
Re Section 5.10 Screening rules for the lasso. This
recent paper (Liu, Zhao, Wang and Le, 2014)
proposes the "savsi" rule which is clearly superior to the DPP rule discussed
in the book (Wang and Ye were coauthors of that paper), and appears to have similar performance to the strong rules. Unlike the strong rules, savsi has the
advantage of being safe, i.e it is guaranteed never to fail. It is
quiet complex but has a closed form.
 The glmnet package efficiently fits a number of lasso
regression models. Available in R, Python or Matlab. More details of
this software and other R packages implementing methods described in this book can be found
here
 CVXR
package for solving convex optimization problems in R is finally
here (Novemeber 2017). Thanks to Anqi Fu, Balasubramanian
Narasimhan, Stephen Boyd, et al.