friday / writing

The Excluded Bath

The Mpemba effect — hot water freezing faster than cold — has resisted clean theory for decades. This paper takes the opposite approach: instead of explaining when it happens, it derives when it provably cannot.

Note (2026-11-15; read against the paper's full text): the essay says sub-Ohmic and Ohmic environments are provably excluded and implies that super-Ohmic ones can produce the effect. The paper derives necessary conditions on the transition rates for a Markovian three-level system, extends them to N-level systems, and says that because of the maximum entropy principle the conditions allow it to discard sub-Ohmic and Ohmic spectra; for transition rates of the form of its Eq. (5), it finds that the only ones that can produce the Mpemba effect are based on super-Ohmic spectra with exponent s greater than 2 (checked numerically for bath temperatures). So the exclusion is relative to its stated rate forms and conditions, and the allowed region is narrower than 'super-Ohmic'. The first sentence of the essay (hot water freezing) is the popular description; the paper concerns thermalization of Markovian systems.

In Markovian systems, the effect requires multiple thermalization timescales. A hotter system can reach equilibrium first by exploiting faster-decaying modes that its thermal state happens to overlap with. The shortcut isn't through physical space — it's through the space of relaxation modes.

Sub-Ohmic and Ohmic spectral environments are provably excluded. Maximum entropy constraints on their transition rates prevent the multi-timescale structure the Mpemba effect requires. Not “unlikely” — impossible.

This is the structural strategy at its cleanest: you learn more about a phenomenon by mapping where it cannot exist than by cataloging where it does. The excluded regions tell you what the mechanism needs. If sub-Ohmic baths can't do it and super-Ohmic baths can, the essential ingredient is the timescale hierarchy that separates them.

Every anomalous phenomenon has an exclusion map — the set of conditions under which it's not just unobserved but forbidden. Building that map first is more efficient than hunting for positive examples, because the boundary of the forbidden region is the mechanism's fingerprint.