Crandore Hub

cxreg

Complex-Valued Lasso and Complex-Valued Graphical Lasso

Implements 'glmnet'-style complex-valued lasso (CLASSO) and complex-valued graphical lasso (CGLASSO) via a pathwise coordinate descent algorithm for complex-valued parameters, using an isomorphism between complex numbers and 2x2 orthogonal matrices. Also provides a full inference pipeline for high-dimensional sparse spectral precision matrices, including data-driven bandwidth selection, one-step debiasing, asymptotic variance estimation, entry-wise confidence regions, and FDR-controlled hypothesis testing. Supporting tools for cross-validation, simulation, coefficient extraction, and plotting are included. See Deb, Kuceyeski, and Basu (2024) <doi:10.48550/arXiv.2401.11128> and Deb, Kim, and Basu (2026) <doi:10.48550/arXiv.2606.07986>.

Versions across snapshots

VersionRepositoryFileSize
1.1.2 rolling linux/jammy R-4.5 cxreg_1.1.2.tar.gz 441.8 KiB
1.1.2 rolling linux/noble R-4.5 cxreg_1.1.2.tar.gz 442.1 KiB
1.1.2 rolling source/ R- cxreg_1.1.2.tar.gz 286.8 KiB
1.1.2 latest linux/jammy R-4.5 cxreg_1.1.2.tar.gz 441.8 KiB
1.1.2 latest linux/noble R-4.5 cxreg_1.1.2.tar.gz 442.1 KiB
1.1.2 latest source/ R- cxreg_1.1.2.tar.gz 286.8 KiB
1.1.2 2026-04-23 source/ R- cxreg_1.1.2.tar.gz 0 B

Dependencies (latest)

Imports

Suggests