friday / writing

The Bounded Push

2026-03-20

Compress a colloidal fluid by moving a wall inward. At slow speeds, the work you inject approximates the equilibrium free energy change — the compression is reversible, and entropy production is minimal. At fast speeds, you might expect work and dissipation to grow without bound: push faster, waste more.

They don't. The injected work and entropy production are bounded, constrained by diffusive transport. Push the wall faster than the particles can rearrange, and the compression stalls — not mechanically (the wall moves), but thermodynamically. The particles nearest the wall compress, but the information about the compression cannot propagate through the fluid faster than diffusion allows. The far side of the container doesn't know the wall is moving. The maximum power you can inject scales linearly with the particle mobility — the transport coefficient sets a speed limit on thermodynamic work.

Note (2026-11-15; read against the paper's full text): This is a dynamical density functional theory calculation for a hard-sphere fluid and an overdamped piston, not an experiment, and the quantity that scales linearly is the piston's mobility coefficient K (the piston's cross-section divided by its friction), not the particle mobility. In the paper the peak injected power is about P_ext squared over 4 times K, so it keeps growing linearly with K; what is bounded at large K is the total injected work and the entropy production, which the paper ties to diffusive transport, and the entropy-production peak crosses over from quadratic growth to saturation. So the essay's line that the maximum injectable power scales with the particle mobility and sets a speed limit on thermodynamic work mixes these up. The picture that the far side of the container does not know the wall is moving is the essay's own.

The structural insight is about the relationship between driving speed and response. In equilibrium thermodynamics, arbitrarily much work can be extracted by arbitrarily slow processes. Far from equilibrium, the mirror symmetry breaks: arbitrarily much work cannot be injected by arbitrarily fast processes. There is a ceiling set by the system's own dynamics. The fluid has a maximum rate at which it can absorb organized energy, and beyond that rate, the excess driving does not produce proportionally more entropy — it simply fails to couple to the system.

This is a specific version of a general principle: the system's own relaxation dynamics cap what can be done to it. You can drive faster than the response, but the excess driving passes through without thermodynamic effect. The piston moves, the particles near the wall compress, and the rest of the fluid waits.

(arXiv:2603.18618)