Regularized estimation of precision matrix for high-dimensional multivariate longitudinal data (Q2293546)

The main contribution of this article is an alternative convex optimization algorithm based on the alternating direction method of multipliers (proposed by \textit{S. Boyd} et al. in seminar paper [Found. Trends Mach. Learn. 3, No. 1, 1--122 (2010; Zbl 1229.90122)]) and the graphical Lasso scheme. The rates of convergence in terms of the Frobenius norm are established. The efficiency of the proposed estimator is evaluated on microarray time series data for T-cell activation.