This week, The Quantum Insider published a press release announcing IBM Quantum and UC Berkeley findings that suggest noisy quantum computers could outperform supercomputers at certain tasks.
They achieved this by showing that quantum computers can provide accurate and reliable results for challenging simulation problems at 127 qubit scale. The scale of this problem goes beyond what could be achieved with the most powerful classic computers.
According to IBM, research “shows that there are now only two options for solving quantum simulation problems at this scale: (1) classical approximation methods, which are more scalable but less accurate than precise classical methods, and (2) hardware quantum.”
To celebrate this achievement, IBM Research created a short film, Quantum Utility: The IBM Quantum Experiment and UC Berkeley chart a path to useful quantum computingpublished on YouTube on June 14 and filmed at the IBM Thomas J. Watson Research Center, Yorktown Heights, NY.
Several experts took part in the video. Here are some things participants said about the experiment:
Abhinav Kandala, Quantum Ability and Demonstration, IBM Quantum:
“So to be honest, I was quite nervous.
We ran extremely challenging experiments to explore the limitations of today’s quantum computers and how they compare to classical computing methods.
We wanted our collaborators at UC Berkeley to use classical simulation methods to verify whether our noisy quantum processor produces reliable results. We also like to push our devices to the limit where their methods might struggle. We don’t want them to come back to us too quickly and say, “Yeah, we solved the problem really easily.”
On scale, IBM is making more progress than anyone else, but noise is a different matter entirely.
We can only do this because we have now built quantum systems of unprecedented scale and quality and developed the ability to manipulate noise in quantum systems at that scale.
The way we control and manipulate hardware at that scale is unprecedented. The answer to this question did not exist before.
It will eventually reach a point where we can actually learn something new from quantum computing but we can’t access the classical alone. A point where we can really push the boundaries of human knowledge.”
Sajant Anand, Graduate Researcher, UC Berkeley:
“I try not to see this as a competition between classical and quantum computing, but I do this experiment in hopes that supercomputers outperform quantum computers.
So my environmental project is to help simulate these quantum experiments on some of the largest and most powerful supercomputers in the United States, including those at the Lawrence Berkeley National Lab, and the National Energy Research Scientific Computing Center.
Using some of the new air mitigation techniques that IBM has introduced, we actually found that quantum devices easily match exact class results and outperform classical approximation results in cases where the right answer is available. Impressively, even as we moved into circuits beyond the precise classical methods, quantum still seemed more reliable than some of the approximate methods we tried on these massive supercomputers.
Understanding the root causes of these differences will allow us to improve both methods and as examples of the benefits of interactions between quantum and classical. Together, we can realize tools that can solve interesting problems.”
Sarah Sheldon, Quantum Capabilities and Demonstrations, IBM Quantum:
“Having a good, reliable 127-qubit device with high coherence time is really a prerequisite. The hardware really made this project possible.
As we increase our number of qubits, reduce our error rates, and improve these techniques, we’ll see a growing set of problems where quantum computers can provide value.
Key takeaways from the experiment and the film:
- Quantum computers are very sensitive to noise, disturbances that cause errors in their calculations.
- For quantum hardware, scale and noise have always been the biggest problems. IBM has spent years developing methods to filter out the effects of noise from quantum computations. This is called error mitigation.
- IBM Research wanted to see if mitigating errors would allow quantum computers to match the accuracy of classical supercomputers at these scales. Using the error mitigation technique used in the paper, IBM runs its own calculations and slowly increases the noise, then extrapolates back to a no-noise solution.
Featured image: Credit: IBM Research