A replayable intervention workflow for data-programmed Kuramoto networks in exploratory unconventional computing
Year: 2026
Type: article
Abstract: Data-driven Kuramoto representations are not new: coupled phase oscillators have previously been used to synchronize and extract structure from multivariate data. Here we address a different question: whether a fixed data-derived oscillator substrate can be interrogated reproducibly through controlled state interventions. Low-dimensional features are compiled into symmetric signed couplings, after which the couplings are frozen and the resulting dynamics are queried through a pause–inspect–edit–resume protocol. The contribution is therefore a replayable intervention workflow rather than a new synchronization theorem or clustering algorithm. A worked Iris example illustrates the protocol, while a small batch audit over five label-aware hard cases and matched same-class controls shows that recovery behavior is sample-specific and does not, in the present data, establish a general dynamical law linking hard-case status to slower relocking. Compact sensitivity checks further show that sign-preserving monotone coupling maps retain closely related pairwise phase organization, whereas shifting the attraction–repulsion threshold can change global order. These results support a cautious interpretation of iDra as a reproducible exploratory protocol built on a known oscillator substrate, with quantitative diagnostics and explicit limitations.
Source: Scientific Reports
Author: Chikoo Oosawa
Institution: Kyushu Institute of Technology
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7225874434
