Back in 2019, Google proudly announced They have achieved what quantum computing researchers have sought for years: evidence that esoteric technology can outperform conventional methods. But this evidence of “quantum supremacy” is being challenged by researchers who claim they got ahead of Google on a relatively ordinary supercomputer.
To be clear, no one is saying that Google lied or misinterpreted its work – the painstaking and groundbreaking research that led to its declaration of quantum supremacy in 2019 is still very important. But if this new paper is correct, the classic versus quantum computing competition is still anyone’s game.
Could you Read the full story of how Google took you from theory to reality in the original article, but here’s the very short version. Quantum computers like Sycamore are no better than classical computers at anything yet, with the possible exception of one task: simulating a quantum computer.
It sounds like a conditional process, but the point of quantum supremacy is to demonstrate the applicability of this method by finding one very specific and exotic task that it can do better than even the fastest supercomputer. Because this puts a quantum foot in the door to expand this library of tasks. Maybe eventually all tasks will be faster in quantity, but for the purposes of Google in 2019, there was only one, and they showed the how and why in great detail.
Now, a team at the Chinese Academy of Sciences led by Pan Zhang has published a paper describing a new technique for simulating a quantum computer (specifically, specific noise patterns it places) that appears to take a fraction of the estimated time of a classical computation to do so in 2019.
I am neither an expert in quantum computing nor a professor of statistical physics, I can only give a general idea of the technique Zhang et al. used. They posed the problem as a large 3D network of tensors, with 53 qubits in sycamore represented by a network of nodes, extruded 20 times to represent the 20 cycles that the sycamore gates went through in the simulation. The mathematical relationships between these tensors (each with its own set of correlated vectors) were calculated using a set of 512 GPUs.
In Google’s original research paper, it was estimated that it would take about 10,000 years to run this scale of simulation on the most powerful supercomputer available at the time (the summit at Oak Ridge National Laboratory) – although this is evident, this was their estimate of 54 qubits. 25 cycles 53 qubits 20 works considerably less complex but will take a few years at your discretion.
Zhang’s group claims to have done it in 15 hours. And if they had access to a proper supercomputer like Summit, it could be done in a few seconds — faster than Sycamore. Their paper will be published in Physical Review Letters. You can read it here (PDF).
These results have not been fully examined and replicated by those familiar with such things, but there is no reason to believe that they are some kind of error or hoax. Even Google has admitted that the wand may be passed back and forth several times before the dominance is firmly entrenched, since it is extremely difficult to build and program quantum computers while traditional hardware and software are constantly being improved. (Others in the quantum world were skeptical of their claims at first, but some are direct competitors.)
Google provided the following comment to acknowledge the progress here:
We said in our 2019 paper that classical algorithms would improve (in fact, Google invented the method used here to simulate random circuits in 2017, and fidelity trading methods for computational costs in 2018 and 2019) — but the key point is that quantum technology is improving exponentially faster. So we don’t think this classic approach can keep up with quantum circuits in 2022 and beyond, despite significant improvements in the past few years.
As University of Maryland quantum scientist Dominik Hangleiter science saidThis isn’t a black eye for Google or a knockout of yours in general by any means: “The Google Experiment has done what it was supposed to do, start this race.”
Google may respond with new claims of its own – they haven’t been consistent either. But the fact that it is even competitive is good news for everyone involved; This is an exciting area of computing and work like Google and Zhang continues to raise the bar for everyone.