Slashdot Mirror
Scientists Break Quantum Entanglement Record At 18 Qubits (zmescience.com)
hackingbear writes: Researchers at the Chinese University of Science and Technology have demonstrated stable quantum entanglement with 18 qubits, surpassing the previous world record of 10, also held by the same team. This represents a step toward realizing large-scale quantum computing, according to a recent study published in the journal Physical Review Letters. Physicist Pan Jianwei and his colleagues achieved the new record by simultaneously exploiting three different degrees of freedom-paths, polarization and orbital angular momentum of six photons, the fundamental particle of light. The outcome combination resulted in a stable 18-qubit state. Full control over the number of entangled particles determines the fundamental ability for quantum information processing, according to the study. There are early-stage quantum computers out there that argue more qubits -- such as IBM's 50-qubit machine and Google's 72-qubit Bristlecone, but in those cases, the individual quantum states of the qubits aren't (fully) controllable. "The team's next step will be to realize a 50-qubit entanglement and manipulation," according to Wang Xilin, a member of the team. The same research team also held the world record on quantum communication distance as well as operating the world's first quantum communication satellite.
I think you have no idea what QC actually is.
QC is limited only by the size of a stable system that you can build. But the stable systems you can build give you any/all/every answers IMMEDIATELY. That's unprecedented in standard computing. The bigger the system you can stabilise, the bigger the questions you can answer immediately (i.e. factorise this 2048-bit number), and you can answer ALL such questions in that same timeframe (so once you can break one 2048-bit key instantly, you can break them all instantly).
That's getting some SERIOUS funding from military sources, not to mention all the other questions that it can answer. Investors won't go away until the horse has been proven dead and flogged for decades, but we still just keep advancing (sure, it's not super fast, but when computing was brand new, one transistor was the size of an apple and very limited... 40 years later, we had GHz microprocessors on the head of a pin).
However, classical processors have hit size limits (too large and the signals have to be asynchronous to propagate at the speed of light and stay consistent, but we haven't even got those kinds of chips yet), speed limits (same problem with speed-of-light), thermal limits, etc. and haven't significantly advanced in years. There's a reason that every desktop is still only "2Ghz" or so (maybe 3/4 in short bursts but not sustained performance), and the fastest mainframes in the world are still in that order of magnitude. I had a 2GHz machine in 2000-something. 18 years hasn't made that much of a difference to classical processor speeds in consumer kit at room-temperature! Sure, we have multiple cores, but that's just making the processing harder, and the processor hotter, and we KNOW that now everything you want to do scales in parallel.
Classical will never make QC irrelevant, it's an entirely different kettle of fish. If anything QC will render traditional computing entirely insecure overnight. We are already having to make QC-safe algorithms now, we're so worried about what might happen if someone builds a decent-size one.
The first nation to perfect QC is going to be a world leader... it'll be bigger than the space race and the nuclear race combined. Not just from a military point of view, but also from the sheer number of problems that your physicists and mathematicians and engineers can just say "Solve this, I've reworded it in QC language" (which we're already preparing and were before physical QC machines even existed!), and they literally get the answer as soon as they press Go. You will literally advance science overnight just by having a single QC machine large enough (enough qubits) to run a decent-size QC algorithm on... it will answer questions that we currently can't answer with a billion years of traditional computation, overnight. 1000 qubits is all you need to change the world, most likely. Given that the record is beat every year, that's not far off.
It means the writer believes that people who understand the phrase "simultaneous exploit of three different degrees of freedom-paths, polarization and orbital angular momentum" might not know what photons are.
No sig today...
The bigger the system you can stabilise, the bigger the questions you can answer immediately (i.e. factorise this 2048-bit number), and you can answer ALL such questions in that same timeframe (so once you can break one 2048-bit key instantly, you can break them all instantly).
That is simply not true.
http://www.quantumforquants.org/quantum-computing/limits-of-quantum-computing/
Probably. With the progress they are making, they will most assuredly not deliver in the next 50 years and without any fundamental breakthrough (not on the horizon and cannot be planned or forced) it may take 1000 years or longer for this to become useful at all. At the moment, they seem to be able to add about 1 Qbit/year for actual computations. And the impression that this may scale sub-linear is not off the table at all.
Time for the hype to die down, there is nothing useful this technology can do.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
QC is limited only by the size of a stable system that you can build. But the stable systems you can build give you any/all/every answers IMMEDIATELY.
Looks like you are the clueless one here. No, it does not give you answers immediately at all. You still have to feed in the data and do computation steps and you have to do this slowly and carefully to avoid decoherence. And if it decoheres, you have to do everything again from scratch. And due to noise, you either have to add a lot of error-correcting steps or run it for a lot of times. There is nothing "immediate" here.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
Actually quite a few more. Even to break an ECC modulus, they need about 300 more. With the scaling of around 1Qbit/year since 2001 (when they factored 15 on a 7 Qbit machine), my guess would be that it will take a few centuries to get there.
Silicon scaled exponentially almost from the beginning and continued to do so for a long time. That is what makes it powerful today. QCs have never scaled better than linearly and are still at a ridiculously useless size as a consequence, after about 30 years of applied research. They may also well scale sub-linearly. The whole thing would have been dropped as a dead-end a while ago, except that many humans run after every hype that tickles their fantasy.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
The answer is immediate. O(1). Not O(n) or worse as in any classical system.
I'm not sure how you could have come to believe this ... but it's incorrect. Quantum computers can use algorithms that scale better than classical ones - say, scaling as O(n^2) rather than O(e^n) - but they don't generate an answer *immediately*.
For example, Shor's algorithm for factorisation runs in O((logN)^2 * loglogN * logloglogN) time, while the classical general field number sieve which does the same thing runs in (roughly) O(e^(1.9 * (logN)^0.33 * (loglogN)^0.66) time. That's a massive improvement - going from subexponential to polynomial time - but it's still not instant.
I think it's almost always a mistake to assume linear progress on R&D -- a single breakthrough can drastically alter time-lines.