• Accepted Paper

Weighted network reconstruction via physical coupling information reservoir computing

Huiyan Wang and Shuai Wang

Phys. Rev. E - Accepted 19 November, 2026

DOI: https://doi.org/10.1103/1wsf-n52f

Abstract

Data-driven approaches, represented by reservoir computing (RC), have demonstrated significant potential in the study of complex networked dynamical systems. However, the reconstruction of weighted networks remains a long-standing challenge. To address this issue, we propose a physical coupling information reservoir computing (PCI-RC) framework. By transforming state prediction into derivative prediction, PCI-RC decomposes the dynamics of each node into an intrinsic term learned by RC and an explicit physical coupling term with a prescribed functional form. When the actual weighted network is unknown, the reservoir readout weights and physical coupling weights are simultaneously estimated through joint ridge regression, allowing the weighted adjacency matrix to be reconstructed. The reconstructed weighted network is then incorporated into the prediction stage to improve the accuracy of dynamics prediction. Experimental results across several representative coupled dynamical systems validate the effectiveness of the proposed method. The proposed framework therefore provides a gray-box approach for jointly reconstructing weighted interactions and predicting network dynamics under prescribed coupling mechanisms.

Export citation

Export citation

Choose format for download:

Download Citation

If the author has provided any supplemental materials with this article they will be available upon publication of the version of record.

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation