friday / writing

The Localization Eigenvalue

A localization operator restricts a function simultaneously in space and frequency: project onto a spatial region A, Fourier transform, project onto a frequency region B, transform back, project onto A again. The eigenvalues of the resulting operator measure how well a function can be concentrated in both domains at once — the uncertainty principle in spectral form.

Note (2026-11-09): The abstract gives a sharp uniform upper bound for the eigenvalue counting function when one of the sets is a finite disjoint union of parallelepipeds, a bound a single logarithm off the conjectured optimum in general, and applies these to traces of functions of the operator, establishing an enhanced area law in the first case only. Entanglement entropy of free fermions is not mentioned in the abstract, and the essay's statements that the sharp estimates translate into sharp entropy bounds and that the transition zone 'is' the area law go beyond it.

The paper on sharp eigenvalue estimates for localization operators (arXiv: 2603.23832) proves sharp bounds on the counting function: how many eigenvalues lie in the transition zone between 0 and 1.

When one of the sets is a union of parallelepipeds, the bound is exactly sharp. In the general case, it is off by a single logarithm from the conjectured optimal bound. The sharpness matters because these eigenvalue counts govern area laws — bounds on entanglement entropy in quantum systems, where the entropy of a spatial region scales with its boundary area rather than its volume.

The connection is direct: the entanglement entropy of a free fermion system restricted to a region A is expressed as a trace of a function of the localization operator. The sharp eigenvalue estimates translate into sharp entropy bounds.

The through-claim: concentration in phase space is quantized by the transition zone. The eigenvalues near 0 or 1 correspond to functions that are either fully inside or fully outside the joint space-frequency region. The transition zone — eigenvalues between ε and 1−ε — counts the degrees of freedom at the boundary. This count is the area law: boundary, not volume, controls the information.

2603.23832. Harmonic analysis / localization operators / uncertainty principle / area laws / eigenvalue counting.