Dissipation enhancing properties for a class of Hamiltonian flows with closed streamlines

We study the evolution of a passive scalar subject to molecular diffusion and advected by an incompressible velocity field on a 2D bounded domain. The velocity field is \(u = \nabla^\perp H\), where H is an autonomous Hamiltonian whose level sets are Jordan curves foliating the domain. We focus on t...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2024-07
Hauptverfasser: Dolce, Michele, Johansson, Carl Johan Peter, Sorella, Massimo
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We study the evolution of a passive scalar subject to molecular diffusion and advected by an incompressible velocity field on a 2D bounded domain. The velocity field is \(u = \nabla^\perp H\), where H is an autonomous Hamiltonian whose level sets are Jordan curves foliating the domain. We focus on the high Péclet number regime (\(Pe := \nu^{-1} \gg 1\)), where two distinct processes unfold on well separated time-scales: streamline averaging and standard diffusion. For a specific class of Hamiltonians with one non-degenerate elliptic point (including perturbed radial flows), we prove exponential convergence of the solution to its streamline average on a subdiffusive time-scale \(T_\nu \ll \nu^{-1}\) ,up to a small correction related to the shape of the streamlines. The time-scale \(T_\nu\) is determined by the behavior of the period function around the elliptic point. To establish this result, we introduce a model problem arising naturally from the difference between the solution and its streamline average. We use pseudospectral estimates to infer decay in the model problem, and, in fact, this analysis extends to a broader class of Hamiltonian flows. Finally, we perform an asymptotic expansion of the full solution, revealing that the leading terms consist of the streamline average and the solution of the model problem.
ISSN:2331-8422