Combatting crosstalk in quantum computers

Saturday, October 23, 2021

In an advance towards better control over large quantum computers, researchers have demonstrated a new tool to compensate for crosstalk in superconducting circuits.

“Ultimately, if we want to be able to build very large quantum computers, we’re going to need to keep errors to manageable levels,” said IQC and University of Waterloo Department of Physics and Astronomy faculty member Adrian Lupascu. “What we did in this paper was to tackle one point where errors creep in, which has to do with control signals.”

A superconducting chip can have many qubits—the quantum version of your classical computer bit—and scientists need to be able to send an electrical or magnetic signal to each one to control them individually.

Independent control is not so easy to maintain, however. With so many control lines so close together on the same chip, parts of signals get ‘borrowed’ by qubits they weren’t meant for, and the result is a noisy mess. This pollution is called crosstalk.

The ideal approach to dealing with crosstalk is to design your device so it doesn’t happen in the first place. But this easier said than done, and not always possible depending on the type of device you’re using.

The alternative approach is to compensate for the crosstalk. If qubit one is getting too much signal because it is being affected by the signal to qubit two, you can alter the level of signal going to each qubit to compensate. But there’s a problem: to compensate, you need to have a very accurate understanding of the crosstalk that’s happening. Large quantum systems are too complicated to simulate, which is part of the reason why they have the potential to be so powerful. But it also means it is difficult to model all their dynamics.

The researchers found another way. They created a fully automated tool that takes advantage of a known fundamental property of superconducting circuits to start with a rough estimate of crosstalk and iterates repeatedly until it has a very accurate characterization.

Circuit diagram

Coupling from the electric control signal (blue) to superconducting circuits (purple and orange). The solid black arrow indicates coupling to the intended loop and the dashed arrow indicates crosstalk to unintended loops. The teal arrow indicates interaction between circuit elements, which hampers crosstalk calibration measurement.

Importantly, the team demonstrated the tool on a device that is one of the largest quantum computers currently in operation, in terms of number of independent control signals, showing promising applications in the ever-growing systems of the future. And because it takes advantage of a fundamental property of superconducting systems, it is device-independent within that set of systems, meaning the specific design of a device is not a barrier to using this tool.

The team also developed a method to quantify the error in their tool’s procedure, letting them know how close they are to determining the actual crosstalk. So far, the results are promising; the error rates are well within the normal range for potentially achieving scalable quantum information processing.

The breakthrough was the result of a program in quantum annealing involving an international collaboration of more than ten partners across academia, government and industry, including MIT Lincoln Laboratory in Massachusetts, where much of the experimental work took place.

“I would say it is the most sophisticated data analysis tool that I have been involved with: being able to process so much data in a way that will be reliable without any human intervention, essentially—that is something that was very interesting to see,” said Lupascu.

First author Xi Dai, a PhD student at IQC and the Department of Physics and Astronomy, played a crucial role in making the breakthrough happen.

“Many implementations of quantum computing rely on electric control to manipulate qubits, and crosstalk is a major challenge to scaling up quantum computers,” said Dai. “We proposed an innovative crosstalk calibration procedure that is automated and implemented on superconducting devices with up to 27 control loops, which are among the largest in the community.”

The researchers’ work is another piece in the puzzle of a large, powerful quantum computer that do things totally impossible for a classical computer.

“If you want to build a meaningful quantum computer, you have to put many qubits together and maintain low error rates,” said Lupascu. “This will likely require a combination of fundamentally new science and major technical developments.”

The ultimate answer may be unclear, but compensating for crosstalk is a small but meaningful step towards a solution.

Calibration of flux crosstalk in large-scale flux-tunable superconducting quantum circuits was published in PRX Quantum on October 20, 2021.

  1. 2021 (37)
    1. November (2)
    2. October (6)
    3. September (3)
    4. August (2)
    5. July (3)
    6. June (3)
    7. May (5)
    8. April (3)
    9. March (2)
    10. February (5)
    11. January (3)
  2. 2020 (48)
    1. December (4)
    2. November (4)
    3. October (8)
    4. September (2)
    5. August (4)
    6. July (2)
    7. June (8)
    8. May (4)
    9. April (6)
    10. March (2)
    11. February (2)
    12. January (2)
  3. 2019 (61)
    1. December (6)
    2. November (2)
    3. October (8)
    4. September (4)
    5. August (2)
    6. July (6)
    7. June (6)
    8. May (4)
    9. April (4)
    10. March (11)
    11. February (6)
    12. January (2)
  4. 2018 (28)
  5. 2017 (34)
  6. 2016 (44)
  7. 2015 (38)
  8. 2014 (31)
  9. 2013 (27)
  10. 2012 (33)
  11. 2011 (31)
  12. 2010 (52)
  13. 2009 (29)
  14. 2008 (31)
  15. 2007 (27)
  16. 2006 (15)
  17. 2005 (4)