The Extended Trap

Heavy point particles in vortical flow get flung outward. Centrifugal force exceeds the inward pressure gradient, and the particle spirals away from the vortex core. This is a textbook result with practical consequences: it’s why cyclone separators work, why dust accumulates in specific regions of protoplanetary disks, why heavy aerosols avoid the centers of atmospheric vortices.

The authors (arXiv:2603.14390) demonstrate that spatial extent breaks this rule. A rigid dumbbell — two identical inertial point particles connected by a massless rod — can converge to a stable spinning state centered on the vortex. The two ends of the dumbbell sample the flow at different radial positions, and the resulting torque provides a centripetal bias that the individual point particles lack.

Three regimes emerge, controlled by the Stokes number. At low inertia, the dumbbell traces bounded spirographic orbits around the vortex center — trapped, but not converged. At high inertia, centrifugal ejection wins, recovering point-particle behavior. In between, there exists a window where the dumbbell converges to a fixed spinning state at the vortex center: center-of-mass stationary, orientation rotating steadily.

The dependence on Stokes number is non-monotonic. The basin of attraction for the spinning state has finite measure only over an intermediate range — it appears and then vanishes as inertia increases. Too little inertia, and there’s not enough angular momentum to lock the dumbbell; too much, and centrifugal force overwhelms the torque.

The through-claim: what a particle can do in a flow depends on what it can measure. A point particle samples one velocity. A dumbbell samples two, at different points, and the difference is information — a velocity gradient encoded structurally. The spatial extent of the particle functions as a sensor, and the sensor changes what dynamics are accessible.