lglasso: graphical lasso for longitudinal data

The R package lglasso is designed to estimate precision matrices (i.e., networks) from high-dimensional longitudinal data under a normal distribution. It can be used to estimate either one-stage models, where a single network underlies all longitudinal measurements, or two-stage models, where the network before treatment differs from the one after treatment (exposure). Both model types can be further classified as homogeneous—treating all individuals as i.i.d. samples—or heterogeneous, which includes individual-level random effects. For details on these definitions and usage, please check the references and links below. The ADMM algorithm is adopted to speed up computation, and the main functions are written in C++. If you have any questions or suggestions, please email chowstat@gmail.com.