The paper examines a mathematical system that combines two types of equations: a conservation law, which describes how some physical quantity spreads through space over time, and a "relaxation oscillator," which is a component that naturally tends to produce rhythmic, periodic behavior. Think of something like a fluid sloshing back and forth while also being driven by a rhythmic pump. Without any diffusion (spreading or smoothing) in the system, there turn out to be infinitely many possible periodic solutions, meaning the long-term behavior is completely unpredictable from the equations alone.
The key finding is that adding even a tiny amount of viscosity, which mathematically acts to smooth and spread things out, breaks this degeneracy and forces the system to settle on one unique periodic solution. This is a phenomenon called "selection": a small physical effect that might seem negligible actually picks out a single preferred behavior from what was previously an infinite menu of possibilities. The authors focus on symmetric initial conditions, under which they can prove that this selected solution is not just unique but also linearly stable, meaning small perturbations will decay rather than grow, so the system reliably returns to that solution if nudged slightly.
The proof strategy is technically sophisticated but conceptually elegant. The system operates on two different timescales, one fast and one slow, and the authors exploit this separation through an averaging technique that simplifies the analysis. They also use tools from spectral theory, which involves studying the characteristic frequencies and modes of the system, to establish stability. The theoretical results are backed up by numerical simulations that illustrate how the selection and stabilization actually play out in practice.