IBM Raises the Bar with a 50-Qubit Quantum Computer (technologyreview.com) 69
IBM said on Friday it has created a prototype 50 qubit quantum computer as it further increases the pressure on Google in the battle to commercialize quantum computing technology. The company is also making a 20-qubit system available through its cloud computing platform, it said. From a report: The announcement does not mean quantum computing is ready for common use. The system IBM has developed is still extremely finicky and challenging to use, as are those being built by others. In both the 50- and the 20-qubit systems, the quantum state is preserved for 90 microseconds -- a record for the industry, but still an extremely short period of time. Nonetheless, 50 qubits is a significant landmark in progress toward practical quantum computers. Other systems built so far have had limited capabilities and could perform only calculations that could also be done on a conventional supercomputer. A 50-qubit machine can do things that are extremely difficult to simulate without quantum technology. Whereas normal computers store information as either a 1 or a 0, quantum computers exploit two phenomena -- entanglement and superposition -- to process information differently.
Imagine a Beowulf cluster of these? (Score:5, Funny)
But can it run Linux?
Re:Imagine a Beowulf cluster of these? (Score:4, Funny)
Yes.
And No.
At the same time.
Re: (Score:2)
Re: (Score:2)
encryption (Score:4, Interesting)
One of the reasons the three letter agencies like to store even encrypted communication is that quantum computers will allow breaking encrypted data in ways that classical computers can't do in any practical sense. An example is Shor's Algorithm for factoring numbers, which runs efficiently in a practical amount of time on a quantum computer and could be used to break public key crypto. If they have saved the current encrypted text they can later break that when quantum computing hits.
Quantum computing is not quite there yet but it is coming up the well.
Re: (Score:2)
I wouldn't be surprised to see a move to lattice based algorithms or crypto that is resistant to quantum factoring in the next few years, once there is a significant key factored. Or, perhaps when a key handshake is done, part of it is keeping a shared secret for a later time, so if the public/private part of the encryption is broken, the shared secret, even though not as secure, would still protect the data.
Re:encryption (Score:5, Interesting)
The practical defenses against the hypothesized quantum cryptopocalypse are:
Grover Issues:
A) Double the key size for symmetric algorithms, MACs
B) Double your hash sizes (you can finesse in which situations, but for practical purposes just double them all)
Shor Issues:
C) Use Hash based signatures for certificates.
D) Replace RSA, DH and ECDH with something else. Lattice crypto is a contender. Some with claim NTRU is fine, but it's not practical.
You shouldn't have been using DSA in the first place. So that's moot.
The dilemma is that the fix for asymmetric key crypto is not clear. Various lattice proposals have come along and been broken. RWLE is a PITA to implement (although that might be getting better soon with some stuff I've seen) and generally we don't know what it's going to be.
On the positive side, it's all BS. They will not build a quantum computer capable of breaking RSA any time soon. TFS makes is sound they they got from 2 bits to 50 bits and so 256 bits are only a short way off. This is grossly misrepresenting the situation. You can make some fragile qbits cohere but you can't do iterative logic on it., You can make a reliable, error corrected qbit, but you can't make reliable error corrected qubits into a memory on which you can perform the quantum logic needed to implement Shor's algorithm. These are the barriers to cross and as far as I can tell, they have remained unsolved for many years. Upping the number of non-ecc qbits doesn't move us towards breaking public key crypto.
I may or may not be proven wrong, but we will have the symmetric upgrades deployed in most new silicon pretty soon and the conference circuit will remain well attended while the lattice crypto work continues. So there will be lots more travel to nice places.
Re: (Score:2)
Citations needed, I think.
Re:encryption (Score:4, Informative)
Really? Isn't this all textbook stuff, except maybe for my DSA snark? Well DSA is very fragile, so I'll keep on snarking.
Here's a common one: M.H. Devoret and R.J.Schoelkopf , Science, Vol 339, 2013.
This has a diagram with a little green arrow from the 3rd stage to the 4th of the 7 stages of development. Saying we're at stage 3 and getting from the 3rd to the 4th stage is the current problem. That was 2013. We're still waiting.We haven't got to stage 4 (logical memory with longer liftime than physical qubits) from stage 3 (QND measurements for error correction and control). Stage 4 through 7 are entirely unsolved.
For all the supposed major advances in quantum computers, by the metric that matters, we haven't moved in 5 years.
Re: (Score:2)
You say "They will not build a quantum computer capable of breaking RSA any time soon." and you're right, for some values of "soon". Unfortunately, the time scale that matters is "how long will it take for us to agree on a replacement for DH, develop it, test it, and have it flush out of all use"? How long has it taken for IPv6 to take over? This is probably worse...
Re: (Score:2)
In the hardware I'm responsible for the answer is by the end of the decade.
I agree that browser security and PKI are basket cases. They will do as they have always done and will make changes when it's too late.
Re: (Score:2)
Quantum is no silver bullet. Basically only public key cryptography is in a need of overhaul, as algorithms we currently use, chosen for nice short keys, are vulnerable. Elsewhere, it's not an issue: for example it's proven that a quantum algorithm can break a hash of no more than twice the length than an equivalent non-quantum computer. Yes, double the hash length is an exponential speed-up, but the only effect is hashes being slightly more cumbersome to read for a human.
Post Quantum Computing Algorithms (Score:2)
Awfully large. (Score:5, Funny)
Re: (Score:2)
The summary is wrong. These must be cubic cubits, as measuring a computer in a single dimension makes no sense at all.
Re: (Score:2)
I think it all comes down to what you base your measurement system from.
Most of the world's measurements are centered around "the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium 133 atom." All of the other basic units require that duration as part of their derivation.
America has realized that it's a poor basis, as the duration can vary significantly depending on one's relative velocity vs a frame of refere
Coherence (Score:3)
I get the feeling that we're going to find out that maintaining coherence requires energy that's exponential in the number of qubits, which would making quantum computing mostly useless.
Our universe has always tended to stop those who try to break the rules; try making a perpetual motion machine, for example.
Re:Coherence (Score:5, Informative)
Re: (Score:2)
I was expecting some crackpot video, but no, it was a real academic with real work.
Nice talk.
Re: (Score:2)
This video is not exactly crystal-clear...
Re: (Score:2)
This video is not exactly crystal-clear...
Noise is correlated between qubits, so it adds up rather than canceling out.
Re: (Score:2)
Perpetual motion is easy. Extracting energy from said machine is impossible.
A machine which appears to be perpetual motion and provides extractable energy is also easy. It fails as soon as you leave whatever environmental conditions you're exploiting.
Re: (Score:2)
I think that the question is : is that quantum computer able to factor integers into prime numbers faster than a classical computer using the same amount of power.
If that's the case, even if it proves too impractical to break cryptography right now, it should at least prove that there is something to be gained from quantum computing.
Re: (Score:2)
Re: (Score:2)
"try making a perpetual motion machine"
Not all that difficult. we're powering ours with human stupidity which is infinite, filtered thru a mesh of hashed bitcoin which are well known to be imaginary. The math -- which involves dividing stupidity by cellphone user intelligence (zero) shows that perpetual motion is not only possible, but inevitable. We'll be taking our product to market just as soon as we handle a couple of engineering glitches.
Getting closer to testing quantum supremacy (Score:5, Insightful)
We're getting closer and closer to testing quantum supremacy- the hypothesis that quantum computers can practically solve problems that classical computers cannot do https://en.wikipedia.org/wiki/Quantum_supremacy [wikipedia.org]. Note that this is a practical statement; anything a quantum computer can do, a classical computer can do, but with potentially exponential slowdown. This follows from the fact that BQP https://en.wikipedia.org/wiki/BQP [wikipedia.org] the set of problems that a quantum computer can do in polynomial time is within is contained in PSPACE https://en.wikipedia.org/wiki/PSPACE [wikipedia.org] the set of things that a classical computer can do with polynomial space (since polynomial space calculations live in EXPTIME, the set of things requiring exponential time, the result follows).
It is very likely that before we see genuinely useful quantum computing (e.g. for factoring large numbers or simulating complicated chemical systems) we'll have an answer to the quantum supremacy question. I suspect that it is more likely that we'll have an answer in terms of boson sampling before we have an answer involving a universal quantum computer.
Essentially, boson sampling works by just looking at the distribution of bosons (well for convenience, photons) as they go through very simple optical objects. Boson sampling has two major advantages: first, we know it is actually *hard* in a technical sense for a classical computer to do unless some conjectures that pretty close to everyone believes are false. In particular, Scott Aaronson and Alex Arkipov proved that if a classical computer can do boson sampling efficiently then the polynomial hierarchy will collapse https://www.scottaaronson.com/papers/optics.pdf [scottaaronson.com]. For those who aren't theoretical compsci people, the polynomial hierarchy not collapsing is a statement which is only marginally stronger than P!=NP and is very widely believed. This is in contrast for example with factoring large numbers where if it turned out that classical computers could efficiently factor the only major conjecture that would turn out to be false would just be the difficulty of factoring itself. Second, boson sampling is much easier in many respects than what IBM is trying to do which requires much fancier systems, supercooled qubits, careful protection from stray particles, careful preservation of entanglement and all sorts of other stuff. Still, what they are doing is important and very necessary if we're going to actually have practical quantum computers ever.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Notes (Score:2)
is it truly 50qb? (Score:1)
D-Wave 1000+ Qubit (Score:1)
Wait, isn't D-Wave already providing a 1000+ Qubit computer? What's the difference?
Re: (Score:2)
Re: (Score:2)
There's no small amount of controversy as to whether D-Wave is even "quantum". It's definitely not general purpose.
D-Wave's current offering is "15x" faster than a single-core silicon microprocessor -- and the tasks it's useful for are embarrassingly parallel. Modern laptops are starting to be offered with 18 or more cores - meaning that even laptop CPU can outperform D-Wave's "1000+ Qubits"
Scientific publications have, by and large, found that a traditional multicore silicon chip can easily outperform what
D-Wave (Score:2)
Canadian-owned and operated D-Wave [dwavesys.com] computer has way more that 50 Q-bits with a 1000Q model available and a 2K in the works.
Re: (Score:3)
D-Wave makes quantum annealing processors - and is only useful for a sliver of useful computing (adiabatic quantum computing).
There's no small amount of controversy [wikipedia.org] as to whether D-Wave systems are truly quantum machines. A number of groups found "no quantum speedup" and have shown better performance using traditional silicon, and studies [sciencemag.org] have been published [arxiv.org] to that effect.
Having worked in supercomputing for a decade, I've looked hard are D-Wave's "quantum" computing, and give it slightly more credibility t
Re: (Score:2)
D-Wave makes quantum annealing processors - and is only useful for a sliver of useful computing (adiabatic quantum computing).
Whew, that's lucky! Quantum computing wouldn't get very far at all without obese diabetic programmers participating.
Re: (Score:2)
Canadian-owned and operated D-Wave computer has way more that 50 Q-bits with a 1000Q model available and a 2K in the works.
D-wave works on a completely different design. Their systems can not manipulate individual qbits, but instead have all their qbits in a big pool functioning together such that they can only manipulate the entire grouping.
Instead of reading out individual qbits, they read the energy level of the entire pool of qbits summed together.
This makes it easier to actually setup all of those qbits in the first place, but they are limited to solving "lowest energy state" problems.
IBM and Google are using designs that
So What? (Score:2)
I don't know how many times I've read the now rote description of quantum computing in some sciencey magazine or blog:
"Our regular computers have bits that can be only 1 and 0. A quantum computer has bits that use Superposition, the bits can be both 1 and 0 at the same time"
Every time I read it, I ask my self, so what? So a bit can be both 1 and 0 at the same time. That didn't explain anything at all.
Re: (Score:2)
Every time I read it, I ask my self, so what? So a bit can be both 1 and 0 at the same time. That didn't explain anything at all.
Right, they forget to mention the hoped-for consequence, which only becomes apparent when you consider a system containing more than one qubit at once.
i.e.
1 qubit = 2 simultaneous states (== 2x potential speedup vs classicalp)
2 qubits = 4 simultaneous states (== 4x potential speedup vs classical)
3 qubits = 8 simultaneous states (== 8x potential speedup vs classical)
[...]
64 qubits = 18446744073709551616 simultaneous states (== 18446744073709551616x potential speedup vs classical)
It's the old rice-on-the-ches [singularitysymposium.com]
Re: So What? (Score:1)
Yes everyone always explains that.
But they never explain how that is in anyway useful.
Hope do you give an input (like a cipher text) and have it spit out an encryption key. Yes 128qbits could take the state of any 128bit encryption key, but how do you collapse it to the correct key?
Re: (Score:2)
It's very weird. Much like everything in the quantum world. Imagine you had 2 classical bits. They could be in the configuration of :
1,1
1,0
0,1
0,0
but only one of those 4 states. With qubits, they are all of those at the same time. As such, where we would just read the bits in a classical computer, referencing the bits is not enough, we have to provide a coefficient as well, to tell which quantum state we are checking. That means essentially we can derive 4 bits (2 coefficients + 2 bits) worth of inform
Re: (Score:2)
There is a library you can download for free that will simulate a number of qubits. But yes, you can use Not and And operations on them. In fact, that is the whole challenge of building an algorithm that takes advantage of them. Because once you apply those rules, and measure the outcome, you collapse the quantum state and they all become a single value, not a superposition. So, think of it this way, it's not that it exists in all states, it is that you don't have to calculate it in all states, you mere
Re: (Score:2)
Unfortunately, only Cubots can understand the theory, and they can also not understand it at the same time.
Re: (Score:2)
That didn't explain anything at all.
Well. It does and it doesn't.
That's not very long (Score:2)
Riiiight. (Score:3)
What's a qubit?