Name: Plamen Krastev
Job Title: Research Computing Associate
How long have you worked for RC?
I've been working with RC since June 2011. This makes 3 years and 2 months.
What led you to a career in HPC?
Before coming to Harvard and RC, I worked at San Diego State University and Lawrence Livermore National Laboratory on a very large and ambitious project with the goal to develop a unified description of all nuclei and their reactions. The project had a very strong computational aspect, and it was a large collaboration between all major national labs and several major universities, with physicists, applied mathematicians and computer scientists all working together.
I was thrilled by the opportunity to have access to the largest and fastest supercomputers on the planet and use them to solve problems, which otherwise would be impossible to attack. To me parallel computing is like being an orchestra conductor, where you "conduct" all computers to work together in perfect harmony towards a common goal. I also discovered that HPC uniquely combines my interests in physics, mathematics and computing, and I became hooked.
What’s the best part of your job?
Together with experiment and theory, computation has become the third pillar in modern science and it is instrumental to the process of scientific discovery. I like the sense of being part of the "discovery process". It also brings a tremendous satisfaction when you can help researchers with their computational challenges in their work.
What’s the hardest part of your job?
Sometimes the things you need to do "right now" are so many that it can be a little difficult to prioritize and do them in the right order. At times RC becomes like a medical emergency room and it becomes difficult to find uninterrupted time to work on a single project. But this is also all part of the "scientific discovery" process. :)
What’s the biggest misconception about RC or HPC in general?
I think the biggest misconception about HPC in general is that if you do your calculations on a supercomputer, they will be done faster by default. This is not true however, because you still need to tell the computer how exactly to solve your problem. You still need to distribute the computational work among all individual computers and orchestrate them to work in harmony. Sometimes computers are like people - they talk more than actually work, especially supercomputers. You need to tell them exactly what to do.
Given all the research conducted on RC’s Odyssey cluster, is there any one project that stands out for you?
There are many extraordinary research projects conducted with the help of our supercomputer at Harvard. In my view, the Illustris project, which aims to model the evolution of the Universe, really stands out for me. It shows 13-billion-year evolution of the Cosmos in breathtaking details. This is simply amazing!
If you could give RC users one piece of advice what would it be?
We are always ready to help our users with their research and questions. My advise is -- If you are not sure about something, or wonder how to do it, just ask. We are here to help!
Red, White, or Blue?
Red -- definitely the color of RC -- lots of dynamics, action, and at times urgency and heat. :)
Copyright © 2014. All Rights Reserved.
Information about how to reuse or republish this work may be available at Attribution.