Is it Turing complete?
Or, to put it another way, it's a physical laboratory set-up that's difficult for a classical computer to simulate but how well can it simulate a classical computer?
Boffins from China say they have managed to detect as many as 76 photons using a quantum computer, a result said to be the second demonstration of "quantum supremacy" or "quantum primacy" – solving a problem that a quantum computer can do far better than a classical computer. Physicists from Shanghai's University of Science …
... the Possibility Exists of Mass Flash Eventual Collateral Damage if Things Go Wrong Both There and Here
Is it Turing complete?
Or, to put it another way, it's a physical laboratory set-up that's difficult for a classical computer to simulate but how well can it simulate a classical computer? ..... Doctor Syntax
You're asking the wrong questions there, DS, and for some others to believe that China is a backward nation and culture because of the state of the roads and sewerage systems in parts thereof, is most telling of the lack of intelligence in those who who express it and offer it as evidence to try prove their invalid point.
If one works with the other premise, at the opposite end of that knowledge spectrum, and one assumes that China/the East is much more advanced than was ever before thought possible, by virtue of information and intelligence which they have nurtured and home grown since forever and/or discovered and uncovered and are further diligently developing, or even been shown by foreign visitors if you wanna believe in those Roswell type incidents/experiments which can provide cover for quantum leaping presentations on Earth and out of this world realisations, the questions morph into .....Is it Turing compete and how well can it stimulate classical computers to compete to simulate the semblance of a comparable advantage before they overheat and collapse and deliver Dependent Operating Systems Meltdowns and Globalised Operating Systems Crashes? ....... hence the Take Care, Beware warning advisory.
That is a fantasy.....
Since it was NOT a unified country but a group of countries that were brought together....
Not to mention that if it is so ahead of its time.. why does it still not have decent roads or a decent sewerage system?
Even the Romans managed that in a much shorter time......
So stop drinking the communist cool aid and start thinking for yourself....
The communist party regime rulers see china as a monoculture, all diversity is to be suppressed. Of the ancient chinese nationalities none are left today, Tibet and several others will join them soon.
That's hardly news to anyone who's not in china, why do I have to explain it to you? Are you in china?
" why does it still not have decent roads"
Almost half of the major roads in the US don't meet the international standard for maximum tolerable roughness of roads.
5.5% of US bridges are classified as being in poor condition. Which isn't many, until you realize that every time you cross your 20th bridge, one of the ones you've crossed has been classified as poor.
Kool-Aid is a proper noun, has a hyphen, and starts with a "k."
So there it is, further evidence that quantum computers really can do certain calculations better than classical computers.
Except it doesn't. It's just more smoke and mirrors, following the recent export ban on Chinese quantum cryptography to make you think they've got something we haven't.
One of things that makes "The Register" such a pleasant read is that the writers knows how to use the power of the understatement. The writer provides facts and context, the reader arrives at their own conclusion. There is too little of that nowadays - most news sites start by instructing the reader what and how to think, and then only sometimes providing context and evidence as an afterthought.
I found that quote to be a stroke of genius journalism - "So there it is ...". Very British.
So we established that quantum systems are hard/expensive to simulate on a computer. But the same is true for almost anything we observe in the world, somewhat depending on the degree of approximation you are willing to accept. I can build a pachinko machine with a board and a few nails and you'll have a hard time to simulate the result of my dropping a ball through it (assuming a proper degree of imprecise and therefore hard-to-model nail positioning on my part). But that doesn't mean my machine is useful for anything but boredom relief for the truly desperately bored.
So what are those "quantum supremacy" experiments good for? And I mean part from getting some scientist a Science paper, which is clearly a good career move.
"you'll have a hard time to simulate the result of my dropping a ball through it (assuming a proper degree of imprecise and therefore hard-to-model nail positioning on my part)"
If we are allowed to measure your board, then it doesn't matter where you hammer nails. The ball is classical. The nails are classical. The board is classical. We can simulate this to very good approximation with the kind of computing power that is being deployed here. TBH, shining an array of lasers pointers would give us a good indication; there's no beam splitters doubling the number of balls nor constructive and destructive interference of these balls - we only have to worry about classical spin, air resistance and frictional losses.
Anyway, Scott's thoughts were this one doesn't have a currently foreseeable use but Google's was done using a practical quantum computer.
And the nails being elastic will vibrate so even is there are no vibrations coming from the environment after the ball hits the first nail the rest of them will be vibrating slightly before it gets to the second. Don't forget the influence of the grain of the wood...
The real world is full of little details that get in the way of predictability.
Proteins fold themselves consistently, they're all built systematically by adding one amino acid to another so it should be possible to calculate the forces that will determine the new configuration from the old as each one is added - but informed guesswork, i.e. AI, is better at working out the configuration from the sequence than predictions based on things we should, in theory, be able to work out.
"We can simulate this [pachinko board] to very good approximation"
And therein lies the rub. How good an approximation is good enough to make meaningful predictions? Classical physics gets really hard really fast except for those few cases where your approximations work well. We don't usually focus on the hard cases because stories on how we cannot model a particularly nasty pachinko board are not really useful.
In the current quantum supremacy stories, it seems like they focus on quantum systems that are easy to implement and hard to model.
Is this a computer at all? It looks like an experiment with light that's being labelled as a quantum computer. The description makes it sound like it's not even configurable to other problems.
Presumably Xi is spending big money on his "china quantum" push, and they're competing for budget, just like the EU did the same a decade ago and Chinese researchers are labelling any laser experiment "quantum computer research" to get some funding, just as happened in the EU.
"quantum supremacy" is presumably aimed at nationalistic politicians, not familiar with what a quantum computer actually is.
Fools and their money....
for which I will probably be shot down in flames.
But rather than run a test then spend half a billion dollars trying to verify it. Why not run run a complex but not too time consuming task on a supercomputer. Then, since the quantum computer is 10^14 times quicker, run the test 10^14 times and see if it gets the same answer as the supercomputer 99.999999% of the time. Unless they have some way of discriminating a "correct" answer from the almost infinite answers produced by the quantum computer, they don't have an announcement to make.
That makes sense. Decoherence is the elephant in the room in QC. How does the decoherence affect "random distribution" experiment result, compared to performing Shors algorithm to crack RSA? Determining if a set of samples from the target distribution is truly random is a "soft" question. Finding the primes p and q used for an RSA key is not a soft question - it has sharp corners.
Start small, just a few bits, but answer a "hard" question. Then extend the number of bits, and experimentally determine how much harder the problem becomes, due to the decoherence, as the number of bits increases.
QC computing power increases exponentially as O(2**n) with number of bits n. Does the cost(*) of decoherence have a less than exponential order? (*cost as in to get sigma confidence in the answer, how many times must the calc be run, or how many many error correcting gates are required).. So far decoherence cost doesn't seem to be a part of the theory, other than "theoretically it is solvable", and is instead labelled as a engineering issue that can be overcome eventually with elbow grease and perseverance. Well, OK - but in that case start solving simple problems of increasing complexity and get an empirical measurement of how that decoherence cost increases with n. That engineering problem will only start improving when there are standard benchmarks to beat.
So [The Chinese team developed an apparatus, dubbed Jiuzhang, that consists of a laser, mirrors, prisms, and photon detectors. The group reported that they achieved a sampling rate that's "~1014 faster than using the state-of-the-art simulation strategy and supercomputers."]
They built an experiment and were able to take measurments 10^14 time faster than simulating the physics in a computer. It is unclear what is new about this state of affairs, this is why most groups use experiments for this type of reaserch, rather than simulations.
"They built an experiment and were able to take measurments 10^14 time faster than simulating the physics in a computer."
I'm glad someone else brought this up, I was starting to think I was missing something. Even reading the paper didn't clear things up (although given it has a typo within the first three words I wasn't expecting anything particuarly high quality). This has absolutely nothing to do with quantum computing. It's an actual physical measurement using a pretty standard photon counting system, which they then claim is faster than doing a simulation of said system. Well yes, no shit.
Biting the hand that feeds IT © 1998–2021