Predicting uncertainty in geometric fluid mechanics

Existence of optimal solutions to problems involving randomness (49J55) Variational principles of physics (49S05) Hamilton's principle (70H25) Variational methods applied to problems in fluid mechanics (76M30)

Abstract: We review opportunities for stochastic geometric mechanics to incorporate observed data into variational principles, in order to derive data-driven nonlinear dynamical models of effects on the variability of computationally resolvable scales of fluid motion, due to unresolvable, small, rapid scales of fluid motion.

Review text

The authors analyze applications of geometric mechanics and especially stochastic variational principles in stochastic fluid dynamics. They introduce the group \(G=Diff_{vol}(\mathcal{D})\) of volume preserving diffeomorphisms of a fluid domain \(\mathcal{D}\), a bounded domain in \( \mathbb{R}^{2}\) or \(\mathbb{R}^{3}\), as the configuration manifold for incompressible fluids. Formally, \(G\) can be considered as a Lie group, whose Lie algebra denoted by \(\mathfrak{g}\) is the space of divergence-free vector fields on \(\mathcal{D}\) parallel to the boundary \(\partial \mathcal{D}\), endowed with the Lie bracket \([u;v]=v\cdot \nabla u-u\cdot \nabla v\). In the Lagrangian description, the equations of motion are given by Hamilton's principle, written as \(\delta \int_{0}^{T}L(g_{t},\overset{.}{g}_{t})\,dt=0\), for all variations of the curve \(g_{t}\) with fixed extremities, or using the Hamilton-Pontryagin principle, as \(\delta \int_{0}^{T}[L(g_{t},v)+\left\langle \pi ,\overset{.}{g}_{t}-v\right\rangle ]dt=0\), for all variations \(\delta g_{t}\), \(\delta v\), \(\delta \pi \), with \( \delta g_{t}\) vanishing at \(t=0,T\). The authors recall the stochastic Hamilton-Clebsch variational principle formally written as \(\delta \int_{0}^{T}[l(u)\,dt+\left\langle p,dq+{\mathfrak{L}}_{dx_{t}}q\right\rangle _{V}]=0 \), with respect to variations \(\delta u\), \(\delta q\), \(\delta p\), for \( \delta q\) vanishing at \(t=0,T\). Here \(l\) is the reduced Lagrangian and the authors assume that \(G\) acts on the right on a vector space \(V\), usually given by a space of tensor fields on \(\mathcal{D}\), \({\mathfrak{L}} _{u}q\) is the infinitesimal generator of this action, for \(u\in \mathfrak{g}\) and \( dx_{t}=dg_{t}X=u(g_{t}X,t)dt+\sum_{i=1}^{N}\xi _{i}(g_{t}X)\circ dW_{t}^{i}\). They derive the Euler-Poincaré equations. They illustrate these results in the cases of 2D and 3D Euler's fluid equations. They then recall the stochastic Hamilton-Pontryagin variational principle written as: \(\delta \int_{0}^{T}[L(g_{t},v)\,dt+\left\langle \pi ,dg_{t}-vdt-\sum_{i=1}^{N}\xi _{i}g_{t}\circ dW_{i}(t)\right\rangle ]=0\), with respect to variations \( \delta g_{t}\), \(\delta v\), \(\delta \pi \), for \(\delta g_{t}\) vanishing at \( t=0,T\). They write this principle in Eulerian formulation and the stationarity conditions. They derive stochastic Hamiltonian formulations. In the last part of their paper, the authors apply these tools to two cases: a deterministic \(N\)-layer quasigeostrophic fluid and a stochastic \(N\)-layer quasigeostrophic fluid for which they write the Hamiltonian formulation.

Recommendations

Cites work

Cited in

(11)

This page was built for publication: Predicting uncertainty in geometric fluid mechanics

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2180365)