Algorithmic warfare: DARPA investigates quantum computing capabilities
A GTRI researcher working on a quantum computer.
Sean McNeil / GTRI Images
The Defense Advanced Research Projects Agency recently funded the second phase of a quantum computing project that aims to expand the use of emerging technology, according to one of the project’s lead researchers.
The second phase of the project led by the Georgia Institute of Technology Research has received $9.2 million in funding for scientists to conduct additional experiments on a quantum computing system configured to connect more computing units than ever before.
The DARPA project – Optimization with Noisy Mid-Range Quantum Devices – aims to “demonstrate the quantum advantage of quantum information processing by overriding the performance of only classical systems in solving optimization challenges.”
Researcher Kriston Herold said that one of the classic problems of optimization that quantum computing systems can solve is called the mobile salesperson.
“One well-known issue is the mobile salesperson problem, where you have a list of addresses you need to take a route and packages for delivery, for example,” he said. “And you want to find the most efficient path, whether it’s in time or distance traveled, the fewest turns left, or at least the gas used.”
He noted that this type of problem appears in a variety of logistical issues in defense and other government business.
Quantum computers use basic units known as qubits instead of 1 and 0 like traditional computers. Its computing power stems from the possibility that all qubits are 1 and 0 at the same time, rather than being limited to one or the other. As a result, a quantum computer can run more complex algorithms and run much faster than a conventional computer.
Herold explained that this research aims to go beyond most advances in quantum computing that have been made so far. Quantum computers exist today, but they are as large as ancient classical computers and have not yet developed computing power to compete with their classical counterparts.
While most quantum computing systems use magnetic traps to isolate ions, one of the team’s researchers, Brian McMahon, has developed an improved “unique” configuration for a more efficient process.
The trapping process – called a Penning trap – uses a combination of magnetic and electric field to trap two-dimensional ion crystals that perform quantum processes.
“Rare earth elements are actually used in permanent magnets, which make up the trap,” Herold said. There are magnets such as neodymium or samarium cobalt. They are very, very strong magnets.”
The trap uses these rare-earth metals instead of “massive cryo-cooled superconducting magnets,” according to the team.
The team has already conducted 18 months of trials and experiments. During that time, researchers built an ionic chain with a length of 10 qubits. A qubit is one of the smallest units in the quantum computing system.
Herold said that building the foundation of research using the short string is a start to research, but that it will eventually go much further than that.
“It was really about testing the control system and showing that this way of operating the device would solve these issues as expected,” he said.
Herold said adding thousands of quantum systems to the chain would cause the computer to compute more precise solutions. Without significantly more systems added, he said, a quantum computer would have roughly the same power as a classical machine.
“At the start of the project, we knew we would need hundreds of qubits to really move the needle to solve an important problem,” he said. “We can still simulate everything that happens on a quantum device, which is too small to attack an optimization problem large enough that we don’t know the answer easily.”
But this does not mean that traditional computing does not play a role in the project. Researchers use classical computing hardware to guide quantum devices to a better starting point, so the system doesn’t have to check every possible solution.
“The classical nature of this is that we use a classical process to sort of observe quantum gear and decide what to do next,” Herold said.
Although the project is promising so far, researchers still face daunting technical challenges. For example, the more complex a quantum system becomes, the more likely it is that a large error rate will occur due to “noise” – a term meaning interference with the state of qubits in a quantum computer.
The research team includes scientists at Oak Ridge National Laboratory, who use a supercomputer there to map the best noise reduction path in the quantum system as it scales.
“With quantum devices, we’re always fighting noise, and at some point, there’s going to be so many bugs that we can’t actually make the devices bigger,” Herold said.
He explained that while part of the research is to find how to mitigate errors, the amount of noise will ultimately limit the number of bits that the string will length and thus the complexity of the system.
However, if researchers can come up with solutions to these challenges for experiments, the results will be important across industries, Herold said.
“This project will show that larger sets of qubits can solve optimization problems in a better way than we know now, and that will have a really transformative impact on the way these problems are solved,” he said.
Threads: Information Technology