Linux Standard Software

Introduction

On Linux workstations in Stronghold, scientific software is managed using a module system. In particular, there are "software modules" and "environment modules". Both of these are accessible using the Linux command line. In general, we prefer environment modules, because they provide software package collections that are known to work well together.

Software Modules

Software modules on a given workstation can be accessed using the Linux command line. In particular, open the Terminal application and type module avail. This will show a long list of software packages that is organized by category (see example below).

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ category: languages ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
R/3.2.3               java/9.0.4            stata/mp-14-20151007
R/dev                 julia/0.4.3           stata/mp-14-20151116
anaconda/2-2.4.0      julia/0.4.6           stata/se-14-20151116
anaconda/2-2.7.0      julia/0.5.0           stata-addons/v1
anaconda/2.3.0        mathematica/11.0.0    stata-addons/v2
anaconda/3-2.4.0      matlab/R2017a         stata-addons/v3
anaconda/3-3.5.0      mro/3.3.1             stata-ado/20160907
anaconda/4.3.0        nodejs/0.12.9         stata-mp/19Dec2017
java/8u162            scala/2.11.11         texlive/2018

Loading a Software Module

In order to use a module, you can simply load if from the command line using the module load command. For example we can load the julia language module by simply typing module load julia from the command line. The software module can now be launched as it normally would be. In the case of Julia, we would simply launch the interpreter by running julia from the command line. Note that this will load the version of Julia that has been designated as the default; as of this writing, that is Julia 0.4.6. However, there is a newer version, which we could load by running module load julia/0.5.0. In this case, we would load version 0.5.0 of Julia.

Environment Modules

Environment modules are essentially a collection of software. For example, a given lab might have a collection of software packages that they use frequently. This collection of packages can be wrapped up in an "environment". Technically speaking, these environments are Conda environments, and we simply store them in the module database as if they were standard software.

We can view the available environment modules by simply running module avail conda. This will show the list of available environment modules (see example below).

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ name: conda*/* ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
conda/bcbi_v0.0.0        conda/cbc_generic        conda/psteytest
conda/bcbi_v0.0.1        conda/datasci_v0.0.0     conda/riipl_v1
conda/bcbi_v0.0.2-alpha  conda/datasci_v0.0.1     conda/riipl_v2
conda/bcbi_v0.2.3        conda/datasci_v0.0.2     conda/riipl_v3
conda/bcbi_v1            conda/datasci_v1         conda/riipl_v4
conda/bcbi_v1_mari       conda/datasci_v2         conda/riipl_v5
conda/bcbi_v2            conda/fgelintest         conda/root
conda/bjorkegren_v1      conda/graphlab_2.1       conda/sing_v0.1.0
conda/cbc_conda_v1       conda/nlp_v0.1.0         conda/stronghold_v1

Each of these environment modules contains a huge list of packages. For example, consider the conda/datasci_v2 environment module. In order to view the available software in this particular module, run conda list --name=datasci_v2to produce a huge list of output that resembles the truncated list below.

# packages in environment at /opt/browncis/conda/envs/datasci_v2:
#
alembic                   0.9.5            py27h0668a64_0
amqp                      1.4.9            py27h5eaba50_0    brown-data-science
anyjson                   0.3.3            py27hdf0fd94_1
asn1crypto                0.22.0           py27h94ebe91_1
atk                       2.25.2               hd48c387_0    brown-data-science
atom-editor               1.21.1               h67387a4_0    brown-data-science
babel                     2.5.0            py27h20693cd_0
backports                 1.0                      py27_0
backports_abc             0.5                      py27_0
billiard                  3.3.0.23         py27h318c9ab_0    brown-data-science
bkcharts                  0.2                      py27_0
bleach                    1.5.0                    py27_0
bokeh                     0.12.7                   py27_0
boost                     1.61.0           py27h2e6ecf4_0    brown-data-science
boto3                     1.4.4                    py27_0
botocore                  1.5.78           py27hd4e8fcf_0
bzip2                     1.0.6                h6d464ef_2
ca-certificates           2018.03.07                    0
cairo                     1.14.10              haa5651f_5
celery                    3.1.25           py27h97bc7ca_0    brown-data-science

Loading Environment Modules

Much like software modules, we can load environment modules with the module load command. In particular, we give the full name of the module. For example we could load the conda/datasci_v2 module by running module load conda/datasci_v2.

Default Environment Modules to load

The conda/cbc_generic module provides R, Python, and Rstudio. In general, a user may start with the conda/cbc_generic module by

module load conda/cbc_generic

If a user needs to access Julia, the user may load the conda/bcbi_v0.3.1 by

module load conda/bcbi_v0.3.1