Skip to content

IBM is strengthening the entire quantum computing stack

    By making small adjustments to the frequency at which the qubits operate, it is possible to prevent these problems. This can be done when the Heron chip is calibrated before being opened for general use.

    In addition, the company has performed a rewrite of the software that controls the system during operations. “After learning from the community and seeing how we could run bigger circuits, [we were able to] It's almost better to define what it should be and rewrite the whole stack on that,” Gambetta said. The result is a dramatic acceleration. “Something that now took 122 hours has been reduced to a few hours,” he told Ars.

    Because people are paying for time on this hardware, that's good for customers now. However, it can also pay off in the longer term, because some errors can occur randomly, so less time spent on a calculation can lead to fewer errors.

    Deeper calculations

    Despite all these improvements, errors are still likely during important calculations. As it continues to work on developing error-corrected qubits, IBM is focusing on what it calls error mitigation, which it first described last year. As we described it at the time:

    “The researchers chose a method where they deliberately amplified the noise of the processor at different levels and then measured them. These measurements are used to estimate a function that produces a similar output to the actual measurements. That function can be then set noise to zero to produce an estimate of what the processor would do without any noise.”

    The problem here is that using the function is computationally difficult, and the difficulty increases as the number of qubits increases. So while it is still easier to perform error mitigation calculations than to simulate the behavior of the quantum computer on the same hardware, there is still the risk of it becoming computationally intractable. But IBM has also taken the time to optimize that. “They have algorithmic improvements and a method that uses tensor methods 1731566617 uses the GPU,” Gambetta told Ars. “So I think it's a combination of both.”