Thoughts and reflections at the end of my Summer of HPC
As summer turns into autumn, we bring to a close the Summer of HPC. I have had a great time learning new things, solving new problems, exploring new areas and working with new people. And here at the end, I suppose it’s time to provide closing thoughts.
Results of the project
As previously discussed, over the course of my project I developed from scratch a program that calculated view factors between surfaces. It took in a surface mesh, triangles describing the shape of the objects, and tested to see if had a line of sight to one another. If they did, the view factor was calculated, computing the double area integral via gaussian quadrature. A simple version of parallelization was also implemented, where each processor was assigned a set of triangle pairs to test and calculate.
What I wish I’d had the chance to do
While I’m proud of what I’ve made, there’s many things I would have loved to do if there had been more time. One of the first things would have been a better implementation of parallelization. Currently, some processors could be assigned a set of pairs where none of them see each other, meaning the processor needs to do little-to-know calculation, while another processor will have many. There are better ways to parallelize, ways that would more evenly distribute the workload across processors.
Additionally, while integration via gaussian quadrature is an acceptably fast and accurate means of calculating the view factor, it is by no means the only or explicitly best one. With more time, I would have hoped to test multiple methods, analyze which works best under circumstances of distances, angles and sizes of the triangle, and changing the program to choose the best calculation methods for each triangle-pair.
One other thing I would have liked to do more of is proper testing on HPC systems. While the project was designed with them in mind, I was running it just on my own laptop the whole time. With more time, I would have liked to grow my knowledge and experience of working on HPC systems.
Working Together from Afar
At this point in September 2021, most people will know the troubles of remote working,. But I had to learn a lot working with a partner in Germany and mentors in France – not least of which being remembering time zones. Keeping up communication, cooperating on solving problems and knowing what others are working on are new skills I’ve had to learn working this project. As things slowly get better, the next piece of work I’ll have to do will probably be in person, but I’m glad for having learnt these skills in an increasingly global remote world of science and technology.
So as I bring my Summer of HPC to a close and begin my next venture – studying a Masters in Quantum Technology, I am very thankful for the opportunities the PRACE’s Summer of HPC has given me and for what I have learnt. I would like to thank PRACE for giving me this opportunity. I also must say thank you to Daniel Pino Muños and Modesar Shakoor for their mentorship over this project, and to Mukund Kashyap for working with me. Thank you all, I hope one day I have the chance to thank you in person.
And with that, I close with thank you, and goodbye.