PHLieNets paper published in Mechanical Systems and Signal Processing

Our paper Beyond Static Models: Hypernetworks for Adaptive and Generalizable Forecasting in Complex Parametric Dynamical Systems, with Konstantinos Vlachas and Eleni Chatzi, is now published open access in Mechanical Systems and Signal Processing (Volume 258, 2026). A preprint is on arXiv.

The paper introduces PHLieNets, hypernetworks that learn an embedding of the system parameters and generate the weights of a forecasting network, so that a single model covers an entire family of parametric dynamical systems and generalizes to regimes unseen during training.