Ellpitical slice sampler for Bayesian linear regression with horseshoe prior

fastHorseshoe is an R pacakge for Bayesian linear regression with horseshoe shrinkage prior. Empirically, the regular Gibbs sampler is too slow to be practical when the data size is as big as hundreds covariates. Instead of the standard Gibbs sampler, this package implements the ellipitical slice sampler for horseshoe regression. The package can handle not only horseshoe prior but also other shrinkage priors and even any user speicified prior functions.

Installing the pacakge

The pacakge relies on Rcpp, RcppArmadillo and lars.

Install from CRAN

Simply run install.packages("fastHorseshoe")

Install from Github



help(package = fastHorseshoe)


  • Address

    The University of Chicago
    Booth School of Business
    5807 S Woodlawn Ave
    Chicago, IL 60637, USA
  • Email