When Geometry Becomes the Price of Perfection
26 Apr 2026, Yanjiang
Geometric complexity of a perfect reset diverges as error vanishes, revealing a fundamental limit of thermodynamics.
The third law of thermodynamics has always had a cosmic sting: you cannot cool anything to absolute zero in a finite number of steps. No matter how clever the refrigerator, how cold the initial bath, the last few billionths of a degree retreat like a mirage in a desert. But what does that cost actually look like? How do we measure the price of perfection — the ultimate effort needed to erase every last trace of thermal information from a system?
A preprint (arXiv:2604.27858) from Tan Van Vu and Keiji Saito at Kyoto University provides a surprisingly elegant answer. The cost, they argue, is geometric complexity — the minimal “distance” along the space of all possible physical maps that separates what your operation does from what you wish it would do. And the result is as stark as it is universal: achieving zero error in any reset operation — whether cooling, information erasure, or state copying — requires the geometric complexity to diverge. In plain terms: you can get as close as you like, but the last step demands an infinite investment of time, energy, or control bandwidth.
Think of it like a journey across a landscape whose terrain is made of operations themselves. The identity map — the “do nothing” operation — sits at one point. The perfect reset map — the one that scrubs every memory of the initial state — sits at another. The geometric complexity is the length of the shortest path between them, measured along a specially designed metric that respects the structure of physical processes. Unlike a real journey, however, the landscape is curved by the constraints of thermodynamics and quantum mechanics: certain directions are harder than others.
To understand why this matters, let’s consider what a reset map actually does. Imagine a box of gas whose every molecule remembers its history — which side of the box it started on, how fast it was moving, whether it ever interacted with a neighbor. A perfect reset would wipe all that information, leaving the gas in a state that depends on nothing but the temperature of its environment. This is the unattainable goal. Vu and Saito prove that, no matter what physical mechanism you use — laser cooling, feedback control, quantum gates — the geometric complexity of implementing that reset map grows without bound as the error shrinks toward zero.
The key insight is a trade-off relation that emerges from the geometry of maps. By analyzing continuous paths on the space of classical stochastic maps and quantum channels, the authors show that the minimal geometric complexity is always at least as large as the inverse of the execution error. If you want the error to drop by a factor of ten, the complexity must rise by at least the same factor. “This is not a bug in any particular experimental setup,” the team writes, “but a fundamental limit imposed by the structure of physical evolution itself.”
What makes this result powerful is its scope. The geometric complexity is not a single resource like time or energy — it is a unified currency that naturally converts between them. A high-complexity operation might require an extremely long duration, a very large energetic cost, or an impossibly precise control pulse. The geometry sees all these as the same thing: distance along a geodesic in map space. This means that the third law’s unattainability principle is not just about cooling; it applies equally to erasing a bit of memory, copying a quantum state, or resetting a qubit in a quantum computer.
This is not a metaphor. The team constructs an explicit Riemannian metric on the space of maps, and the geodesic length is a concrete mathematical object. For classical stochastic maps, the metric is derived from the Fisher information of probability distributions. For quantum channels, it emerges from the Bures metric on density matrices. In both cases, the geometry encodes the physical difficulty of transforming one map into another.
Perhaps the most striking implication is for quantum computing. Every quantum algorithm relies on reset operations — refreshing ancilla qubits, initializing registers, cooling the processor. If every reset step carries a geometric cost that diverges as error vanishes, then the total complexity of a long computation may be bounded by more than just gate fidelities. It may be bounded by the geometry of the reset itself. “The unattainability of perfect reset,” the authors note, “establishes a strict geometric limit on the physical realization of any thermodynamic control or quantum computation.”
For decades, physicists have understood the third law as a statement about time: cooling to absolute zero would require an infinite number of steps. Vu and Saito’s work recasts it as a statement about shape — the curved shape of the space of all possible operations, and the impossibility of traversing certain paths without paying a steep toll in complexity. The third law’s “unattainability principle” is thus revealed as a geometric theorem, as fundamental as the Pythagorean theorem but applied to the much stranger landscape of physical maps.
The road ahead is clear. The team has identified a universal bound, but many questions remain: are there practical protocols that can approach this bound? Can geometric complexity be measured in an experiment? And does the bound tighten when additional physical constraints — like causality or energy conservation — are imposed? The framework, built on the geometry of maps, is designed to accommodate these extensions.
In the end, this work does not merely quantify the cost of perfection. It offers a new language for thinking about the limits of control — a language written in geodesics and curvature, where the third law becomes a theorem about the shape of possibility itself. And that shape, it turns out, is not a straight line, but a curve that rises ever more steeply as you approach the point that can never be reached.
Yanjiang is an online editor of Loom Science
References
- Tan Van Vu et al., Geometric complexity in thermodynamics, arXiv:2604.27858