UROS 2018 Project: Surface Scanning with High Throughput Computing
by Benjamin Kendrick
Suppose you want to run a simulation of the universe in its initial expansion. The explosion of the big bang is a random process in the program you have created, so you want to run it many times. The problem is that each simulation takes hours to run so it’s incredibly impractical to do each run back to back on your computer. What if instead we could run the simulations on a whole room of computers at once, how much time would this save? How energy efficient is this system? How do we even do it? These questions and many more were the basis of my research project.
The project itself was split up into two main parts. Firstly, to set up the system of computers, or cluster, that could talk to each other and more importantly, send a job from one machine to all the others. For this step we looked to HTCondor. This software was used to sort out the networking side of the project.
The second half of the project was to create and run a full-scale production job on the cluster to really investigate its capabilities. For this we used a program called GROMACS, which is a molecular dynamics simulation software, mainly used for protein simulations but in our case, we used GROMACS to run simulated annealing on different clusters of Calcium Fluoride. The idea being that the heating and cooling of the atoms would be a random process and the final positions and energies would be interesting to analyse. To do this we used VMD, a program which visualises data from molecular dynamics. One use of this was to make the gif seen below.
I have learnt an incredible amount of new skills throughout this project, more than I ever thought I would. Firstly, I had never used Linux before, so learning this new OS as well as other skills such as bash scripting has been very useful, as well as being a lot of fun too. As previously mentioned GROMACS, HTCondor and VMD were all introduced to me during this project as well as the languages involved with them. To be given the opportunity and time to learn how to use this range of programs and coding languages has been wonderful, and I thoroughly thank Matt Watkins for this as well as his continued support and enthusiasm.
Looking ahead to the future I can already tell how useful the UROS project scheme will prove to be. From the technical side, the chance to gain familiarity with this whole new area of computing for me is invaluable. From the personal side I feel significantly more confident with tackling problems that require tools and methods that I have never used before, and this kind of adaptability is such an important skill in today’s working world.
As well as the standard “skills I have learnt” stuff, I feel incredibly happy with the work and research we have done, and lucky that I had the chance to do so.
*To view Benjamin’s project poster, please click on the thumbnail below: