3 ms·
Ordinary computers can beat Google’s quantum computer after all
- smckk 4y agoWhy are there so many quantum computing deniers though? From the speed, to the solving of intractable problems, to even the building of powerful q-computers, there seems to be a lot of push-back on all milestones and achievements for this new tech.
- pyinstallwoes 4y agoNew tech that is available for purchase in a physical computing package? Or more like new tech that is only available like playing a MMORPG?
- cwillu 4y agoThere's a lot of hype and misinformation. For one, there haven't been any quantum computers built that could be remotely considered “powerful”. The theoretical basis is solid, and there have been plenty of proofs of concept, but nothing has yet been built which solves actual real-world problems any better than any analog computer would (let alone better than the digital computers we're having this conversation over). The combination of the massive asymptotic-scaling potential, the subtleties of just what that scaling could and could not accomplish, and the engineering difficulties in actually implementing that scaling, make a wonderful soil for the weeds of know-it-all skeptics and know-nothing enthusiasts to grow and choke out the slower growing… ‹gives up on the terrible metaphor› :p
- StrictDabbler 4y agoOne problem is that D-Wave spent the last ten years pushing their quantum annealers as "quantum computers". Everybody got very excited about having functional quantum computers but the D-Wave system can't do any of the stuff we really want or fear from quantum tech. No factoring of large numbers or simulating atomic processes. So there's a lingering air of vaporware over the whole scene. This new google computer can (mostly) produce a very specific type of result using entanglement. It's very cool but it's basic research on how to hook up gates and verify they're hooked up. It can't be used to solve any problem except "what should the output of these random quantum gates be". We're waiting for the IAS but we're still pre-Eniac. Probably about 1925, seven years after the flip-flop was invented.
- dekhn 4y agoI predicted this would happen back when Google announced the original results. It's pretty obvious- there's a long history of people speeding up classical algorithms and the QC people keep misrepresenting their achieved speedups.
- cwillu 4y ago“Still, the Google demonstration was not just hype, researchers say. Sycamore required far fewer operations and less power than a supercomputer, Zhang notes. And if Sycamore had slightly higher fidelity, he says, his team’s simulation couldn’t have kept up. As Hangleiter puts it, “The Google experiment did what it was meant to do, start this race.””
- cwillu 4y agoAlso (from the preprint): “At the same time, our experiments also reflect that Google’s hardware has several advantages over our algo- rithm. The most significant one is that Google’s hardware is much faster in sampling the quantum circuits with sufficient depth, while our algorithm has exponential complexity, hence is not scalable to both depth and qubit number”
- dekhn 4y agoThe scaling part is obviously the relevant part of QC, and when somebody can come up with a QC that solves a Real Problem that couldn't be solved by a crack team of supercomputer developers, we'll all be very happy and then the community can say they've achieved "supremacy".
- dekhn 4y agoIf you want to convince people that QC works, don't say you solved something a supercomputer takes 10K years to solve. The Google experiment didn't start any races- it was a flag planted in the ground, but the ground wasn't stable. Nobody is racing to improve this algorithm except people who want to show their classical computer chops, because the algorithm doesn't do anything useful.
- rrss 4y agoRelated March 2021 preprint: https://arxiv.org/abs/2103.03074 https://arxiv.org/abs/2103.03074 Jan 2022 paper: https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.128.030501 https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.12...
- timcavel 4y ago