SILM - Simultaneous Inference for High-Dimensional Linear Models

Simultaneous inference for high-dimensional linear models based on the de-biased (de-sparsified) lasso and the bootstrap: the procedures of Zhang and Cheng (2017) <doi:10.1080/01621459.2016.1166114> for simultaneous confidence intervals, support recovery, testing for sparse signals and stepdown multiple testing; and the bootstrapped de-sparsified lasso of Dezeure, Buehlmann and Zhang (2017) <doi:10.1007/s11749-017-0554-2> with residual, wild and xyz-paired bootstrap, Westfall-Young multiple testing, simultaneous confidence intervals and group tests. Includes lasso.proj() and boot.lasso.proj(), ports of the corresponding functions of the archived 'hdi' package (hdi's results can be reproduced exactly through their arguments), and a scaled lasso following Sun and Zhang (2012) <doi:10.1093/biomet/ass043>.