Reproducibility package for "Cone Extended Rayleigh Quotients for Directed Graph Learning: Minimax Spectral Certificates, Sensitivity, and Adaptive Control"
Year: 2026
Type: software
Abstract: This record contains the code and saved numerical outputs used to reproduce the computational experiments in the manuscript “Cone Extended Rayleigh Quotients for Directed Graph Learning: Minimax Spectral Certificates, Sensitivity, and Adaptive Control”. The package includes experiments on certified learning for directed stochastic block models, boundary complementary quasi-pairs, right-left spectral sensitivity validation, comparison with graph symmetrization, and adaptive spectral control on the directed Cora citation network. The archive contains the principal Python scripts, canonical CSV/NPZ numerical outputs, and a README describing the structure of the reproducibility package. Random seeds and experimental parameters are specified in the corresponding scripts.
Authors: Il'yasov, Yavdat, Valeev, Nurmukhamet
Institution: Institute of Mathematics with Computer Center
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204254423