Quantum computing techniques are prone to errors that make classical computers sweat

Despite steady improvements in quantum computers, they are still noisy and error-prone, leading to problematic or incorrect answers. Scientists predict that they won’t truly surpass today’s “classical” supercomputers for at least another five or 10 years, until researchers can adequately fix the bugs that cause them. entanglement for quantum bits or qubits.

But a new study shows that, even without effective error correction, there are ways to minimize errors that could make quantum computers useful today.

Researchers at IBM Quantum in New York and their collaborators at the University of California, Berkeley and Lawrence Berkeley National Laboratory reported today (June 14) in the journal Nature that they compared a 127-qubit quantum computer with a state-of-the-art computer. supercomputers are state of the art and, for at least one type of computation, have surpassed supercomputers.

This calculation was not chosen because it is difficult for classical computers, but because it is similar to calculations physicists perform all the time, the researchers say. Importantly, the calculation can be made increasingly complex to test whether today’s noisy, error-prone quantum computers can produce accurate results for some of the most common types of calculations. decide or not.

The fact that quantum computers produce verifiable correct solutions as calculations become more complex, while supercomputer algorithms produce incorrect answers, gives hope that quantum computing algorithms Elements with the ability to minimize errors, rather than making fixing them more difficult, can solve more advanced problems. physics problems, such as understanding the quantum properties of superconductors and new electronic materials.

“We are entering a regime where quantum computers can do things that current algorithms on classical computers cannot,” said Sajant Anand, a UC Berkeley graduate student and study co-author. can do”.

Sarah Sheldon, senior director of Quantum Theory and Capabilities at IBM Quantum, added: “We can start to think of quantum computers as a tool to study problems that are otherwise unexplored. We cannot study it.”

Conversely, quantum computers defeating classical computers could also spark new ideas to improve quantum algorithms currently used on classical computers, according to co-author Michael Zaletel, deputy UC Berkeley physics professor and holder of the Thomas and Alison Schneider Chair in Physics.

“Going into this, I’m pretty sure the classical method will perform better than the quantum one,” he said. “So I have mixed feelings that IBM’s version of noiseless extrapolation works better than the classical approach. But thinking about how quantum systems work can really help us figure out how ancients work. dictionary is suitable to approach the problem. While quantum computers have done that.” what the standard classical algorithm cannot do, we think it is the inspiration to make the classical algorithm better so that classical computers work as well as quantum computers in the future.”

Noise booster to block noise

One key to IBM’s quantum computer’s seeming advantage is quantum error mitigation, a new technique for dealing with noise that comes with quantum computing. Paradoxically, the IBM researchers controlled the noise in their quantum circuit to get an even noisier, less accurate answer and then extrapolated back to estimate the answer the computer would receive if there were no noise. This depends on understanding how noise affects the quantum circuit and predicting how much it will affect the output.

The noise problem is caused by IBM qubits, which are sensitive superconducting circuits that represent zeros and ones of binary calculus. When qubits are entangled together for computation, unavoidable annoyances, such as heat and vibration, can alter that entanglement, causing errors. The greater the entanglement, the worse the impact of noise.

Additionally, calculations operating on one set of qubits can introduce random errors in other qubits that have not been processed. Additional calculations then compound these errors. Scientists hope to be able to use more qubits to track such errors so that corrections can be made – so-called fault-tolerant error correction. But achieving scalable fault tolerance is a major technical challenge, and whether it will work in practice with ever greater numbers of qubits, Zaletel said.

Instead, IBM engineers came up with an error-reduction strategy they called zero-noise extrapolation (ZNE), which uses probabilistic methods to increase noise on a quantum device significantly. control. Based on a recommendation from a former intern, IBM researchers approached Anand, postdoctoral researcher Yantao Wu, and Zaletel to ask for their help in evaluating the accuracy of the results obtained using the Use this error mitigation strategy. Zaletel develops supercomputer algorithms to solve difficult calculations involving quantum systems, such as electronic interactions in new materials. These algorithms use tensor lattice simulations, which can be directly applied to simulate interacting qubits in quantum computers.

Over a period of several weeks, Youngseok Kim and Andrew Eddins at IBM Quantum performed increasingly complex quantum calculations on the advanced IBM Quantum Eagle processor, then Anand attempted the same calculations using modern classical methods on the Cori supercomputer and the Lawrencium Cluster at Berkeley Lab and the Anvil supercomputer at Purdue University. When Quantum Eagle launched in 2021, it had the highest number of high-quality qubits of any quantum computer, seemingly exceeding the simulation capabilities of classical computers.

In fact, accurately simulating all 127 entangled qubits on a classical computer would require an enormous amount of memory. The quantum state would need to be expressed as a power of 2 with 127 distinct numbers. It’s a 1 followed by 38 zeros; Ordinary computers can store about 100 ratios, 27 orders of magnitude too small. To simplify the problem, Anand, Wu and Zaletel used approximation techniques that allowed them to solve the problem on a classical computer in a reasonable amount of time and at a reasonable cost. These methods are a bit like compressing jpeg images, in that they discard less important information and retain only what is needed to reach the correct answer within the limits of available memory.

Anand confirmed the accuracy of the quantum computer’s results for less complex calculations, but as the depth of the calculations increased, the quantum computer’s results differed from those of the ancient computers. dictionary. For certain specific parameters, Anand can simplify the problem and calculate exact solutions to verify quantum calculations against classical computer calculations. At the largest depth considered, there is no exact solution, but quantum and classical results disagree.

The researchers warn that, although they cannot prove that quantum computers’ final answers to the most difficult calculations are correct, Eagle’s successes in previous runs have helped them. I am confident that I am right.

“The success of quantum computers is not like a fine-tuned accident,” Zaletel said. It really works for the circuit it is being applied to.”

Friendly competition

Although Zaletel is cautious in predicting whether this error reduction technique will work with more qubits or higher-depth calculations, the results are still encouraging, he said.

“It kind of fosters a sense of friendly competition,” he said. “I have a feeling that we can simulate on a classical computer what they are doing. But we need to think about it in a smarter and better way – quantum devices are in a regime that suggests we need a different approach.” approach.”

One approach is to simulate the ZNE technique developed by IBM.

“Now, we are asking if we can take the same error minimization concept and apply it to classical tensor network simulations to see if we can get classical results,” says Anand. better or not”. “This work gives us the possibility of being able to use quantum computers as a verification tool for classical computers, flipping the script on what is typically done.”

Anand and Zaletel’s work was supported by the U.S. Department of Energy under an Early Career Award (DE-SC0022716). Wu’s work is supported by a RIKEN iTHEMS Fellowship. Cori is part of the National Energy Research Scientific Computing Center (NERSC), the primary scientific computing facility of the U.S. Department of Energy’s Office of Science.