Paperlane

no. 16 · Tuesday 17 November 2026

Totally Positive Interpolation and Critical Sampling of Positive Operator Orbits

Ilya A. Krishtal and Jian Wu

math.FA

Abstract

How few observations are needed for the stable recovery of the initial state of a linear dynamical system? We study this question for systems governed by a positive operator, with fixed sensors and freely chosen sampling times and positive weights. Our main tool is an interpolation theorem for translates of a nonzero continuous integrable totally positive function. For every uniformly separated sequence of centers, we choose one node near each center so that the corresponding collocation operator is boundedly invertible. The displacement, node separation, and operator bounds depend only on the kernel and the lower separation of the centers; no upper bound is needed on their gaps. If a single integer orbit is a frame, we construct weighted Riesz bases whose sampling counts attain the spectral lower bound up to a bounded additive error. The same method yields bases from continuous functional calculus, including explicit resolvent bases. In the case of normal operators, we give a single-orbit frame for which every admissible weighted design has sampling count growing faster than the radial spectral count.

fields
Functional Analysis, Classical Analysis and ODEs
pdf ↗ page arXiv:2611.20122 ↗