{"id":"arxiv/2603.05325","url":"https://arxiv.org/abs/2603.05325","tags":["math.NA","physics.flu-dyn"],"score":0,"title":"Approaching the optimal closure: equivariance, inductive bias, and Reynolds-number generalization in data-driven LES","domain":"math.NA","source":"arxiv","authors":["Syver Døving Agdestein","Benjamin Sanderse"],"reviews":0,"subject":"math.NA","summary":"Data-driven closures for large-eddy simulation (LES) are commonly built to respect the symmetries of the Navier--Stokes equations, on the premise that this improves accuracy and generalization. We test this premise in a controlled comparison of three data-driven LES closures that share a pointwise, Galilean-invariant velocity-gradient construction but span non-equivariant, octahedral-equivariant, and tensor-basis designs: an unconstrained multi-layer perceptron (MLP), a group-convolutional network whose exactly equivariant weights we synthesize in closed form, and a tensor-basis neural…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-03-05","reactions":{}}