Differentiable Analysis Blueprint

Everything that can go wrong [10'+5']

5 Mar 2026, 16:00

15m

4/3-006 - TH Conference Room (CERN)

Conceptual Challenges

Speaker

Frederic Renner (Deutsches Elektronen-Synchrotron (DE))

Description

My journey through pitfalls I discovered building an autodiff workflow for an ATLAS search.

Presentation materials

what_could_go_wrong.pdf