The iterative Koopman–Beltrami (K–B) operator extracts slow-mode dynamics from a time-series observation of a dissipative system, one shell at a time, via AMG resolvent inversion. Paper 2a (Three Rings II) showed that the number of shells can be predicted a priori from a single DMD fit using a Jensen–Szegő spectral diagnostic. This paper proves that the extraction converges: under three necessary conditions (Leray dissipation, Weyl spectral-gap persistence, finite Jensen count), each pass contracts the current shell's slow-mode energy by a computable factor η < 1, and the iteration terminates after exactly p passes with bounded residual energy. The proof is formalized in Lean 4 as an independent chain (KostpmOperator) using three classical oracles (Leray 1934, Weyl 1912, Davis–Kahan 1970). No sector decomposition is required — the operator works on shells alone, with sectors reserved for the surrogate-model extensions of Paper 2c (Three Rings IV) and the NS-specific analysis of Paper 3. We demonstrate the operator on five systems of increasing complexity. Two are single-pass: a 179-bus power grid where the operator reproduces the topo-RAS vulnerability ranking, and a multiphysics aircraft where the operator recovers the gateway C/F sensor classification. One is a two-pass system: a cylinder wake at Re = 200 where the operator separates the Kármán shedding mode from the entrainment dynamics. Two are scheduling systems analyzed non-combinatorially: highway construction earned-value management (1 shell — the resolvent energy ranks work packages by schedule impact), and geo-distributed data center workload scheduling (5 shells — the richest structure in the test set, with per-datacenter scheduling-impact ranking and 24-hour carbon-aware forecasting). The construction and data center examples demonstrate the operator's utility beyond physics: the same mechanism that identifies vulnerable grid buses and separates turbulent scales also ranks work packages by EVM impact and data centers by carbon-scheduling significance — all without solving a combinatorial optimization problem. v2 (2026-05-09): code supplement added. Two artifacts now accompany the paper. The first, kostpm-lean-v0.1.0.zip, is the Lean 4 formalization comprising KostpmOperator/ (the certified K–B operator with resolvent-energy bound and convergence proof) and KostpmCerts/ (Lie-algebra structure-constant certificates for SO(3), SL(2,ℝ), SL(3,ℝ), Heisenberg, U(1), and the Standard Model gauge algebra), totaling 26 source files; lake exe cache get note that the empirical results of Three Rings IV (Sobol × KOSTPM sensitivity, Szegő saturation testing, the SustainCluster and EVM validations) are reproduced from this same package — see https://doi.org/10.5281/zenodo.19571618 for the IV companion paper.
Building similarity graph...
Analyzing shared references across papers
Loading...
Andrew Strelzoff
Benjamin Strelzoff
Tulio Sulbaran
University of Michigan
The University of Texas at San Antonio
Charles River Analytics (United States)
Building similarity graph...
Analyzing shared references across papers
Loading...
Strelzoff et al. (Sat,) studied this question.
www.synapsesocial.com/papers/6a0171ce3a9f334c28271d96 — DOI: https://doi.org/10.5281/zenodo.20099858