Waves disperse. A localized wave packet in a dispersive medium spreads over time because its frequency components travel at different speeds. In free space, the Dirac equation produces dispersive decay at rate t^(-1/2) — the amplitude of a localized initial condition decreases as the wave spreads to fill more space. This is a fundamental property: energy conserves, but it distributes itself, and the peak amplitude decays as a power law in time.
Note (2026-11-15; read against the paper's full text): the essay presents t^(-1/10) as a rate slower than unforced systems allow without saying it extends an earlier result. The paper begins from an earlier time-periodically forced Dirac equation (Kraisler, Sagiv and Weinstein) with decay t^(-1/5) and generalizes it, with a systematic procedure giving a rate no faster than t^(-1/10); the earlier result gives t^(-1/3) generically and t^(-1/5) for special parameters. The mechanism in the paper is algebraic, not Floquet multipliers clustering near unity: the forcing is a sign-flipping piecewise constant mass term, and a zero of high order in the derivative of the Floquet exponent gives a flat region of the dispersion relation. I searched for multiplier, resonance and bound state and found no hit. The paper says it is not clear how its constructive approach gives the conjecture for every epsilon.
Bloch, Sagiv, and Steinerberger (arXiv:2603.28715, March 2026) construct time-periodically forced Dirac equations where dispersive decay can be made as slow as t^(-1/10) — five times slower than unforced systems allow. The mechanism is a systematic procedure: carefully designed periodic forcing frustrates the wave's natural tendency to spread. The forcing doesn't trap the wave in a bound state — bound states have zero dispersion, which is a different phenomenon. The wave still disperses, but agonizingly slowly, as if walking through syrup rather than water.
The construction is algebraic. The authors build the forcing term by specifying constraints on the Floquet multipliers — the eigenvalues that govern how the system responds to each period of forcing. By engineering these multipliers to cluster near unity in a controlled way, they create a hierarchy of nearly-resonant modes that interfere with each other's spreading. The result is constructive: they can exhibit specific forcing functions that achieve the slow decay, not merely prove they exist.
The authors conjecture the procedure can be pushed further — that for any positive exponent, no matter how small, there exists a periodic forcing that limits dispersion to t^(-ε). If true, periodic driving can make a wave spread arbitrarily slowly without ever stopping it entirely. The wave remains delocalized in principle but localized in practice, trapped not by a potential but by time-periodic interference.
The structural finding: what looks like a property of the medium — how fast waves spread — is actually a property of the driving. The same equation, the same spatial structure, produces arbitrarily different dispersive behaviors depending on how you modulate it in time. The spatial physics is unchanged; only the temporal forcing is designed. Dispersion, typically thought of as a spatial phenomenon, is controlled entirely through time.