Summer Interns Gain HPC Skills, Professional Development at the OLCF
Students supplement classroom learning with firsthand experience
Each year, students of all ages and backgrounds intern at the US Department of Energy’s (DOE’s) Oak Ridge National Laboratory (ORNL) through opportunities offered by the lab and Oak Ridge Associated Universities. Hundreds of individuals participate in these programs every summer, from undergraduate and graduate students to high school graduates and PhD candidates with diverse interests ranging from biochemistry to astrophysics.
For students interested in coding, artificial intelligence, and other aspects of high-performance computing (HPC), the Oak Ridge Leadership Computing Facility (OLCF), a DOE Office of Science User Facility located at ORNL, provides internships to help jumpstart their careers. With the guidance of staff mentors, the 26 interns at the OLCF this summer are programming for HPC systems, developing software, and honing other skills in support of their research projects. Five interns from this group reflected recently on their accomplishments to date and described their goals.
Ruth Hammond, a recent graduate of Oak Ridge High School, interns with computational scientist Arnold Tharrington to benchmark the performance of the OLCF’s supercomputers. Her internship in the Higher Education Research Experiences (HERE) program lasts until mid-August, when she will begin her undergraduate studies at the Rose-Hulman Institute of Technology in Terre Haute, Indiana.
Hammond’s project involves studying a “box” of water molecules using the Tip4p model and the LAMMPS Molecular Dynamics Simulator code on the new IBM AC922 Summit supercomputer. She aims to validate Summit’s performance by running simulations and comparing the results to those of the Cray XK7 Titan supercomputer and other HPC systems.
With the exception of a single high school class, Hammond had little computer science knowledge before beginning her internship. Since then, she has learned about the hardware components in supercomputers (CPUs and GPUs) and practiced using resources including the LAMMPS code, UNIX operating systems, and the CUDA programming model.
Hammond plans to pursue a major in biomedical engineering and a minor in computer science. The resources and personnel at the OLCF have helped her explore the intersection of these interests and provided her with hands-on experience before she sets foot on her college campus.
“Being exposed to the systems here has been amazing,” she said. “I get to walk by Summit every day on my way into the office, which is just about the coolest thing ever.”
Jack Hutchins has also spent the summer between his high school graduation and first semester of college at ORNL. Before graduating from Clinton High School, he earned 63 college credit hours and received an associate’s degree. He plans to study computer engineering when he enters the University of Tennessee, Knoxville, as a junior this fall. Currently, he helps support and improve everyday operations at the OLCF.
Using a program called Dash, Hutchins is developing a dashboard to monitor Titan and display key information such as the total power used during a specific period or the effectiveness of a certain application.
“I’m hoping to deploy the dashboard onto a web server so that any computer on the ORNL network can view it and see what’s going on with Titan at any given time,” he said.
Using Dash requires knowledge of the Python programming language, which Hutchins first encountered during his time as lead programmer for his high school robotics team. At the OLCF, he puts his coding skills into practice on a much larger scale.
Throughout his HERE program experience, Hutchins has become more proficient in Python, enhanced his knowledge of general computing concepts, and dabbled in machine learning. He enjoys speaking with ORNL staff, including his mentor, OLCF Operations Manager Stephen McNally, about their research and responsibilities as he embarks on his own HPC journey.
“The hands-on opportunities I have here make the learning curve a lot easier and more interesting,” he said. “The lab is a great learning environment that can’t be replicated in a classroom setting.”
Zhonghua Zheng leverages HPC resources to seek a better representation of atmospheric aerosols, which include extremely small and potentially harmful airborne particles. He interns with the OLCF’s Advanced Data and Workflow Group to search for more accurate and cost-effective methods of modeling this microscopic mixture of solid and liquid matter.
“My PhD research uses computer simulations, observational data, and artificial intelligence via machine learning to study the critical impact aerosols have on weather prediction, climate change, and human health,” said Zheng, a participant in the Advanced Short-Term Research Opportunity (ASTRO) program. “Even if the air looks clear, you are likely inhaling tiny particles every time you take a breath.”
Under the supervision of his mentor, computational climate scientist Valentine Anantharaj, Zheng implements deep-learning studies on NVIDIA DGX-1 machines. Insights from research in this vein could eventually help scientists mitigate environmental problems that these particles cause.
Zheng, a PhD student with Professor Nicole Riemer at the University of Illinois at Urbana-Champaign, studies environmental engineering with a concentration in computational science and engineering. He holds a master’s degree in agricultural and biological engineering from the same school and a bachelor’s degree in biosystems engineering from Zhejiang University in China. Zheng continues to broaden his knowledge by making professional connections with staff and interns from different disciplines.
“I’m fortunate to intern with the world-class scientists here, and I enjoy gaining a fundamental understanding of various research fields through dynamic conversations on a daily basis,” he said.
After earning his PhD, Zheng aspires to become a university faculty member to continue conducting research while also training the next generation of scientists and engineers.
Julita Inca Chiroque came to ORNL from Peru with a professorial mindset and extensive knowledge of Linux operating systems. After earning a master’s degree in computer science in Peru, she taught at the university level, worked as a Linux administrator, and volunteered for GNOME and Fedora, organizations that provide free software. She has won multiple international scholarships, one of which rewarded her with the opportunity to present her HPC research at the ISC High Performance Workshop in 2016.
As an ASTRO participant over the past 5 months, she has focused on two projects with mentorship from Arjun Shankar, who leads the OLCF’s Advanced Data and Workflow Group and directs the Compute and Data Environment for Science (CADES) at ORNL.
First, Inca improved HPC user documentation guides by conducting a user survey, compiling information and examples, and writing more integrated and intuitive content. She primarily focused on the CADES Scalable HPC Condos guide, a resource now available to facility users online.
“When you have a solid understanding of technical topics, it can be difficult to simplify explanations for others with less experience,” she said. “It’s rewarding to address this challenge and see users exceed their own expectations.”
Second, her current focus involves using a system called FederationID in combination with OpenStack software solutions and OpenID Connect protocol to evaluate user responses after signing into ORNL computing systems such as the OLCF and CADES. Her pilot project aims to connect these separate systems with a single authentication process.
Additionally, Inca hosted a workshop introducing fundamental tools and technologies in the field to ORNL staff and interns new to HPC. “I collaborated with other interns who have expertise in topics like deep learning, parallel computing, and Python to help newcomers get their footing,” she said. In September, Inca will begin a master’s program in HPC at the University of Edinburgh.
Whitney Nelson, a participant in the GEM fellowship program, is building a prototype for an online data catalog to accelerate scientific discovery through data analytics. This resource could not only provide large amounts of data necessary for artificial intelligence applications but also guide simulations on OLCF supercomputers and experiments at other ORNL facilities.
“The catalog is Linux-based, so I’m doing a lot of reconfiguring. Sometimes I write Python scripts or complete other miscellaneous tasks, but mainly I’m looking into how we can make this framework suitable for work done at ORNL,” she said.
Nelson relies on CADES infrastructure for this project. Although she consults with her mentor, the OLCF’s Suhas Somnath, to discuss her plans, Nelson leads the project and makes key decisions along the way, which she considers to be one of the most exciting aspects of her internship.
“Because this product is a prototype, I get to pitch a lot of my ideas and enhance my leadership skills,” she said.
Nelson recently graduated with a bachelor’s degree in computer science from Hampton University, and she will begin a master’s program in human–computer interaction at Georgia Tech this fall. She wants to design technologies for educational purposes, a combination she wants to see implemented more often.
Other students interning at the OLCF this summer were Hazem Abdelhafez, Matthew Bachstein, Aaron Barlow, Tyler Beichler, Cade Brown, Cooper Colglazier, Noah Crum, Sajal Dash, Alfred Farris, Shubhankar Gahlot, Thomas Hill, Ronan Hix, Harsh Khetawat, Monika Kodrycka, Seth Maxwell, Nelson Robbins, Tyesha Ruffin, Luke Schlabach, Jesse Vomfell, Amil Williamson, and Matthew Wilson.
ORNL is managed by UT-Battelle for the Department of Energy’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit https://science.energy.gov.