At realistic LES filter ratios on channel flow, an octahedral-equivariant nonlocal CNN is more accurate, parameter-efficient, and data-efficient than an augmented non-equivariant CNN, while pointwise models fail to beat the Clark baseline.
A recursive neural-network-based subgrid-scale model for large eddy simulation: application to homogeneous isotropic turbulence , volume=
1 Pith paper cite this work, alongside 11 external citations. Polarity classification is still indexing.
1
Pith paper citing it
11
external citations · OpenAlex
fields
physics.flu-dyn 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Rotational equivariance and locality in data-driven subgrid-scale closures
At realistic LES filter ratios on channel flow, an octahedral-equivariant nonlocal CNN is more accurate, parameter-efficient, and data-efficient than an augmented non-equivariant CNN, while pointwise models fail to beat the Clark baseline.