UNIVERSITY PARK, Pa. — The 2020 cohort of Penn State’s Institute for Computational and Data Sciences (ICDS) Faculty Fellows — formerly known as faculty co-hires — includes an interdisciplinary group of researchers who will use the University’s computational resources to probe the world’s biggest societal and social challenges. Their expertise ranges from investigating the inner workings of quantum mechanics to understanding the creation of galaxies. They seek to answer questions as broad as why do we exist and how can we exist more sustainably — to how can we make sure our data is more secure and create artificial intelligence that is fair?
“ICDS collaborates with colleges and departments across the University to recruit and support a cadre of faculty fellows — talented and creative researchers who are doing data- and computationally-intensive research that spans the disciplinary landscape” said Jenni Evans, professor of meteorology and atmospheric science and director of the Institute for Computational and Data Sciences.
“Penn State’s exceptional computational and data sciences community continues to advance with the arrival of this group of creative interdisciplinary researchers,” said Evans. “While this class are from different disciplines to many of the current ICDS Faculty Fellows, they are united in the belief that interdisciplinary research has the potential to unlock science’s biggest mysteries and to solve society’s biggest challenges. It’s our mission at ICDS to offer them, and all Penn State faculty whose research requires it, the computational tools, capabilities and resources to fulfill that mission.”
Here are this year’s ICDS faculty fellows and a look at their research areas — and where this work could lead.
"Helping Science Take Quantum Leaps"
Fodor is a professor of physics whose work is focused on better understanding lattice quantum chromodynamics (QCD) and investigating QCD at non-vanishing temperatures and densities. His research interests also include the study of elementary particle and nuclear physics, as well as fundamental quark and hadron properties.
Fodor said that he is interested in basic, but vexing, mysteries of our world and universe.
“The goal of my research is to understand basic puzzles: why do we exist, where does mass come from, what happens to nothing if we heat it up, etc.,” said Fodor.
Fodor said collaborating with ICDS will not just offer access to computational resources, but the partnership will also help him broaden his vision. Taking that vision and adding the power of the ICDS’s Roar supercomputer could lead to new insights into elementary particles, he added.
“One of the projects I’m hoping to explore is to solve a long-standing puzzle about the magnetic field around elementary particles,” said Fodor.
"Exploring Galactic-Sized Questions"
Leja is an assistant professor of astronomy and astrophysics and is interested in understanding one of the great mysteries of science — how do galaxies form?
“Galaxies are the cosmic engine through which the universe converts primordial gas into stars, and thus into heavy elements, planets, and eventually, very far down the line, things like humans,” Leja said. “The formation and growth of galaxies is the first step in that process, and it occurs on the grandest scale. How do galaxies form from the cosmic void? What processes occur to transform them from chaotic explosions of gas and energy in the early universe, to grand, well-structured spiral galaxies toady? How can we use these bright candles in the distance universe to understand how the universe evolves as a whole — i.e., cosmology?”
Leja said that his work relies on large surveys and statistics — and that requires a lot of computing power.
“I specialize in fitting flexible models to galaxy photometry and spectra, in building and exploring analytical and theoretical models of galaxy evolution, and in learning about and applying astrostatistics to big problems in galaxy evolution,” Leja said. “I've harnessed millions of hours of supercomputer time running specialized code to build a more complete picture of how galaxies form and evolve.”
This is where ICDS comes in.
“ICDS is full of cutting-edge researchers using advanced computational and statistical methods to analyze their — often very large — data,” Leja said. “I am very much looking forward to learning how my fellow researchers efficiently sort, analyze, process, and disseminate their findings using computational techniques, and learning from this process.”
He added that working with ICDS’s Roar supercomputer will help him explore those galaxy-sized scientific questions.
“Analyzing a single galaxy to understand the gas, stars, dust, and black hole content takes about 20-25 hours on a single core,” said Leja. “Modern surveys will detect about 200,000 galaxies today, and about 200 million galaxies in ten years. The computational resources of ICDS-ACI will be key to being competitive in this field, and collaboration with the ICDS will be critical to develop the new computational tools absolutely necessary to analyze the surveys of tomorrow.”
In the future, Leja said he and his team will access data from the Subaru Prime Focus Spectrograph survey, a massive international spectroscopic survey of galaxy evolution expected to be available in the next year or two.
“I am excited to use these data in the upcoming years to measure stellar population properties of galaxies as a function of cosmic time, including stellar masses, star formation histories, chemical abundances, and dust contents, on a massive scale,” he said.
"Imagining — and Imaging — New Planets"
Czekala, an assistant professor of astronomy and astrophysics, studies how planets form using observations with radio telescope interferometers like the Atacama Large Millimeter Array (ALMA), located in Chile. These telescope arrays, like the VLA featured in the movie Contact, use a network of interconnected antennas to improve image resolution. ALMA can directly detect the cold gas and dust in protoplanetary disks. Czekala uses molecular tracers like carbon monoxide to build a three-dimensional representation of the disk structure and velocity field.
With the recent generations of observations, we don’t just observe the disk rotate, but can actually sense the ways in the velocity of the field deviates from its expected rotation at a very subtle level.
“It’s like watching a river flow downstream. At first you see the bulk flow, but when you look beyond at the eddies and turbulent wakes you can infer that there might be a submerged rock,” said Czekala. “In this case, that rock might be a newly formed protoplanet.”
As an interferometer, ALMA uses 66 antennas spaced up to 15 kilometers apart to effectively mimic a telescope with a 15-kilometer diameter, with some complications for image fidelity. Czekala is working on developing new algorithms to push the level of inference possible from large and multifaced datasets using GPU processing power. A lot of the image processing questions tend to crop up in other fields.
“There’s a lot of natural collaborations that I’m looking forward to starting within ICDS,” he said.
"Making Data Safe for Democracy"
Hu is a professor of law and of international affairs. She also serves as Penn State Law's inaugural dean for non-JD programs.
Hu’s research is focused on the intersection of immigration policy, national security, cybersurveillance and civil rights.
In the past, Hu served as special policy counsel in the Office of Special Counsel for Immigration-Related Unfair Employment Practices (OSC), Civil Rights Division, U.S. Department of Justice. She was also a former senior policy adviser in the Obama Administration.
Hu has published several works on dataveillance and cybersurveillance and is currently working on “The Big Data Constitution,” a forthcoming book being published by Cambridge University Press. She is currently a member of the advisory board of the Future of Privacy Forum, a nonprofit think tank in Washington, D.C., that promotes responsible data privacy policies.
"Shining Big Data’s Focus on Star Power"
Villar will be joining Penn State in 2021 as an assistant professor of astronomy and astrophysics. She is the Ford Foundation Dissertation Fellow at Harvard in the astronomy department.
Villar's data-intensive approach to investigating how stars die through explosions and collisions.
“In particular, I take a data-driven perspective and use machine learning to ‘classify’ these events — in other words, I train neural networks to identify the underlying physical processes,” Villar said. “Additionally, I study so-called "multimessenger" astrophysics. Specifically, I study the light that comes from colliding neutron stars. called kilonovae.”
Villar said big data techniques are becoming a necessary part of work, one reason she’s excited to be part of the ICDS team.
“We discover roughly 10,000 stellar explosions (called "supernovae") each year,” Villar said. “But a new telescope, called the Vera Rubin Observatory, is coming online which will increase this discovery rate by two orders of magnitude! We will need to use cutting edge machine learning techniques and high-performance computing to quickly and accurately classify these events in real time in order to actively hunt for new astrophysics. In a way, we're solving a really generic problem — ‘How do we study driving mechanisms of noisy sparse time series?’ ”
Villar is also glad to be part of ICDS’s interdisciplinarity.
“I am so excited to bring in ideas from other fields through ICDS,” she said. “On top of that, ICDS’s Roar supercomputer will be necessary to handle this deluge of data, both in realtime and for archival statistical studies.”
"Driving Materials Design and Development for Energy Innovation"
Reinhart is an assistant professor of materials science and engineering who uses a data-driven approach to materials design and discovery.
“This means I use physics-based simulations to understand the way materials behave at the micro-, meso-, and macro-scales and couple that with machine learning to find new and improved materials for a variety of applications,” said Reinhart. “These can include new metals for additive manufacturing, semiconductors for optical computing, or even new building materials for homes or transportation infrastructure.”
This investigation into materials design and discovery could lead to practical applications, particularly in the energy sector, such as new battery chemistry, more efficient solar cells and improved metal manufacturing processes.
While Reinhart plans to use the ICDS Roar supercomputer, some of his work may also be aimed at improving supercomputing itself.
“I also hope to improve the sustainability of the built environment with improved building materials and create more energy-efficient super-computing platforms with optical computing meta-materials,” he said.
Interdisciplinarity is another benefit as an ICDS faculty fellow.
“My work is interdisciplinary by nature — I make use of state-of-the-art concepts from computer science and fundamental materials science, and borrow heavily from a variety of engineering disciplines, such as chemical, electrical, mechanical, civil, etc.,” Reinhart said. “So, I am excited to share a common space with computational and data science researchers from a broad cross-section of the research community here at Penn State to be exposed to many new ideas and hopefully integrate them into my own research projects.”
To learn more about the ICDS faculty fellows program, visit the ICDS website.