets, graph networks, and diffusion models — and hybrid models, in which a numerical solver itself becomes a neural network layer. Both classes can learn to solve real physical problems, and their end-to-end differentiability makes them well suited to optimization and inverse problems.

The workshop opens with an introduction to the field, followed by two lightning talks from early-career researchers, and fosters active participation through live tutorials and a structured discussion. We assume familiarity with machine learning in general; no background in scientific machine learning or differentiable simulation is required.

Programme

The session is designed as an interactive tutorial with a short framing talk.

Meet and greet (3-5 min)
Icebreaker warmup where all shake hands with a person in the room.

Landscape framing (10 min)
We set the scene: what Physics-Informed AI is, why it matters for engineering, and how differentiable programming and surrogate modeling fit together. Illustrated with a real-world case study.

Live tutorial (40 min)
Narrated, end-to-end walkthrough of a small but complete engineering problem (1D/2D PDE). The presenter drives the notebook on screen; participants can follow along on a pre-loaded cloud notebook or simply watch.

Steps:
Wrap a physical solver as a reusable Tesseract component → generate training data → train a neural surrogate live → compare solver vs. surrogate → gradient-based inverse design through both → compose them into a hybrid pipeline.

Lightning talk 1 (10 min)
Early-career researcher: applied Physics-Informed AI case study.

Lightning talk 2 (10 min)
Early-career researcher: applied Physics-Informed AI case study.

Structured discussion (15 min)

• Opens with a quick audience poll to surface the room’s composition (simulation practitioners, ML researchers, both).
• The moderator then poses prepared questions e.g., “What existing simulation workflow in your work would benefit most from gradients?” and “What’s blocking you from adopting these methods?” to seed the conversation.
• Closes with a round of concrete next steps: contact exchange, interest in a follow-up meetup, and collaboration ideas.

Speakers’ list

• Nikolas Borrel-Jensen / Dion Hafner, Pasteur Labs: “Physics-Informed AI for real-world engineering: landscape and a case study”
• Matteo Calafà , PhD student: Lightning talk on SciML for acoustics
• Melissa Ulsøe Jessen, Incoming PhD student, DHI / DTU Compute: Lightning talk on SciML for free-surface flows.

Level

Intermediate: For attendees who have basic understanding or some experience with the subject but are not yet advanced.

Organizers

• Lead Organizer: Nikolas Borrel-Jensen,  Simulation Intelligence Engineer, Pasteur Labs, nikolas.borrel@simulation.science
• Dion Hafner, R&D Lead, Pasteur Labs, dion.haefner@simulation.science
• Allan Engsig-Karup, Associate Professor, DTU Compute, apek@dtu.dk
• Melissa Ulsøe Jessen, Incoming PhD student, DHI / DTU Compute. muje@dhigroup.com
• Jakob Tanderup, Research Assistant, DTU Compute , jakob@tanderup.net