A blob of vorticity in a two-dimensional incompressible flow starts compact and spreads. How fast does it spread? In free space, the answer depends on whether viscosity is present (Navier-Stokes, where diffusion drives the spreading) or absent (Euler, where the inviscid dynamics alone must account for any growth).
The paper works in an infinite cylinder — a two-dimensional strip, periodic in one direction, infinite in the other. For non-negative, compactly supported initial vorticity, the vorticity support's diameter grows at most like (t log t)^(1/3) in the Euler case. This refines the previous best bound of t^(1/3) log²(t) — the logarithmic factor is reduced from squared to linear.
The improvement comes from an iterative scheme that exploits the antisymmetry of the Biot-Savart kernel in the cylinder geometry. The kernel maps vorticity to velocity; its antisymmetry ensures that the induced velocity field has cancellations that a generic kernel would not. These cancellations limit how efficiently the flow can transport vorticity outward, which is what the confinement bound captures.
The result quantifies a qualitative fact: vorticity in an incompressible inviscid flow wants to stay together. The flow field that the vorticity generates tends to recirculate rather than disperse, and the confinement bound measures this tendency. The diameter grows sublinearly — slower than ballistic transport — because the self-generated velocity field is partially self-confining. The vortex is its own cage, and the cage is nearly tight: (t log t)^(1/3) is close to the conjectured optimal t^(1/3).