Rearrangements of functions, saddle points and uncountable families of steady configurations for a vortex (Q910302): Difference between revisions

Kelvin studied steady planar flows of an ideal fluid, confined in a bounded region with solid boundaries. He supposed the vorticity to have a given value in a region of fixed area, outside which the flow was irrotational, and considered possible configurations of the vortex. In the case of a circular region, he observed that infinitely many steady configurations exist, and one can easily verify that any radially symmetric configuration is steady. Less obviously, he claimed that in a dumb-bell shaped region, infinitely many steady configurations can be obtained by dividing the vorticity between the ends of the dumb-bell in an arbitrary proportion, and seeking the configuration that maximises the kinetic energy subject to this restriction. In this paper we prove two results on the existence of infinitely many steady configurations for a vortex, based on Kelvin's principle of stationary kinetic energy. We admit flows in which the vorticity may be nonconstant in the vortex core (the region of non-zero vorticity), and seek steady flows in which the vorticity is a rearrangement of a given function; this more general formulation is based on a theory of 3- dimensional vortex rings; additionally, we prescribe the circulations of the velocity around the connected components of the boundary. We first show that in a bounded planar region of arbitrary shape, for a given non- negative function \(\zeta_ 0\), the kinetic energies of steady ideal fluid flows whose vorticities are rearrangements of \(\zeta_ 0\), and having the prescribed circulations, realise all values between a maximum and a minimum. This result is proved using a saddle-point theorem for a convex functional relative to a set having convex sections. Our second result applies to a simply connected region comprising two cavities connected by a constricted aperture, generalising Kelvin's dumb- bell. We prove the existence of steady flows for which the vorticity vanishes in the aperture, and for which the restriction of the vorticity to each cavity is a rearrangement of a prescribed non-negative function. This is proved by a maximisation argument, and yields a disconnected vortex core that avoids the boundary of the region. By assigning varying proportions of the vorticity to the two cavities, uncountably many steady flows are obtained, all of whose vorticities are rearrangements of one given function.