{"id":"arxiv/2609.24332","url":"https://arxiv.org/abs/2609.24332","tags":["math.NA","physics.flu-dyn"],"score":0,"title":"Autoencoders vs. Numerical Analysis--Informed Manifold Learning for Navier--Stokes Flows","domain":"math.NA","source":"arxiv","authors":["Alessandro Della Pia","Lucia Russo","Ioannis Kevrekidis","Constantinos Siettos"],"reviews":0,"subject":"math.NA","summary":"Autoencoders (AEs) have become a dominant approach to nonlinear latent-space construction in data-driven reduced-order modelling (ROM), with their decoders lifting latent representations back to the ambient state space. Their prominence, however, has overshadowed an established alternative: manifold-learning methods grounded in classical numerical analysis. We revisit this alternative using Parsimonious Diffusion Maps (PDMs), benchmarking them against Proper Orthogonal Decomposition (POD)-based ROMs and several convolutional AE architectures for the two-dimensional incompressible flow past a…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-21","reactions":{}}