ScaleAware-JEPA combines Constrained Diffusion Decomposition with a scale-tied JEPA objective to learn label-free latent coordinates that recover coherent morphology in multiscale fields such as MHD turbulence and interstellar gas.
Particles and fields in fluid turbulence
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
The understanding of fluid turbulence has considerably progressed in recent years. The application of the methods of statistical mechanics to the description of the motion of fluid particles, i.e. to the Lagrangian dynamics, has led to a new quantitative theory of intermittency in turbulent transport. The first analytical description of anomalous scaling laws in turbulence has been obtained. The underlying physical mechanism reveals the role of statistical integrals of motion in non-equilibrium systems. For turbulent transport, the statistical conservation laws are hidden in the evolution of groups of fluid particles and arise from the competition between the expansion of a group and the change of its geometry. By breaking the scale-invariance symmetry, the statistically conserved quantities lead to the observed anomalous scaling of transported fields. Lagrangian methods also shed new light on some practical issues, such as mixing and turbulent magnetic dynamo.
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
ScaleAware-JEPA: Latent Representation for Discovery in Multiscale Physical Fields
ScaleAware-JEPA combines Constrained Diffusion Decomposition with a scale-tied JEPA objective to learn label-free latent coordinates that recover coherent morphology in multiscale fields such as MHD turbulence and interstellar gas.