Capable of faster and more complex computing than a traditional computer, quantum computing is poised to quickly take machine learning and artificial intelligence to the next level.
Credit: University of Pittsburgh
Capable of faster and more complex computing than a traditional computer, quantum computing is poised to quickly take machine learning and artificial intelligence to the next level.
New work at the University of Pittsburgh blends physics with computer engineering to accelerate quantum computing capabilities while taking inspiration from something more familiar: a tree. The work recently received a $300,000 New Initiative Grant from the Charles E. Kaufman Foundation.
“One of the challenges of quantum computers is that the interactions tend to be noisy—there isn’t 100 percent fidelity,” said Alex K. Jones, professor of electrical and computer engineering at Pitt, who is leading this project. “The state changes create noise over time, so from input to output, it’s a race against decoherence, the loss of information. We’re working to create better gates so that the time for each operation is shorter, resulting in better error correction and higher fidelity.”
Qubits are the basic unit of information in quantum computing. Where binary code in computer science uses bits, either 1s or 0s, qubits function together in a system, like atoms, and can be entangled with other qubits. That means anything done to one qubit happens to the entangled ones, as well.
These properties make them much more powerful than bits—and much more complicated to work with.
To push quantum computing toward its full potential, Jones is partnering with Michael Hatridge, associate professor of physics at Pitt. They realized that in order to optimize the way these qubits talk to one another, the classic lattice structures used in IBM and Google’s quantum computers were limiting.
Instead, they are arranging the qubits in the shape of a tree, a methodology from classical parallel computer networks.
Jones and Hatridge are using a device called a SNAIL that allows them to create interactions between qubits as if they form elements, like “leaves,” on a tree, building a rich interaction space. In order for leaves on different “branches” to communicate, they must connect through the “trunk” of the tree, reaching out to their destination. With this SNAIL device, five or six qubits can interact with each other at the same time, opening the door for researchers to scale up this tree or other flexible approaches.
For instance, the team has proposed a “Corral” topology unique to both physics and computer science in a paper that will appear at the February 2023 High-performance Computer Architecture (HPCA) Conference in Montreal..
“We realized the tree structure and these novel structures like the Corral made it easier to move data around and opened richer computational space,” said Jones. “With this award from the Kaufman Foundation, we are looking at the interaction between qubits within individual modules. What can we learn about these nodes, and how do we pick the best computational interactions among them to advance the power of quantum computing so that each qubit can accomplish more than before?”
Noting that this is only the first of several in-dept collaborations in this area, Jones added,
“We’ve only just scratched the surface.”
The New Initiative Grants “encourage investigators with strong research records to establish interdisciplinary collaborations requiring expertise beyond that of any single researcher and take a novel approach to the topic in question,” according to the Pittsburgh Foundation’s news release. This is the 11th time a Pitt researcher has received this grant.
