Granular gases — dilute collections of particles that collide and lose energy — develop interesting viscosity under shear. In the simplest models, the energy lost per collision (the restitution coefficient) is constant. Kikuchi, Kobayashi, and Takada (arXiv: 2603.19605) let it depend on collision velocity: faster impacts lose more energy.
The result: the shear viscosity, plotted against shear rate, develops an S-curve. Two stable branches — low viscosity at low shear, high viscosity at high shear — connected by an unstable middle segment. As shear rate increases, the system rides the low branch until it disappears, then jumps discontinuously to the high branch. Both branches show Bagnold scaling (viscosity proportional to shear rate), but with different prefactors.
The through-claim: the viscosity jump comes from kinetics, not friction. In dense suspensions, discontinuous shear thickening (the Wyart-Cates mechanism) comes from particles jamming and friction activating. Here, the gas is dilute — no contacts, no jamming, no friction. The discontinuity comes purely from the velocity dependence of energy dissipation: at high shear, collisions are faster, lose more energy, and the granular temperature (the velocity fluctuation amplitude) can't keep up. The system undergoes a kinetic phase transition between two dissipation regimes.
This means discontinuous rheology — sudden jumps in viscosity — is a broader phenomenon than the dense-suspension literature suggests. You don't need contacts. You need velocity-dependent dissipation. The bistability is in the energy balance, not the microstructure.
Kikuchi, Kobayashi & Takada, 2603.19605. Granular physics / rheology / shear thickening / kinetic theory.