README.md

In jolars/LookAheadScreening: Safe Look-Ahead Screening for the Lasso

output: github_document

Code for the Hessian Screening Rule

Results

The results from the simulations, which were run on a dedicated HPC cluster, are stored in the results folder. The source code for the actual paper, including figures, is found in paper/.

Reproducing the Results

The results from our paper were run through a singularity container. Check the releases for pre-built singularity containers that you can download and use.

To reproduce the results, always use the singularity container. To run an experiment from the singularity container, call

singularity run --bind results:/Project/results container.sif <script>

where <script> should be a name of a script in the experiments folder, such as experiments/simulateddata.R.

Re-building the Singularity Container

If you want to re-build the singularity container from scratch (or simply want to clone the repo to your local drive), you can do so via the following steps.

1. Clone the repository to your local hard drive. On linux, using SSH authentication, run

shell git clone [e-mail]:jolars/LookAheadScreening.git

1. Navigate to the root of the repo and build the singularity container by calling

shell cd LookAheadScreening sudo singularity build container.sif Singularity

Then proceed as in Reproducing the Results to run the experiments.

Running Experiments without Singularity (Not Recommended!)

Alternatively, you may also reproduce the results by cloning this repository and starting R in the root directory of this folder (which will activate the renv repository) and then run

renv::restore()

to restore the project library. Then build the R package (see below) and run the simulations directly by running the scripts in the experiments folder. This is not recommended, however, since it, unlike the Singularity container approach, does not exactly reproduce the software environment used when these simulations where originally run and may result in discrepancies due to differences in for instance operating systems, compilers, and BLAS/LAPACK implementations.

R Package

If you want to build and experiment with the package, you can do so by calling

 R CMD INSTALL  .

provided you have cded to the root folder of this repository. First ensure, however, that you have enabled the renv project library by calling renv::restore() (see the section above).

Data

The datasets used in these simulations are stored in the data folder. Scripts to retrieve these datasets from their original sources can be found in data-raw/.

jolars/LookAheadScreening documentation built on Dec. 21, 2021, 2:15 a.m.