Today’s quantum computers are computationally small—the chips in your smartphone contain billions of transistors, and the most powerful quantum computers contain the quantum equivalent of a few hundred transistors. They are also unreliable. If you’re doing the same calculations over and over, they’re likely to come up with a different answer each time.
But with their inherent ability to consider many possibilities simultaneously, quantum computers don’t have to be very large to solve certain intractable computing problems. On Wednesday, IBM researchers announced that they had devised a way to manage unreliability in a way As follows: will lead to solid, useful answers.
“What IBM is showing here is really a very important step towards serious quantum algorithm design,” said Dorit Aharonov, a professor of computer science at the Hebrew University of Jerusalem, who was not involved in the research.
While Google researchers claimed in 2019 that they had achieved “quantum supremacy” — a task that can be performed on a quantum computer much faster than a classical computer — IBM researchers said they had achieved some new Something more useful, despite a more modest name.
“We’re entering what I call the utility phase of quantum computing,” said Jay Gambetta, vice president of Quantum at IBM. “Practical Era.”
The team of IBM scientists working for Dr. Gambetta describe their results in a paper published Wednesday in the journal Nature.
Modern computers are called digital or classical because they process bits of information that are either 1, 0, on, or off. Quantum computers perform calculations on quantum bits, or qubits, that capture more complex information states. Just as physicist Erwin Schrödinger’s thought experiment posited that a cat could be in a quantum state that is both dead and alive, a qubit can be both 1 and 0 at the same time.
This allows quantum computers to perform multiple calculations at once, whereas digital computers must perform each calculation separately. By accelerating computing, quantum computers have the potential to solve large and complex problems in fields such as chemistry and materials science that are out of reach today. Quantum computers may also have a darker side, threatening privacy by breaking the algorithms that protect ciphers and encrypted communications.
When Google researchers declared their supremacy in 2019, they said their quantum computer completed a calculation in 3 minutes and 20 seconds, which would take about 10,000 years on a state-of-the-art conventional supercomputer.
But some other researchers, including those at IBM, have dismissed the claim, calling the problem an artifact. “Google’s experiment is impressive, and it’s impressive, but it’s doing something that isn’t interesting for any application,” said Dr. Aharonov, who is also chief scientific officer at quantum computing company Qedma.
Google’s calculations were also less impressive than they first appeared.A group of Chinese researchers were able to perform Perform the same calculation on a non-quantum supercomputer in a little more than five minutesmuch faster than the 10,000 years estimated by the Google team.
The IBM researchers in the new study performed a different task, one that has intrigued physicists. They used a quantum processor with 127 qubits to simulate the behavior of 127 atomic-scale magnetic bars — small enough to be governed by the weird rules of quantum mechanics — in a magnetic field. This is a simple system known as the Ising model, which is often used to study magnetism.
The question is too complex to compute an exact answer on even the biggest, fastest supercomputers.
On a quantum computer, the calculation can be completed in less than one-thousandth of a second. Every quantum calculation is unreliable — fluctuations in quantum noise will inevitably creep in and introduce errors — but each is fast, so it can be repeated.
In fact, for many calculations, extra noise is intentionally added, making the answers even less reliable. But by varying the amount of noise, the researchers can tease out the specific characteristics of the noise and its impact at each step of the computation.
“We can amplify the noise very precisely, and then we can rerun the same circuit,” said Abhinav Kandala, quantum capabilities and demonstration manager at IBM Quantum and an author on the Nature paper. “Once we have these results with different levels of noise, we can extrapolate the results without the noise.”
Essentially, the researchers were able to subtract the effects of noise from unreliable quantum computing, a process they call error mitigation.
“You have to get around it by inventing really clever ways to mitigate the noise,” Dr. Aharonov said. “That’s what they do.”
The computer performed a total of 600,000 calculations to arrive at the answer for the overall magnetization produced by the 127 bar magnets.
But how good is the answer?
For help, the IBM team turned to physicists at the University of California, Berkeley. Although the Ising model with 127 magnets is too large, with too many possible configurations, to fit into a conventional computer, classical algorithms can produce approximate answers, a technique similar to how JPEG image compression discards less important data to reduce file size size while preserving most of the image details.
Michael Zaleter, a professor of physics at Berkeley and an author on the Nature paper, said that when he started working with IBM, he thought his classical algorithms would do a better job than quantum ones.
“The results were a little bit different than I expected,” Dr. Zaletel said.
Certain configurations of the Ising model can be solved exactly, and both classical and quantum algorithms agree on simpler examples. For more complex but solvable instances, quantum and classical algorithms yield different answers, and the quantum algorithm is correct.
So for other cases where quantum and classical computations diverge and the exact solution is not known, “there is reason to believe that the quantum computation results are more accurate,” says Sajant Anand, a graduate student at Berkeley who has done a lot of work on classical approximations.
It is not yet clear whether quantum computing has an undisputed victory over the classical techniques of the Ising model.
Mr. Anand is currently trying to add an error-mitigating version of the classical algorithm, and it has the potential to match or exceed the performance of quantum computing.
“It’s not obvious that they’ve achieved quantum supremacy here,” Dr Zaletel said.
In the longer term, quantum scientists expect a different approach, error correction, which detects and corrects computational errors, will open the door to speeding up quantum computers for many purposes.
Error correction has been used in traditional computers and data transmission to fix garbled characters. But for quantum computers, error correction may still be years away, requiring better processors to handle more qubits.
IBM scientists believe that error mitigation is a temporary solution that can now be used to solve increasingly complex problems outside the Ising model.
“This is one of the simplest physical science problems in existence,” Dr. Gambetta said. “So it’s a great start. But now the question is, how do you generalize this and move on to more interesting natural science questions?”
These could include clarifying the properties of exotic materials, accelerating drug discovery and simulating fusion reactions.