Securing Supply Chains with Quantum Computing

A leading idea for programming quantum optimization algorithms has involved coupling quantum computers and conventional ones to solve a problem together, called the variational approach. The conventional computer performs an optimization of control settings that dictate the behavior of the quantum computer.

One issue with this approach is that its impact is constrained by the ability of the conventional computer to solve optimization problems with a large number of parameters.

Sandia scientist Kenneth Rudinger, who also worked on the project, said the variational approach might not be practical when quantum computers finally become capable of living up to their promise.

“We have good reason to believe that the size of the kinds of problems you would want to solve is too large for the variational approach; at that scale it becomes essentially impossible for the conventional computer to find good settings for the quantum device,” he said.

New framework to solve intricate problems

The Sandia team succeeded in greatly reducing the role of classical computing. With the new framework, called FALQON — short for Feedback-based Algorithm for Quantum Optimization — the classical computer does not do any optimization. It only needs the computational power of a calculator, letting the quantum computer do all the heavy lifting and theoretically allowing it to work on much more complicated problems, like how to efficiently reroute a shipping fleet when a major port suddenly closes.

A framework, in this case, means a structure for how to write an algorithm. Sandia’s core concept is for a quantum computer to repeatedly adapt its structure as it moves through a calculation. Layers of quantum computing gates, the building blocks of quantum algorithms, are determined by measurements of the output of previous layers through a feedback process.

“After I run the first layer of the algorithm, I measure the qubits and get some information from them,” Magann said. “I feed that information back to my algorithm and use that to define the second layer. I then run the second layer, measure the qubits again, feed that information back for the third layer, and so on and so forth.”

Sarovar said, “It defines another class of quantum algorithms that operate through feedback.”

Until quantum computers become more powerful, the framework is largely a theoretical tool that can only be tested on problems classical computers can already solve. However, the team believes the framework shows great potential for formulating useful algorithms for the medium-to-large-scale quantum computers of the future. They are eager to see if it can help develop quantum computing algorithms to solve problems in chemistry, physics and machine learning.