/isgl-paper

Simulations in the iterative sparse-group lasso paper

Primary LanguageRGNU General Public License v3.0GPL-3.0

Instructions

Simulations and real-data analysis conducted in the paper from Laria, Aguilera-Morillo, Lillo (2018) An iterative sparse-group lasso. Detailed instructions are provided with each particular simulation. In general, to run all the simulations you should have a working R installation in your system. To run some simulations, you will also need python2, so we strongly suggest to install both R and python2 through the conda environment. It is also required to install the R package sglfast

Some scripts in these simulations are designed to run in a high performance cluster with a PBS queue, specifically a Rocks 6.1.1 Sand Boa cluster, with the HPC Roll. You will probably have to comment the following lines if you want to run the R scripts in a PC.

library(Rmpi)
library(doMPI)
cl <- startMPIcluster()
registerDoMPI(cl)
closeCluster(cl)
mpi.quit()

However, they are not designed to run in a PC, so they might take some time.