AI RESEARCH

Sparse Autoencoders as a Steering Basis for Phase Synchronization in Graph-Based CFD Surrogates

arXiv CS.LG

ArXi:2604.04946v1 Announce Type: cross Graph-based surrogate models provide fast alternatives to high-fidelity CFD solvers, but their opaque latent spaces and limited controllability restrict use in safety-critical settings. A key failure mode in oscillatory flows is phase drift, where predictions remain qualitatively correct but gradually lose temporal alignment with observations, limiting use in digital twins and closed-loop control. Correcting this through re