Quantum computers have the potential to revolutionize the way scientific problems are solved. They are expected to excel in certain tasks, such as simulating quantum systems, beyond what is possible with classical computers.
However, in their current state, quantum computers are too noisy to reach this potential. Noise can arise from environmental factors such as stray electromagnetic fields or material defects. This noise can cause errors in calculations.
Pacific Northwest National Laboratory (PNNL) scientists created a way to make sure that quantum calculations are less noisy.
“The success of quantum computing in the noisy intermediate scale era depends upon noise mitigation” said Ang Li, co-author of the research nominated for a “Best Paper” award at the 49th Annual International Symposium on Computer Architecture (ISCA 2022).
Using ensembles against noise
PNNL research associate Samuel Stein is no stranger to quantum computing. As a Master’s student at Fordham University, Stein noticed that some quantum computers performed certain tasks better than others. Not only that, but their performance would decrease depending on how long it had been since they were calibrated.
“When researchers use quantum computers, they generally don’t know the status of the computer, such as how long it has been since calibration or if this particular computer would work well for their algorithm over another potentially less busy processor” said Stein, lead author of the publication.
Once he arrived at PNNL, Stein used his experience to make quantum computing better for researchers. He partnered with Li, as well as PNNL researchers Bo Peng, Karol Kowalski, Nathan Baker, and James Ang, PNNL joint appointee Nathan Wiebe from the University of Toronto and Yufei Ding from the University of California, Santa Barbara to accomplish this.
Together, they created Ensembled Quantum Computing (EQC)—a way for multiple quantum computers to come together to overcome noise and improve variational quantum algorithm performance.
With EQC, a so-called “master node” observes multiple quantum computers and distributes jobs to different computers through “client nodes”. The client nodes measure the quality of the output of their assigned quantum computer and report back to the master node. The master node then re-evaluates the task distribution and optimizes it based on the computers’ current state.
“It basically acts as a load balancer,” said Stein. “When one computer’s performance starts to decrease, the master node shifts the work to a more performant computer. Then when the first computer has recovered or is calibrated, it can shift the work back.”
When performing these types of distributed calculations with EQC, the team observed a substantial increase in speed and accuracy with variational quantum algorithms.
“Though EQC isn’t perfect, it offers a promising look into a future where quantum computers can cooperatively solve problems,” said Li. “This way, we can maximize the fidelity and performance of noisy devices.”
This research was supported by the Department of Energy (DOE), Office of Science, National Quantum Information Science Research Centers, Co-design Center for Quantum Advantage, the DOE, Office of Science, Office of Basic Energy Sciences (BES), Division of Chemical Sciences, Geosciences, and Biosciences, and the National Science Foundation.