Dr. Tim Zuehlsdorff is using supercomputers to do once-impossible experiments. As the leader of one of six teams recently awarded support from the Huang Complex Supercomputing Seed Funds, he and his team are designing intricate simulations to deepen our understanding of quantum dynamics. “What we’re really interested in is what happens to a molecule once you excite it, once you hit it with light,” Dr. Zuehlsdorff says. “Where does this energy go? How does it relax? How does it distribute into vibrations in the molecule, or trigger chemical reactions, or any of these kinds of things?”
The Huang Complex Supercomputing Seed Funds were created to fund Oregon State University research projects involving the Huang Complex NVIDIA supercomputer. The six projects range from improving nuclear fusion to breaking down the physics of breaking waves.
Dr. Zuehlsdorff and his team are specifically studying the wave function of porphyrin molecules. These molecules, related to chlorophyll in plants, play a role in how living things harvest light for various uses. Porphyrin molecules have a lot of applications in energy production, medicine, and agriculture. For example, Dr. Zuehlsdorff notes, molecules related to porphyrin can be used in solar cells. However, scientists don’t yet completely understand how light affects porphyrin molecules on a quantum level. That’s where Dr. Zuehlsdorff and his team come in.
The short-term goal of his research, Dr. Zuehlsdorff says, is to provide a clear explanation for the ways that molecules interact with light. By learning how molecules respond to light on a quantum level, scientists and engineers can build better solar panels, sensors, and other photovoltaic technologies.
For the team to get the answers they’re looking for, they need to store the porphyrin molecules’ wave function in a specialized way. “The wave function is this really high-dimensional object that you can’t even really store on a normal computer for anything that has more than a couple of electrons in it,” Dr. Zuehlsdorff says. To store the function efficiently while not losing any critical details, Dr. Zuehlsdorff and his team use tensor networks. A tensor network is, in this context, a mathematical representation of the wave function that preserves complexity without being impractical to evaluate. “It also happens to be that it can be evaluated very fast on modern supercomputers,” Dr. Zuehlsdorff notes.
To run their tensor network calculations, Dr. Zuehlsdorff and his team are using the Huang Complex NVIDIA supercomputer to perform incredibly powerful calculations called hero runs.
If normal computing power is like a typical athlete’s 50-meter dash, then a hero run is like an Olympian’s sprint. It lies at the extreme end of what a computer is capable of. To make a hero run possible, Dr. Zuehlsdorff needs a tremendous amount of computer power, nearly all of it dedicated to a single task.
“The idea with the new supercomputer and the hero run is to go way beyond what was possible previously,” Dr. Zuehlsdorff says.
Now that Dr. Zuehlsdorff’s team has made initial progress, their next move is to more from small-scale to large-scale computations, running faster and more robust calculations that can only be done on a supercomputer. Their results could have significant implications for future quantum research.
What makes quantum mechanics challenging for any scientist is that “it’s very, very difficult to simulate the full wave function of any object that isn’t a handful of atoms,” Dr. Zuehlsdorff explains. Supercomputers could change that.
“What we are trying to do is find efficient ways of simulating the full complexity of the quantum mechanical wave function on modern supercomputing architectures,” he says. “That’s the hope.”
