The science needed to solve the problems of a rapidly changing society will spring from the ideas and talents of people with diverse backgrounds and skill sets.
As part of Carnegie Mellon University’s ambitious future of science initiative, researchers in the Mellon College of Science and the university’s other six schools and colleges are focused on eight areas ripe for world-changing breakthroughs accelerated by bringing together Tartans’ expertise in the foundational sciences, artificial intelligence (AI), engineering and data analytics.
“Driven by interdisciplinary research and the convergence of disciplines, the Mellon College of Science is transforming scientific inquiry,” said Rebecca Doerge, Glen de Vries Dean of MCS. “Our scientists are harnessing vast new sources of data and computational methods to accelerate exploration.”
Here are a few of the scientists asking and answering critical questions for the next generation of discovery.
Computational Finance: Collaborative Risk Taking
Martin Larsson’s research in mathematical finance connects to areas as diverse as hedge fund strategies, interest rate securities and energy finance.
“Mathematical finance combines interesting math with a broad range of applications,” said Larsson, a professor in the Department of Mathematical Sciences. “Finance is about sharing risk and making efficient use of limited resources. Mathematics is essential for understanding those mechanisms.”
This research also feeds into education. CMU’s graduate and undergraduate degrees in computational finance blend theory and computation into a curriculum that prepares students for active roles in the financial industry.
“A lot of what’s going on in industry would not have been possible without conceptual breakthroughs in the ’60s, ’70s and ’80s,” Larsson said. “But those ideas must be operationalized, and this requires computation. Financial firms rely on staggering amounts of computational resources.”
CMU’s Master of Science in Computational Finance program is jointly managed by the Mellon College of Science, the Dietrich College of Humanities and Social Sciences, the Tepper School of Business and the Heinz College of Public Policy and Information Science.
“There is a long history of cross-disciplinary collaboration in this space. That is very much consistent with the ethos of the future of science initiative,” he said.
Cosmology: Universal Collaboration
Advances in space- and ground-based telescopes and gravitational-wave detectors have ushered in a golden age of astronomical data.
But making sense of the vast amount of data being collected on hundreds of millions of stars requires a team effort.
“No one can be an expert in cosmology, chemistry, detector physics, machine learning, statistics. You need five separate people, all working together,” said astronomer Katie Breivik. “When collaboration happens, you get magic way better than the sum of the parts.”
Breivik, the McWilliams Center for Cosmology’s newest faculty member, studies how binary stars evolve from birth to death. Binary stars — two stars orbiting a common center of mass — impact almost all aspects of astronomy, from shaping galaxies to making black holes. But they are tricky to study.
“There are no surveys that target binary stars only. We’re taking what we can get from what’s already out there,” Breivik said.
Breivik teases useful information from datasets on stellar objects like black holes, stellar cores and normal stars to develop simulations of binary star interactions. Along with machine learning experts, she combines these models with data from gravitational-wave and electromagnetic surveys.
“The McWilliams Center makes it possible for people who have disparate expertise to talk to each other in a comfortable space, and that’s really critical,” Breivik said.
Life Sciences Breakthroughs: Synthetic Organelles
With explosive growth anticipated in genomic data, Huaiying Zhang, an assistant professor of biological sciences, is combining her engineering background with an interest in cellular structures.
Her lab examines the phase transitions in cells that lead to division, and she is working to create synthetic organelles within cells.
“We use engineering approaches where we are trying to study these cell compartments,” Zhang said. “If I cannot create it, I don’t understand it, so we’re creating it for the sake of understanding it. You can control the chemistry in cells and therefore control their behavior.”
Zhang controls cells by injecting them with photosensitive chemicals, which respond to different parts of the light spectrum. Once she captures images of the cells’ reactions, she feeds the information into an AI program she created with the help of researchers in computational biology to further understand what changes in the cells.
“That’s the future of imaging,” Zhang said. “We rely a lot on our eyes for most analysis and some automation to obtain information. AI is more powerful than that. We have all these movies from experiments — terabytes of data that we have already invested in. AI can mine that information for future discoveries.”
Mathematical Foundations of AI: Theorem Developments
Jeremy Avigad, a professor of philosophy and mathematical sciences and director of the Charles C. Hoskinson Center for Formal Mathematics, is working on ways to use computers and AI to support and develop mathematics.
Avigad and others at CMU have been using a programming language and proof assistant known as Lean, developed at Microsoft Research, to help mathematicians reason more reliably and precisely.
“Lean allows you to write mathematical arguments in such a way that a computer can understand and check it and verify that all the steps are correct,” Avigad said.
In the future, Avigad said that he hopes that the program will evolve so mathematicians at all levels can check proofs in real time.
“This has the potential to be a really transformative technology for mathematics,” Avigad said. “In the last century, mathematics has gotten more complex. Arguments have become more complicated. Mathematics needs human reasoning and thought and reflection, but there are also things that computers can do extremely well. As we learn how to leverage those things and take advantage of them, I think it’ll really improve our capacity to do mathematics.”
Materials of the Future: Crunching Bits into Atoms
The design and development of new materials is crucial to improving materials, drugs, new catalysts and more. Which is why Gabe Gomes and other CMU researchers are transforming chemistry as a discipline.
“We are making the field as we are going. That is really, really awesome and terrifying,” said Gomes, an assistant professor in CMU’s Departments of Chemistry and Chemical Engineering. “The future is this amalgamation of digital and physical.”
Gomes was named one of 2022’s Talented 12 by Chemical and Engineering News. His lab group leverages computational strategies to address problems in chemical reactivity; and he and CMU’s Olexandr Isayev are part of the Center for Computer Assisted Synthesis, a multi-institution team funded through a $20 million grant from the National Science Foundation that will bring together chemistry and machine learning to advance molecular synthesis.
With the impending arrival of the world’s first academic cloud lab at CMU, Gomes said he expects cloud experimentation will mature within the decade.
“We have to educate our chemists and chemical engineers with a strong computational foundation. It is no longer nice to have — it’s a requirement. The cloud lab allows us to do that,” he said. “It changes how we think about problems, and I love that. It will be transformational for all of us going forward.”
Neuroscience: Sense and Sensibility
Kate Hong, an assistant professor of biological sciences and a member of CMU’s Neuroscience Institute, investigates how the brain mediates Sensory-guided behavior.
During her postdoctoral work, Hong discovered that even when the primary somatosensory cortex, which processes information about touch, is removed, mice rapidly recover their ability to detect objects. Animals without a somatosensory cortex also were able to learn a tactile detection task as fast as normal mice.
“The ability to detect objects is not unique to animals with a neocortex. We take the view that the cortex evolved to modify preexisting subcortical sensory and motor circuits,” Hong said. “Uncovering how the cortex modulates subcortical brain areas during behavior is the key to understanding how the brain processes sensory information.”
Hong said investments in the Neuroscience Institute and the school’s collaborative atmosphere drew her to CMU. Hong works with researchers in the Electrical and Computer Engineering and Biomedical Engineering departments to analyze and model data using information theory.
“The field is continuously developing novel strategies to make sense of large populations of neural data,” Hong said. “Having close collaborators at the forefront in developing these strategies allows us to extend our research beyond what we would be capable of on our own.”
With these new techniques, Hong aims to better understand how sensory information flows along multiple, interconnected networks, and further, how these networks may rapidly recover from cortical injury.
Sustainability Science: Harnessing Hydrogen
In the quest for a clean energy future, hydrogen is a potential game-changer. But hydrogen fuel is costly to produce. Devices that split water into hydrogen and oxygen rely on platinum — an expensive precious metal — making large-scale use of this approach a major challenge.
Associate Professor of Chemistry Kevin Noonan is meeting the challenge head-on. His lab is part of a Department of Energy Frontier Research Center, which brings together a multidisciplinary team to tackle the toughest scientific questions preventing advances in energy technologies.
“My group collaborates with chemists, physicists and engineers to build next-generation electrolyzers and fuel cells,” Noonan said. “We’re each designing a different component, but we need to put everything together. That’s where cross-pollination is key.”
Noonan is developing materials for membranes, a critical component in electrolyzers — devices that use electricity to split water into hydrogen and oxygen. The membrane conducts ions between two electrodes, facilitating the water splitting reaction.
Noonan is making membranes out of hybrid organic/inorganic materials using elements like sulfur and phosphorus and running the reaction under basic conditions, an approach that can help move away from precious-metal-based electrocatalysts and dramatically lower the cost of clean hydrogen fuel production.
“We’ve spent a lot of time talking to other people about how to design the membrane more efficiently, how to model it, how to study it,” Noonan said. “It truly is a cross-disciplinary endeavor.”
Quantum Information: Foundational Exploration
Jyoti Katoch, an assistant professor of physics, and other CMU researchers investigate two-dimensional materials. Unusual physical properties arise when two graphene sheets, just a few atoms thick, twist to a “magic angle.”
“What you get is superconductivity in this material, which is mind-blowing,” said Katoch, who studies electronic, optical and spin-dependent properties of novel quantum materials. “These materials can help us understand a lot of fundamental questions in physics. We also are learning how these materials work and how to take them a step closer to real-world applications.”
Those could include space exploration, consumer electronics, biosensing equipment and more.
Katoch’s group probes materials’ electronic band structure using NanoARPES, or a Nano angle-resolved photoemission spectroscopy. With a handful of research institutions having the capability, Katoch partners with Diamond Light Source, the United Kingdom’s national synchrotron, and the Advanced Light Source research facility at Lawrence Berkeley National Laboratory in California.
Along with CMU’s top-notch Claire and John Bertucci Nanotechnology Laboratory and collaborations with the Pittsburgh Quantum Institute, the Department of Physics uses interdisciplinary approaches to explore potential breakthroughs.
“Data science and machine learning can fast-track the discovery of quantum materials by identifying virtual systems with desired topological properties,” she said. “They also help analyze the four-dimensional datasets that we generate. This work can have a revolutionary impact on future technologies and bring about the next quantum revolution.”
■ Kirsten Heuring, Amy Laird, Heidi Opdyke