Discover the hidden "noise" that can kill qubits

Researchers at MIT and Dartmouth College have developed, for the first time, a tool to detect new features of environmental "noise" that can destroy the fragile quantum state of qubits, fundamental components of quantum computers. This breakthrough could lead to a better understanding of microscopic noise mechanisms and new methods of qubit protection.

Qubits can represent the two states corresponding to the classical binary bits, 0 or 1. But they can also maintain a "quantum superposition" of the two states simultaneously, allowing quantum computers to solve complex problems virtually impossible for conventional computers.

However, the quantum "coherence" of a qubit, that is, its ability to maintain the overlay state, can collapse under the effect of noise from the environment around the qubit. Noise can come from the control electronics, heat or impurities in the qubit material itself, and can also result in serious miscalculations that can be difficult to correct.

Researchers have developed statistical models to estimate the impact of unwanted noise sources around qubits to create new ways to protect them and better understand the noise mechanisms themselves. But these tools generally capture a simplistic "Gaussian noise", essentially the collection of random disturbances from a large number of sources. In short, it is like a white noise from the murmur of a large crowd, where there is no specific disruptive pattern that stands out, so that the qubit is not affected in any way by any particular source. In this type of model, the noise probability distribution would form a standard symmetrical bell curve, regardless of the statistical significance of the individual contributors.

In an article published today in the journal Nature Communications, the researchers describe a new tool that, for the first time, measures the "non-Gaussian noise" affecting a qubit. This noise has distinctive patterns that usually come from some particularly loud noise sources.

Researchers devised techniques to separate this noise from Gaussian background noise and then used signal processing techniques to reconstruct very detailed information about these noise signals. These reconstructions can help researchers build more realistic noise models, which could lead to more robust methods of protecting qubits from specific types of noise. Researchers now need such tools: Qubits are made with fewer defects, which could increase the presence of non-Gaussian noise.

"It's like being in a crowded room. If everyone is talking with the same volume, there is a lot of background noise, but I can still maintain my own conversation. However, if some people speak particularly loudly, I can not help keeping them from their conversation. It can be very distracting, "says William Oliver, an associate professor of electrical engineering and computer science, professor of physics practice at the Lincoln Laboratory Fellow at MIT, and associate director of the Electronics Research Laboratory (ELT). "For qubits with many flaws, the noise will decode, but we usually know how to handle this type of aggregate, usually Gaussian noise. However, as qubits improve and there are fewer defects, individuals begin to stand out and the noise may no longer be simply Gaussian in nature. We can also find ways to handle this, but we must first know the specific type of non-Gaussian noise and its statistics. "

"It is rare for theoretical physicists to be able to design an idea and to find an experimental platform and experimental colleagues willing to invest in it," says Lorenza Viola, co-author, professor of physics at Dartmouth.. "It was great to achieve such an important result with the MIT team."

The following authors join Oliver and Viola: first author Youngkyu Sung, Fei Yan, Jack Y. Qiu, Uwe von Lüpke, Terry P. Orlando and Simon Gustavsson, all of RLE; David K. Kim and Jonilyn L. Yoder of the Lincoln Laboratory; and Felix Beaudoin and Leigh M. Norris of Dartmouth.

Pulse filters

For their work, researchers have taken advantage of the fact that superconducting qubits are good sensors for detecting their own noise. More precisely, they use a qubit "flux", which consists of a superconducting loop capable of detecting a particular type of disturbing noise, called magnetic flux, coming from its environment.

In the experiments, they induced a non-Gaussian "phase shift" noise by injecting an artificial flow noise that disturbs the qubit and causes it to lose coherence, which is then used as a measurement tool. "Usually, we want to avoid decoherence, but in this case, qubit decoherence tells us something about the noise in its environment," Oliver says.

Specifically, they fired 110 "pi pulses" – which are used to invert qubit states – in specific sequences over tens of microseconds. Each pulse sequence effectively created a narrow frequency "filter" that masked much of the noise, except in a particular frequency band. By measuring the response of a qubit sensor to bandpass filtered noise, they extracted the noise power in this frequency band.

By changing the pulse sequences, they could move the filters up and down to sample the noise at different frequencies. In doing so, they notably observed how non-Gaussian noise made the qubit distinctly unchecked, providing a large spectrum of non-Gaussian noise.

Deletion and correction of error

The main innovation behind this work is to carefully design the pulses to act as filters to extract the properties of the "bispectrum", a two-dimensional representation providing information on the distinctive temporal correlations of non-Gaussian noise.

Essentially, by reconstructing the bispectrum, they might find the properties of non-Gaussian noise signals that affect qubit in time – properties that do not exist in Gaussian noise signals. The general idea is that, for Gaussian noise, there will only be a correlation between two instants in time, called "second order temporal correlation". But, for a non-Gaussian noise, the properties at a given moment will directly correlate with the properties of several future points. These "higher order" correlations are the mark of non-Gaussian noise. In this work, the authors were able to extract noise with correlations between three moments.

This information can help programmers validate and personalize error suppression and error correction codes for qubits, which helps correct noise-induced errors and ensure accurate calculation.

Such protocols use noise model information to make implementations more efficient for practical quantum computers. But as the details of the noise are not yet well understood, the current error correction codes are designed with this standard bell curve. Using the researchers' tool, programmers can either evaluate the effective operation of their code in realistic scenarios, or begin to focus on non-Gaussian noise.

Oliver maintains the same analogy with a crowded room: "If you know there is only one strong person in the room, you will design a code that effectively stifles that person rather than trying to handle all the scenarios possible. "