Nuffield Research Placements at EPCC

Three secondary-school students have joined EPCC for a three-week research visit, each investigating a separate topic proposed by EPCC. They are being supervised by EPCC technical staff, principally Joseph Lee, Lorna Smith, and James Richings.  

While based at EPCC, the Nuffield students have also visited EPCC's Advanced Computing Facility, where they were given a tour of the facilities we host there. 

Following their research placements, each student will produce a report and an academic-style poster to be showcased at an event organised by the Nuffield Foundation.  

Student projects at EPCC 

EPCC staff outline the projects they will each supervise. 

Machine Learning: supervised by James Richings. 

The project is designed to enhance the student's understanding of the Python programming language and the basics of machine learning, starting with training a model for the binary classification of images according to the type of object depicted.  

If the student progresses well, we will then look at Generative Adversarial Networks (GANs), which aim to generate an image to ‘fool’ a second network ("the detective") into thinking the generated image is a real image and not generated by an AI. This process is iterated to produce a better detector and generator of images. 

Completion of the project will give the student a very good understanding of how to use machine learning in a useful context. The key challenges will be to understand the principles of neural networks and tuning the parameters to train the system efficiently.  

Navier Stokes Solver: supervised by Lorna Smith. 

This project focuses on taking the code used in one of EPCC’s Outreach demonstrations which, using a simple Navier Stokes solver, simulates the amount of lift generated by modifying three parameters for the wing of an aeroplane: the camber (shape of the wing); the angle of attack against the air flow, and the thickness of the wing.  

The code is used in a demo run by Wee Archie, the mini supercomputer used by EPCC to demonstrate the principles of high-performance computing at outreach events. It has been adapted to use GPUs as the next version of Wee Archie may employ NVIDIA Jetson boards, which have embedded GPUs. The project will thus test the performance across a number of different GPU and CPU systems.   

Benchmarking A Quantum Chemistry Library: supervised by Joseph Lee. 

The focus of this project is the benchmarking of a quantum chemistry library (CP2K) and a computational fluid dynamics library (XCompact3D) on Cirrus, the Tier-2 supercomputer hosted by EPCC.  

These codes are heavily used by scientists to accurately simulate different physical systems, and are computationally intensive. It is important to understand and optimise their scaling performance by tuning a range of parameters, such as the number of threads and processes per node. This research project will be a good opportunity for the student to learn about the various components of supercomputing, including hardware architecture and software development practices.  

The Nuffield Foundation 

The Nuffield Foundation is an independent charitable trust with a mission to advance educational opportunity and social well-being. Its activities include the provision of opportunities for young people to develop skills and confidence in science and research. 

Links 

The Nuffield Foundation 

Nuffield Research Placements