Find out how to run RStudio, MATLAB, and Stata on Datahub and PARASOL (formerly DSMLP), how licensed software gets installed, and which software you can install yourself. The standard images cover most courses and projects and receive priority support. See Standard Images, Tags, and Pinning.
R is included in every standard image, alongside Python and Julia, and runs as a
Jupyter kernel. Entering ?command_name in a cell displays the documentation for
a function.
RStudio does not require a separate sign-in.
In a new account, install.packages() does not work until a personal package
library exists. Create the library once, in the RStudio Console, during the
first RStudio session:
dir.create("~/R")
dir.create("~/R/library")
.libPaths("~/R/library")The library is stored in the home directory and counts against its quota. See Storage Quota.
rstudio-notebook derives from datascience-notebook, the CPU image, rather
than from scipy-ml-notebook, and is therefore not GPU-enabled. See
Standard Images.
A course that needs both RStudio and a GPU requires a custom image or a second
environment.
See also: Getting Help
MATLAB can be used on the platform but is not a standard feature. The ITS scope of support classes MATLAB, as Jupyter kernels or as the Web UI, and GNU Octave among the capabilities in Complex & Experimental Capabilities. These capabilities are regularly used on the platform but fall outside the normal bounds of ITS support, and their use in a course is led by the instructor or the Technical Point of Contact (TPOC).
Research IT documents running MATLAB from a small shell script that sets a
license-file environment variable. The license covers course work and research
work alike. The script exports
MLM_LICENSE_FILE, points to a MATLAB installation held on the cluster, and runs
MATLAB headless: with -nojvm -nodisplay -nosplash for interactive use, and
with -batch for a script.
MATLAB runs inside a job and never on the login node, where running computation
is prohibited. See
What the Login Node Is For.
Scripts are made executable with chmod before they are run.
Research IT maintains the MATLAB release, the installation directory, and the license file. Confirm the current paths with Research IT at rcd-support@ucsd.edu before building a course around them.
Stata runs in the scipy-ml container for users with provisioned licensing.
Research IT Services installs Stata into the member's home directory, and the
member then runs it from inside a container as ~/stata-se.
The platform does not provide Stata licenses, and a license is required before installation. A holder of a Stata license through a department or project arranges the installation by writing to rcd-support@ucsd.edu.
Installing licensed software is permitted. Purchasing the license is the responsibility of the user or the user's sponsoring department. Research IT Services can assist with the installation. Some versions of some products are not compatible with a containerized cluster environment.
Before purchasing a license, confirm the following with the relevant contact:
For research use, the contact is rcd-support@ucsd.edu. For a course, the contact is parasol@ucsd.edu.
Licensed data is subject to separate rules. See Restricted & Licensed Datasets.
The ITS scope of support names the following capabilities as available on the cluster but outside normal support.
| Capability | Documentation |
|---|---|
| MATLAB (Jupyter kernels or Web UI), GNU Octave | MATLAB, Octave & Other Complex Applications |
| Spark clusters | ITS publishes no documentation. The only description of a multi-node Spark topology on this platform is a DSC 102 assignment, which is course material rather than documentation. |
| ArcGIS integration | ITS publishes no documentation. |
| Postgres and other persistent services | Launched from Kubernetes manifests and reached by in-cluster service name. See Direct Kubernetes Use and Session Events. |
| Background batch processing and analysis pipelines | Job Modes |
| Visual Studio Code integration | Remote Editor Setup |
| Containers not derived from a standard image, and student-built containers | Building & Publishing a Custom Image |
The complex or experimental classification concerns support, not capability. These capabilities work, and courses use them. Incorporating one into a course requires the instructor or TPOC to become independently familiar with the underlying technology and then to serve as primary support for students' use of it. ITS gives technical guidance but, without an advance agreement, does not take on implementation or front-line support. The scope of support is published in Complex Customizations & Experimental Features.
Book a 1:1 Consultation at least one full quarter before any planned use of one of these capabilities, to discuss feasibility. Support & Technical Consultation describes consultations.
A member can install anything that installs into the member's own home directory: Python packages into a virtual environment with its own Jupyter kernel, R packages into a personal library, and a custom kernel. The procedures are in Customizing an Environment.
Software that installs into the operating system requires a custom image. A
container has no sudo and no flag that grants it, and root is available at
image build time instead, on a different machine. See
Root Access and System Packages.
Building & Publishing a Custom Image
covers building an image.
Package lists change with every image build. The current contents of each image are published from the image repository. See Finding the Package List.