The room is large, with half a dozen students spaced out from one end to the other. It’s overcast outside, and the lights are dimmed inside. The room is still with the hum of quiet whispers and tapping on keyboards; students are engaged with their computer screens, though they occasionally talk to the student at the next table over or call for a professor.
A casual observer might not suspect that those students were answering the questions of the universe and asking a few of their own at the same time.
This was the scene during NAU’s inaugural astroinformatics boot camp, a joint venture between the Department of Astronomy and Planetary Science (DAPS) and the School of Informatics, Computing, and Cyber Systems (SICCS). DAPS Chair David Trilling, SICCS assistant professor Michael Gowanlock and SICCS master’s student Daniel Kramer ran the bootcamp, which was funded by the Las Cumbres Observatory and the Heising-Simons Foundation with the goal of increasing diversity in astronomy- and computer science-related fields.
It’s an important goal, not only for NAU but also for these disciplines. Recent studies show that physical sciences and computer sciences are near the bottom in terms of bachelor’s degrees awarded to many underrepresented groups, and that means less diversity of ideas and in creative problem-solving. It also limits career opportunities in stable fields; according to the NSF, unemployment is lower for scientists and engineers than other occupations, and they’re on average higher-paying jobs. As a Hispanic Serving Institute, it’s part of NAU’s mission as well.
“By providing hands-on training to students from underrepresented backgrounds, we are helping to create potential career paths for students from Arizona and our region, as well as contributing to the growth of the national STEM economy,” Trilling said.
Seven students, who come from a range of majors and backgrounds, spent a week in August before the semester started immersed in big data—universe-sized data, in fact. They were working with images from large telescopes to study asteroids. The problem is that there is an enormous amount of data produced by these all-sky astronomical surveys. So the students made the universe smaller. They wrote and tested algorithms that measured asteroid properties, looked for unusual or changing asteroid behaviors and identified the most efficient computer science algorithms to employ.
These are the skills that the students—all seniors this year—will take into their job search or graduate career regardless of the field they pursue.
“One of the best things you can do, especially if you don’t know which career path you’re going into, is to just get as much experience as you can and try new things,” participant Michael Rocha said.
Seven students participated in the bootcamp:
- Kyle Young, a computer science major from Phoenix
- William Havas, an astronomy major from Redmond, Washington
- Savannah Chappus, a mathematics and animation major from Flagstaff
- Scott Austin, a computer science major from Southern California
- Charlie Saluski, a computer science major from the Phoenix area
- Loren Larrieu, an electrical engineering major from Flagstaff
- Michael Rocha, a computer engineering major from Phoenix
The bootcamp was open to all students who had taken certain courses, with an emphasis on groups who have been historically underrepresented in astronomy, data science and computing. And the organizers looked for ways to ensure those students could participate. Not every student can take an unpaid week off—so students were paid to participate. Not every student had all the needed skills—so the professors taught the skills, walked the students through the tasks and supported them while letting the students take the lead. The goal was to give every student the opportunity to do research and become stronger candidates when they graduate.
“The students learned some of the skills that encompass the role of a data scientist,” Gowanlock said. “This is currently one of the fasting growing jobs in industry, and in terms of other academic fields, the skills are highly transferable; a student from the bootcamp could apply what they learned to other disciplines, such as those in STEM, business and social sciences, among others.”
Astroinformatics and the art of ‘needle in a haystack’ science
This new field is the combination of astronomy and computer science. Trilling said astronomy is undergoing a revolution; an increasing amount of research is driven by massive all-sky telescopic surveys, including a new 10-year, billion-dollar effort beginning in 2024. They offer unprecedented views of the universe and a mind-boggling amount of data.
“These surveys produce so much data that no human could ever look at the data; instead, we need computational algorithms to analyze the data,” Trilling said. “So, this revolution is really being driven by technological advances that allow us to create such large-scale surveys and to analyze them through software.”
Astroinformatics, then, is necessary to put this large-scale survey data to use. Trilling said scientists have never known so
much about all the categories of “things” in the universe. They also provide the opportunity for “needle in a haystack” science by finding rare astrophysical objects from which astronomers can learn about extreme and unusual events in the history of the solar system and the universe.
This big data, of course, requires big computing. The students used Jupyter and Python to write and run code and used Monsoon, NAU’s supercomputer.
A day at bootcamp
Early in the week, the students split into three general groups, although their work split further. Chappus, Austin and Saluski looked at data available through current telescopes in preparation for the opening of a large-scale telescope in Chile. Using NAU’s SNAPS program—Solar System Notification Alert Processing System—they used a standard deviation model to look for outliers in the data.
This will help analyze data from the new telescope, with its 3.2-billion-pixel camera with the goal of understanding how our solar system formed.
“It would take 542 laptops to display one of the photos, so we’re not looking at the photos, we’re looking at the output data and figuring out how to find interesting objects for the public to then consume and go, ‘oh, there’s this interesting asteroid, I want to go look at it,’” Chappus said. “That’s the idea.”
Saluski was using the same data to look at changes in behavior that could indicate a collision in the asteroid belt. Scientists rarely see this happen, but they see evidence of it. Larrieu and Rocha’s work was looking at data that showed asteroids’ color—red means old, but blue could mean a collision happened, knocked off the older red crust and allowed for fresher ground.
Young was writing code to figure out which parameters, out of 32,000 combinations, showed correlation among asteroids. Correlation could give scientists information about whether an asteroid had a collision or what other features it might have based on what asteroids it was like or was not like. Havas used available data to locate asteroids in the sky and determine whether their features necessitated more research.
There were a lot of ups and downs—codes had errors, sometimes the algorithm worked but didn’t find the object, other times finding an answer just meant more questions. There also was the logistics of the work—this amount of computing required a massive amount of computer memory. But they kept going.
Twice a day, the group came together and presented their work for the day—their results, processes and stumbling blocks. Students and professors asked questions, challenged ideas and prompted ways to overcome difficulties.
“We all sat in a room for a week and worked together on these problems,” Trilling said. “Working side-by-side—especially during a week when there were relatively few outside distractions—really helped the students make a lot of progress, and I think they had fun doing it—and so did the instructors!”
Heidi Toth | NAU Communications
(928) 523-8737 | heidi.toth@nau.edu