D-Wave Makes Waves
Image: King et al, arXiv:2403.00910
Buckle up for some serious quantum stuff! The company D-wave, which is pursuing a unique approach to quantum computing, has reported a breakthrough. They say they’ve managed a computation that would have taken a conventional computer millions of years which, if true, is the most impressive quantum computation we’ve seen so far. Let’s have a look.
To understand what D-wave has done, I first need to tell you how D-wave’s approach differs from other quantum computing approaches. D-wave doesn’t attempt to execute logical operations on single qubits. They instead use what’s called “quantum annealing”. Sounds like a hippie retreat, but means you set up a large collection of qubits in a suitable configuration, then let those qubits relax into an energetically ideal state -- which they do on their own -- and that ideal state gives you the answer you want to know.
D-wave has been working with thousands of qubits for years, whereas IBM only just reached 1000 late last year. However, because the approaches are so different, it makes no sense to just compare the numbers of qubits.
The D-Wave team used two of their devices to simulate these phase transitions of the coupled magnet system for a larger collection of magnets than anyone has been able to calculate before. The tricky bit is verifying that it’s right because you know, no one’s been able to do it before.
So what they do is that they compare the result of their calculation, that’s the blue squares in the right figure with the cases of small numbers of magnets where it can be calculated on a conventional computer, that’s the line. As you can see, they agree very well.
And for a large number of magnets, there are some mathematical approximations from which we know that some quantities of the magnets’ system obey general scaling behaviours. Again their simulation fits remarkably well to the expectation. You can see this in this figure on the right, where the data points are from their simulation and the rectangles are the theoretical predictions.
They then estimate how long it would take to do this calculation on the best current supercomputers. As an example they use the Frontier supercomputer located at Oak Ridge National Laboratory in Tennessee. It’s currently one of the most powerful computers in the world and performs more than a billion billion operations per second.
The D-wave group writes that even for a number of magnets smaller than what they did their calculation for, the “hypothetical runtime on the Frontier supercomputer would surpass millions of years.”
It’s actually really impressive. Honestly, I’m much more impressed by this than by the IBM result from last year. Though I want to point out that this is a pre-print and has not yet been peer reviewed.
Keep reading with a 7-day free trial
Subscribe to Science without the gobbledygook to keep reading this post and get 7 days of free access to the full post archives.