PhysICL
In-Context Learning of Physical Fields on Unstructured Meshes from Procedural Priors
Paper title PhysICL: In-Context Learning of Physical Fields on Unstructured Meshes from Procedural Priors
Paper arXiv 2611.15965PDF
A single model pretrained only on procedurally generated physical tasks predicts fields, with quantiles, directly on each query mesh from a set of labelled simulations, without updating its weights. The same checkpoint accepts a variable number of context examples, scalar and field inputs, any number of output fields, and meshes of varying size and connectivity.
Share on LinkedIn, X or Bluesky, or or .
Know related work that belongs here? Suggest a link
Something wrong on this page? Tell the editor
More papers
All papers- PretrainingUniversal latent spaces — Self-Supervised Learning of Universal Latent Spaces for Operator Learning
- PretrainingCross-domain pretraining — Cross-domain pretraining for steady neural CFD surrogates
- PretrainingGeometry meets physics — Data-efficient pre-training for unstructured neural PDE solvers