University of California, Berkeley
Hannah Klion attended a small language-immersion school while growing up in Indianapolis, but also inherited her psychologist father’s tinkering tendencies. She learned fluent Spanish and conversational French and joined the school robotics team.
Klion tested the idea of pursuing a STEM degree during the Summer Science Program, an intense six-week course in Santa Barbara, California, between her junior and senior years of high school. She tracked the positions and orbits of asteroids and attended lectures, including one in computational astrophysics. “I don’t think I really understood the power of simulations beforehand,” Klion says, remembering plots of accretion disks that encircle black holes. “It was really cool to see how you could use first-principles physics to simulate some complicated processes that we can’t even observe directly.”
In 2011, Klion headed to the California Institute of Technology and sampled computational research in both astrophysics and particle physics. She settled into exploring the first 25 milliseconds or so of violent explosions of massive stars: rapidly rotating core-collapse supernovae. That experience helped establish the niche for her Ph.D. research, which is supported by a Department of Energy Computational Science Graduate Fellowship (DOE CSGF).
Klion explores high-energy phenomena on the scale of stars, involving objects such as black holes, neutron stars and white dwarfs. These systems have complicated physical properties that interact in interesting ways, she says. “It’s just a regime that doesn’t exist on Earth and it’s not a regime that we can directly do experiments on.” Astronomy primarily relies on understanding objects and interactions from information transmitted within packets of light, or photons. With simulations, Klion can set up experiment-like scenarios, letting her tweak parameters and see what happens. She’s particularly interested in understanding astronomical changes that occur relatively quickly – on a human timescale.
Working with Eliot Quataert at the University of California, Berkeley, Klion has examined rotation in red giants, the objects that launch supernovae. More recently she has simulated neutron star mergers, when these supremely dense objects spiral into each other and fuse violently, producing gravitational waves, ejecting and attracting matter around them and launching a jet of gamma rays.
Klion has been using a model of neutron star mergers to simulate how light moves through the resulting stellar material, aiming to predict the light signatures Earth-based astronomers should observe from these cataclysmic celestial events. Astrophysicists first observed a neutron star merger in August 2017, but not until 12 hours after it appeared. Klion focuses on earlier times, up to three hours after an explosion. So far she’s concentrated on how the jet interacts with the stellar material, and she wants to study the ultraviolet light signatures the intense heat of these explosions produce. She did much of this work on Cori, the Cray XC40 supercomputer at the National Energy Research Scientific Computing Center, a DOE Office of Science user facility.
During her 2017 Oak Ridge National Laboratory practicum, Klion reworked part of the message passing interface (MPI) code within FLASH, a high-performance computing (HPC) application for supernovae and other high-energy physics problems. Calculating self-gravity within stellar objects and nuclear burn requires lots of communication between processor nodes. The code had run these tasks sequentially, slowing large simulations on HPC systems. Working with Bronson Messer, Klion improved the code’s multitasking, handling communication for self-gravity calculations in the background while computing the nuclear burn and then returning to finish the self-gravity results.
Klion enjoyed the opportunity to “just focus on HPC for a couple of months and really get a better understanding of MPI,” a standard HPC tool. She worked with Titan, the Cray XK7 at the Oak Ridge Leadership Computing Facility and Summitdev, the test platform Summit, the facility’s newest supercomputer.
The time focusing on HPC alone has helped Klion think about the code she develops for her Ph.D. work, improving both performance and usability. “I’m able to be a little more HPC-oriented with that now, which is nice,” she says.
Klion is interested in entering academia after graduating in 2021, but also is considering other options, including national laboratory posts. She spends her free time among the stars and during college knitted a shawl with a celestial map of the Northern Hemisphere. She also coordinates Astro Night, a series of astronomy outreach events at Berkeley that attracts dozens of visitors for talks and star gazing.
Image caption: This two-dimensional simulation shows a jet launching from within material ejected from a neutron star. This is the starting point for Hannah Klion’s radiation transport simulations. Credit: Paul Duffell.