Statement

Let d1d\ge1, 0<ν1/30<\nu\le1/3, and 0<h<1/1000<h<1/100. Suppose X[1,1]dX\subseteq[-1,1]^d is ν\nu- from scales hh to 11, and Y[h1,h1]dY\subseteq[-h^{-1},h^{-1}]^d is ν\nu- from scales 11 to h1h^{-1}. Then constants C,β>0C,\beta>0, depending only on ν\nu and dd, satisfy

suppf^Yf1X2Chβf2\operatorname{supp}\widehat f\subseteq Y \quad\Longrightarrow\quad \|f\mathbf1_X\|_2\le Ch^\beta\|f\|_2

for every fL2(Rd)f\in L^2(\mathbb R^d).

Proof architecture

after all dilations, bounded thickenings, and translations needed by the Han–Schlag iteration. The then converts their fast decay into a single-scale observability estimate and iterates that estimate through the holes of XX.

Necessity of directional control

For d2d\ge2, two orthogonal can be ball porous while Fourier duality maps the natural measure on one to the natural measure on the other. Line porosity excludes this model obstruction.

References
  1. Alex Cohen, “Fractal uncertainty in higher dimensions,” 2024. arXiv record. Relevant: Theorem 1.1.