Institutes

• Here is a list of the available software.
• Here is their policy regarding machine learning.

1. Prepare results which satisfy statistical disclosure control rules and put them in the "Ready for Checking" directory in a directory with the name YYYY_MM_DD_<descriptivename>_v1.
2. Include a completed copy of the output request form (which you can find in the SRS project folder.)
3. Email statistical.support@ons.gov.uk with a subject "[date] Output request for project [number]". "The email body should note project number, name of folder for output, and any useful context or considerations that may assist SRS officers."
4. If they are approved, the output will be emailed to you. Remember to notify the SRS Impact team (IDS.impact@ons.gov.uk) if results are publised!

Stick to .R, .csv, and .txt.

Only very common file types are supported. But there are some surprisingly common ones that are not included such as .json, and .md and .rmd.

For code, you are pretty much limited to .R files, and these need to have clear comments about the script's purpose/use and cannot contain any data or system information.

The SRS guidance says

• For outputs requests and queries: email statistical.support@ons.gov.uk
• For ingests, transfers and all other SRS requests and queries: email srs.customer.support@ons.gov.uk

1. Create an account, (there details on the intranet about how to do this.)
2. SSH into login.hpc.cam.ac.uk (for CPU) or gpu.hpc.cam.ac.uk (for GPU) using your CRSid as the username.
   • module avail <str> will list the modules that are available.
   • module load <modulename> will load a particular module.
   • mybalance takes minutes to run.
3. Use Slurm to run jobs, there are details of useful commands here.
   • This is done with a Slurm "submit script", which describes resources requested and the commands to run. This gets submitted with sbatch <path/to/submit.txt>. Output will be written to slurm-<jobnum>.out files.
   • squeue -u <username> will print out your jobs.
   • "Array jobs" are used for embarrassingly parallel tasks. Keep in mind that jobs are limited to 36 hours. A rule of thumb is to keep each array task at about an hour of runtime (remember the scheduler will try to fit jobs in so smaller ones may get run earlier.) Roughly, each array task will run as a separate instance of the Slurm script (which is really just a bash script), but with a different SLURM_ARRAY_TASK_ID environment variable value.
   • Use --cpus-per-task to specify the number of CPUs (on the same node) to use for a task.
   • Use the template! for setting up Slurm tasks! Read through it carefully and pay attention to the notes on the intranet as a way to understand the settings.
4. Remember to store large material that doesn't need to be backed up in /home/<CRSid>/rds/hpc-work (up to 1TB per user). There is also /home/<CRSid> (up to 40GB per user backed up.) You can get a listing of your space with the quota command. The file system is shared across the login node and the compute nodes so you only need to move your code onto the login node and it will be available everywhere. Git is an option.

#+end_src

Log into the MRC BSU intranet and follow the links for the HR pages.

Log into timetastic and select the dates.

Use the University library service to access the Dow Jones Factiva database which has all their articles index.

Point your browser to https://rstudio.mrc-bsu.cam.ac.uk/ and log in with the same credentials you use to access Brahms.

The University of Melbourne has a service to verify qualifications here. Using the free online Verification of Qualification service allows anyone to check if someone has completed a University of Melbourne qualification/s. The service lists the names of the qualification/s, and the year that the qualification/s was conferred.

There are some notes on how to connect to the VPN here. The takeaway is that you need to use GlobalProtect VPN on linux and the portal is vpn.unimelb.edu.au.

It has been proposed students should register for a desk every 6 months.

University of Melbourne

School page

• Information session slides (2024)

AMSI

https://srs.amsi.org.au/

ANU

https://science.anu.edu.au/study/summer-research-scholarship-program

Cancer Council Victoria

https://www.cancervic.org.au/research/grants/research_sum_vacation_studentships

You can print via the web portal available from the following link:

https://papercut.unimelb.edu.au/user

These dates are not definitive, please refer to the official source at LINK.

Handbook Page

LINK TO MATERIALS

This is a workshop looking at how git and GitHub can be used to assist in the pursuit of reproducible research.

Tutorial Notebooks