California Institute of Technology
Even though he’s studied them for years, Sherwood Richers still has difficulty conveying the scale of gamma ray bursts (GRBs).
“It’s very hard to conceptualize,” says Richers, a Department of Energy Computational Science Graduate Fellowship (DOE CSGF) recipient. GRBs are focused beams of energy at the high-frequency end of the electromagnetic spectrum, visible only to orbiting gamma-ray telescopes. They come from distant parts of the universe and emit more energy in 10 seconds than our sun will in 10 billion years.
“It really is amazing how other-worldly these things are,” Richers says. “These are the brightest events in the universe, the most powerful explosions in the universe and, you know, they’re just cool. Things blowing up are cool.”
GRBs also are a mystery. Astrophysicists think they come from supernovae – exploding stars – and/or the collisions of neutron stars – the densely compressed remains left when massive stars collapse.
Because gamma-ray bursts come from cosmological distances, it’s difficult to observe their sources in detail. And, obviously, researchers can’t recreate GRBs in a lab. That means they must depend on simulations run on high-performance computing (HPC) systems to help understand these powerful phenomena.
The simulations, however, are so computationally demanding that researchers must heavily approximate the fundamental physics to make them tractable on HPC systems. Those compromises mean the calculations are less able to unravel what causes GRBs.
Richers’s research with Christian Ott at the California Institute of Technology targets neutrino transport, which astrophysicists believe is a key component of the supernovae and neutron star mergers that generate GRBs. Neutrinos are nearly massless subatomic particles that interact weakly with matter. Trillions pass through our skulls – and the rest of our bodies – each second with nary a hitch.
“But in supernovae and merging neutron stars, they’re very important,” Richers says, because in those conditions matter is so extraordinarily dense neutrinos can’t help but interact with it. Neutrino interactions, in fact, could be what drives GRBs. Yet “it’s a hard problem. Simulating any kind of radiation transport is very difficult. It’s even more so in the midst of an exploding star.”
While on his 2013 DOE CSGF practicum with Lawrence Berkeley National Laboratory’s Daniel Kasen, Richers worked on simulating neutrino transport through the discs of matter that remain after neutron stars merge. To do the job, he combined aspects of NuLib, a library of computer routines to calculate neutrino opacities, and Sedona, a Monte Carlo radiation transport program. Ott collaborated with former graduate student Evan O’Connor to develop NuLib; Kasen was a developer on Sedona.
Monte Carlo methods rely on random sampling to efficiently model phenomena and is especially useful for simulating particle interactions. The practicum project was intended as a one-time exercise, Richers says, but he realized that Monte Carlo methods could help overcome shortcomings in how neutrinos are treated in large simulations of supernovae and neutron star collisions.
The result is Sedonu, a code Richers created almost from scratch by rewriting and restructuring the original Sedona program. He’s lead author on a paper published in
“What I’m working on now is generalizing it – including more physics and making it faster” so Sedonu can be coupled to bigger magnetohydrodynamics codes, which track fluid flow under the influence of magnetic fields. “The bottom line is Sedonu works well, but there’s a lot more to be done” to incorporate it into those large simulation codes. “That work includes increasing the capabilities but also extensively testing it to make sure everything works as it should.”
The research, coupled with observations, could help constrain nuclear physics, like that governing reactors, but also help us understand how heavy elements like gold and platinum came into being.
Integrating Sedonu is the final project in Richers’ doctoral research. He recently received a Blue Waters Graduate Fellowship that will provide financial support – and supercomputing time – for what he hopes is his final year of studies. After a projected spring 2017 graduation, he’ll seek a postdoctoral research post – but he’s also pursuing a spot as an Air National Guard pilot.
“It’s a weird branching in my career, where either path might be possible,” Richers says. He earned his pilot’s license after joining Caltech’s flying club and now flies aerobatic maneuvers. Richers enjoys it, but admits most of his aviation is just about getting into the air. “Turning this hobby into something that’s valuable as an Air National Guard pilot would be very rewarding.”
The guard gig becomes part time after training, Richers adds, “so my ultimate dream, if I can manage it, would be to fly part time and do academics part time.”
He chuckles. “We’ll see if that’s possible.”
Image caption: This visualization shows a failed attempt to produce gamma-ray jets in a three-dimensional simulation of a rapidly rotating core-collapse supernova. Warm colors indicate magnetic strong fields and cool colors indicate weak ones. Rather than easily launching a jet, magnetic fields in 3-D simulations are wound tightly by the rotating core to form flux tubes that are crumpled in bipolar lobes. The simulation showed that a mechanism researchers thought might produce a jet doesn’t work after all. Previous two-dimensional simulations behaved completely differently, emphasizing that large 3-D simulations are vital to understand the inner workings of these supernovae. Image courtesy of Sherwood Richers and Philipp Mösta.