Towards Stellarator Geometry Optimisation for Nuclear Fusion

Sep 8, 2026·
Tobias Weißberg
Moritz Heep
Moritz Heep
,
Zorah Lähner
,
Florian Bernard
· 1 min read
Abstract
We present a local geometry refinement method and a 2d parameterisation for visualisation that helped us beat the ConStellaration leaderboard for stellarator geometry optimisation on the geometric task.
Type
Publication
ECCV Workshop on Geometric Intelligence: From Vision to Scientific Discovery

For additional information, please read the preprint.

Shape Optimisation Nuclear Fusion Geometry Finite Differences
Tobias Weißberg
Authors
Tobias Weißberg
PhD Student in the ‘Learning and Optimisation for Visual Computing’ Group
Moritz Heep
Authors
Moritz Heep
Researcher
Zorah Lähner
Authors
Zorah Lähner
Head of the ‘Geometry in Machine Learning’ Group
Florian Bernard
Authors
Florian Bernard
Head of the ‘Learning and Optimisation for Visual Computing’ Group