Quantum ain't just for leaping

February 9, 2007 1:01 AM

So.. who's ready for Quantum Computing?

British Colombia-based D-Wave says they've got one and they're going to demo that sucker in Mountain View, CA on Feb 13th and then at the Telus World of Science in Vancouver, Canada on February 15th.

Quoting from TechWorld :

"Multiple quantum states exist at the same time, so every quantum bit or "qubit" in such a machine is simultaneously 0 and 1. D-Wave's prototype has only 16 qubits, but systems with hundreds of qubits would be able to process more inputs than there are atoms in the universe."

Naturally, the tech-savvy blogosphere is skeptical. But what do you think? (previously, previously)
posted by revmitcz (54 comments total)

Full disclosure - I don't understand thing #1 about quantum mechanics, quantum theory, quantum cryptography or quantum computing. I posted this mostly to watch MeFi talk about quantum computing in some fashion that I could understand and then explain why this probably won't happen - or at least not this soon.

That having been said, this "qubit" stuff reminds me of Q-bert, so I'm kinda looking forward to dropping that term in casual geek conversation sometime soon.
posted by revmitcz at 1:07 AM on February 9, 2007

I know enough to know that I don't know jack about this.
posted by Heywood Mogroot at 1:31 AM on February 9, 2007

I can't get my head around this concept. It's too impossible. Why can scientists come up with this, but not, say, a viable alternative to oil or a miracle crop that ends world hunger?

Still, i await our new Q-puters. i bet they'll be able to run Photoshop CS2 with no lag -- while downloading the Chemical Brothers' entire catalog!

Swank.
posted by ELF Radio at 1:34 AM on February 9, 2007

Somebody Should Do Something About All the Problems
posted by Human Flesh at 1:39 AM on February 9, 2007

Why can scientists come up with this, but not, say, a viable alternative to oil or a miracle crop that ends world hunger?

We Can Put A Man On The Moon, But We Can't Make Killer Robot Police?
posted by Human Flesh at 1:47 AM on February 9, 2007

So.. who's ready for Quantum Computing?

I'm undecided.
posted by srboisvert at 1:55 AM on February 9, 2007 [4 favorites]

Ah, Edith Heare...... the prophet of a new and better world.
posted by ELF Radio at 2:00 AM on February 9, 2007

I'm undecided.

I'm simultaneously for and against it. Sort of.
posted by joe lisboa at 2:22 AM on February 9, 2007 [1 favorite]

There exists in that box, simultaneously, a computer and no computer at all.

We won't know until they open the box.
posted by loquacious at 2:26 AM on February 9, 2007 [1 favorite]

I suspect Dwave is deluded if not an outright hoax. It bears some of the marks of a free-energy scam; instead is the first 'free computation' scam.
Their device is not actually a 'quantum computer' in the sense you have read about. The fact that they have purposely muddied the waters here is just one of the many problems I have with their claims. The fact that they are demoing the device to laymen before its been shown to work to experts in the field is a big red flag.

On the other hand, if Dwave is legit and able to pull of their ultimate claim of being able to solve np-complete problems, their device is substantially more powerful than the 'quantum computers' talked about today; it truly would be solving a big problem which would go a long way towards helping scientists solve the types of problems in ELFs post.
posted by Osmanthus at 2:30 AM on February 9, 2007

The thing that's always struck me about QC is that very very small proofs of concept can often be made to work, but scaling the thing up to achieve anything that's actually hard for a conventional computer to do always seems to run into decoherence problems.

I have a sneaking suspicion that information vs. reliability of information will turn out to be an uncertainty relationship, much like time vs. energy or position vs. momentum, and that this will stop QC being as fabulous as its boosters would have us believe.

But I'd be happy to be proved wrong.
posted by flabdablet at 3:16 AM on February 9, 2007

Jokes for nerds!
posted by oxford blue at 4:03 AM on February 9, 2007

Cat's dead, cat's alive, whatever: who's going to clean the poop out of the box?
posted by stavrosthewonderchicken at 4:28 AM on February 9, 2007

stavroscluck screwed my joke.
posted by toma at 4:42 AM on February 9, 2007

I've never understood this at all well (I don't think very many people do), but from what I do get, the point of a quantum computer is not to 'compute' in a traditional sense.

Right now, computers are deterministic devices; they are always in a known state, and transition in known ways to other states (as long as they are working correctly, anyway :) ). Each step is sequential. Your computer can 'multitask', but it's not really running multiple programs at once... it's just so damn fast that it can jump from program to program to program faster than you can see. It feels like they are all running at the same time, even though they're actually being executed in turn. (note: multi-CPU systems actually do run more than one program at a time, but fundamentally they're doing the same task-switch thing... there's just two or more CPUs doing it instead of one.)

Some things are very hard to do with this kind of computing. At the moment, for some kinds of problems, the only way we know how to solve them is to try different combinations of possibilities until we get to one that works. Factoring prime numbers is one of these problems, and it's the foundation of modern encryption. Encryption keys are pairs of prime numbers that mathematically interact with each other somehow. (sorry I'm not clear on the details). Given the public key and a ciphertext message, it requires an unbelievable amount of time to find the other prime number and decode the message. It boils down to repeatedly trying number after number..."does this decrypt it? does this? how about this?". There are a lot of potential numbers to check... a decently strong encryption system would require the lifetime of the Universe to crack this way.

This is one of the things that quantum computers are believed to be able to solve. They are made of 'qubits'; each qubit is a single atom that has been linked with other qubits around it. By carefully arranging inputs in some fashion (and this gets very very very hairy), scientists are able to describe the parameters of the problem... and the qubits will automatically fall into a correct answer for the problem. In theory, by feeding the public key and the ciphertext to a large enough quantum computer, it would instantly determine the correct decryption key, forced into that state by the rules of quantum interactions.

This would mean, of course, that all public key encryption methods would be rendered obsolete as soon as a quantum computer was built with enough qubits to crack that particular encryption.

This should also be applicable to other things as well, problems that are presently very hard to solve... if we can posit a question clearly to a quantum computer, the Universe itself will, in essence, answer. Some branches of thought look at reality as a gigantic computation system... a quantum computer would be hijacking some of that computing power for our own use.

That said, we haven't gotten this working yet, and it doesn't look likely anytime soon. As flabdablet is saying, perhaps it can't work... I've had a similar hunch to the one he describes, but he put it into words better than I ever have.

I've probably butchered this explanation.. don't take anything I'm saying as gospel, just a very rough first approximation. Hopefully, if I've really messed it up, someone more knowledgeable will be irritated enough to correct me. :)
posted by Malor at 4:55 AM on February 9, 2007 [2 favorites]

My tiny comment slits wavicled your jocularity beam and hived off a coupla new universes. Cluck you, buddy: first observed, best dressed!
posted by stavrosthewonderchicken at 4:56 AM on February 9, 2007

but will it tell me where the final Wonka's Golden Ticket is located?
posted by medium format at 4:58 AM on February 9, 2007

It seems the Adams' quotient shall soon be quoted.

Yes. I can feel it.
posted by oxford blue at 5:17 AM on February 9, 2007

First off, I agree with Osmanthus. I'm ridiculously highly suspicious of technology claims made to the mass public.

For anyone who's interested, here are the US patents that I could dig up from D-Wave. I probably won't read these because they look a bit tricky. It might be important to note that the other company taking out a bunch of patents in this field is IBM.
posted by muddgirl at 5:25 AM on February 9, 2007

This is kewl, but I wish we could get fusion and desktop nanotech first....
posted by pax digita at 5:25 AM on February 9, 2007

All I know is that there's a good chance this thing will rip a whole right through the space-time continuum...

...and with the oncoming energy crunch, nobody's going to be able to afford a computer that sucks up 1.21 gigawatts of power.
posted by Muddler at 5:28 AM on February 9, 2007

I heard that was the beauty of the quantum computer.
You don't even need to run a program to get the correct answer.
I'm not even sure if you need to write the complete software for said un-run program.
You probably don't even need the q-puter.

Just thinking about the problem should solve itself quantumly.
posted by Balisong at 5:37 AM on February 9, 2007

This will only be true and actual when Apple does it.

But for an explanation with drama, watch that episode where Worf is splintered into endless quantum universes that he travels about in and people die and he gets it on and stuff and we also discover that he didn't win the bat'leth tournament but that's only because the other guy cheated but of course in other splinters he did win and also didn't even go. At least it ended with champagne.
posted by juiceCake at 6:01 AM on February 9, 2007

I've covered quantum computing for a number of years -- in fact, my latest book is on information theory and discusses quantum information and quantum computation.

I'm quite skeptical.

Some context: the biggest operating quantum computer that I know of has fewer than 10 quantum bits; these use NMR to store quantum bits and execute algorithms by manipulating nuclear spins. This method cis not scaleable -- it won't get much beyond where it is today.

There's a number of other methods out there that might be scaleable -- and one that holds a lot of promise is by storing qubits and executing algorithms by manipulating magnetic fluxes or electric currents in solid-state circuitry. The problem is, though, that unlike what happens in NMR quantum computing, it's hard to couple quantum bits. Without coupling, there's no executing a quantum algorithm.

So these guys are claiming that they've coupled 16 quantum bits to one another in solid-state circuitry. It would be a big leap forward if true. I doubt it is.

My reasons:
1) This would a major advance -- if true, it would be easily publishable in Nature or PRL. Much of what led up to it would also be similarly publishable. Yet, despite the standard PRL/arXiv format, I don't see evidence that they have submitted their work for publication.

2) There are certain quantum algorithms which are standard: the Deutsch-Josza algorithm, the Grover algorithm, Shor factoring, etc. These are the canonical test problems for quantum computers. If I wanted to prove to the world that I had a quantum computer, the first thing I'd do is execute one of these to prove that I've got what I'm claiming. Simulating an Ising spin glass is a very weird choice.

3) A public demonstration is bizarre, especially since the public has no means of determining what's doing the computing and whether, in fact, the computation is faster than classical. It smacks of fund-raising -- and shady fund-raising at that. The money would come in if they built their reputation on some peer-reviewed publications.

4) They're touting practical applications. There's no way a 16-bit quantum computer can do anything useful; regular computers are so much bigger and more efficient that it'll take dozens more bits before quantum computers can offer even special-purpose processors. Solving tiny NP-complete problems that can be exhausted by a conventional computer isn't a realistic application, IMHO.

5) I've never heard of these guys before. Obviously, people can come out of nowhere (and my knowledge of the quantum computation community isn't perfect) but the quantum computation community is pretty small, and you get to know the names of the big players reasonably quickly.

So -- while I'm not saying it's impossible, I wouldn't let these people within 500 yeards of my wallet.
posted by cgs06 at 6:01 AM on February 9, 2007 [13 favorites]

Well, you just got another book sale.
posted by geoff. at 6:09 AM on February 9, 2007

"but will it tell me where the final Wonka's Golden Ticket is located?"

It could, but that would be cheating.

Anyway, haven't we already had a quantum computer? As I recall, early Pentiums said 3 was simultaneously 3 and not 3.
posted by mr_crash_davis at 6:33 AM on February 9, 2007

Nobody saw him, the quantum man,
But still he lay moaning:
I was much further out then you thought
And not a wave but a particle.
Poor chap, he always loved larking
And now he’s indeterminate
It must have been too cold for him his heart gave way,
They said.
Oh, no no no, he was a wave always
(Still the indeterminate one lay moaning)
I was much too far out all my life
And not a wave but a particle.
posted by Falconetti at 6:39 AM on February 9, 2007 [2 favorites]

This will only be true and actual when Apple does it.

Actually, Apple already unveiled it at the last Macworld expo -- it sparked a near riot in the conference hall. You just happen to be in the reality where it didn't happen, and they talked about some stupid phone instead.
posted by pardonyou? at 6:44 AM on February 9, 2007

The third link in the FPP has an interesting thread at the bottom which says there is a "roadmap" to have 1,000 qbits by 2008. It seems like it is tailored for a highly specific application. I am cautiously hopeful.

I have added Charles Seife's book to my wishlist, and also recommend Quantum Enigma.
posted by stbalbach at 6:53 AM on February 9, 2007

Will it be able to clean the track marks out of underwear?
posted by Captaintripps at 7:04 AM on February 9, 2007

I'm highly skeptical as well.

Look at their optical picture of the qubit processor. It has the 16 qubits on a circuit board connected by wiring. In a real quantum computer, the qubits must be maintained in a state of entanglement during processing exchanging no information with the outside world. I'd expect a quantum computer to consist of molecules in close proximity at near absolute zero temperatures, not little processors connected by wires.
posted by justkevin at 7:15 AM on February 9, 2007

We won't know until they open the box.

You mean until we make them open the box.
posted by Horace Rumpole at 7:35 AM on February 9, 2007

cgs06: It's comments like yours that make MetaFilter occasionally so damn awesome. Thanks for the insight.
posted by TBoneMcCool at 8:14 AM on February 9, 2007

Since I don't understand any of this, I'll just point out this awesome bit from the Q*Bert link:

Dedicated upright cabinets for Q*bert contain a solenoid that creates a knocking sound inside the cabinet whenever a character falls off the pyramid, simulating the sound a character might make if it actually fell to the bottom of the cabinet. In some units, this sound is created by a bean bag inside the case rigged to fall.


Now quantum mechanics I can wrap my brain around.
posted by Terminal Verbosity at 8:21 AM on February 9, 2007

The beauty of a quantum computer is that while you don't necessarily get the right answer, in some of the universes that get forked off there is a version of you that does.
posted by George_Spiggott at 8:23 AM on February 9, 2007 [1 favorite]

Can I play Diablo on it?
posted by Mister_A at 9:08 AM on February 9, 2007

Is anybody else feeling all hot and bothered with all this talk of processing inputs? I need to take a cold shower...
posted by dobie at 9:11 AM on February 9, 2007

Leonid Levin explains here why quantum computing cannot work as claimed, even in principle:

The major problem is the requirement that basic quantum equations hold to multi-hundredth if not millionth decimal positions where the significant digits of the relevant quantum amplitudes reside. We have never seen a physical law valid to over a dozen decimals. Typically, every few new decimal places require major rethinking of most basic concepts. Are quantum amplitudes still complex numbers to such accuracies or do they become quaternions, colored graphs, or sick-humored gremlins? I suspect physicists would doubt even the laws of arithmetic pushed that far. In fact, we know that the most basic laws cannot all be correct to hundreds of decimals: this is where they stop being consistent with each other!
posted by metaplectic at 9:13 AM on February 9, 2007

...and with the oncoming energy crunch, nobody's going to be able to afford a computer that sucks up 1.21 gigawatts of power.

1.21 Gigawatts!?!?!!!
posted by BoatMeme at 9:20 AM on February 9, 2007

cgs06 -The investing/venture capital market in Canada is tiny and extremely risk averse. Despite any nationalist rhetoric you might hear, there isn't really the climate of fostering innovation that exists in most of the developed world. This might actually just be their cute and unsophisticated way of trying to get a bit of funding.
posted by dobie at 9:22 AM on February 9, 2007

cgs06 - I bought your book for my girlfriend to explain some of the complex ideas that I'm just poor at communicating. Picked it up last week and found myself sucked in. Nice work!
posted by BoatMeme at 9:26 AM on February 9, 2007

The smartest guy I know in the field is Dave Bacon. Granted, he is also the only guy I know. I took his Quantum Computing class at the UW. Here is what he has to say on the topic.

In short: He's skeptical.
posted by jeffamaphone at 10:41 AM on February 9, 2007

Boatmeme, TBone -- thanks much.

Dobie -- that's a shame, because the scientific community in Canada has been extremely savvy about where they spend their limited resources. For example, the Sudbury Neutrino Observatory was a way to make a world-class contribution to physics. And I think the Perimeter Institute (which is privately funded) is a real good idea, especially in conjunction with U. Waterloo. It would be a shame if the only way to get capital is to PR everything to the hilt.

Geoff and St.Balbach -- I hope you enjoy!
posted by cgs06 at 2:00 PM on February 9, 2007

This will only be true and actual when Apple does it.

Product names, anyone?

My bet's on iQube.
posted by weston at 2:35 PM on February 9, 2007

Would they publish if it meant they lost their head start of unique intellectual property? Publishing really matters to academics, and I understand it can be important for credibility, but equally I could see they might not want to let the, uh, cat out of the box.
posted by Rumple at 2:37 PM on February 9, 2007

Rumple writes "Would they publish if it meant they lost their head start of unique intellectual property? Publishing really matters to academics, and I understand it can be important for credibility, but equally I could see they might not want to let the, uh, cat out of the box."

Well, as muddgirl pointed out, they've already done a lot of patenting. Once they hold a patent on a technology, there's no reason not to publish on that technology; in fact, there's every incentive to publish. And it would be foolish for them to publicly demonstrate something they haven't patented yet, since that could be construed as a public disclosure of the invention, which would make it impossible to get an international patent.

Plus, I'm not sure the public is in a very good position to understand this demonstration; this is pretty esoteric stuff, and a very specialized field for even professional computer scientist and physicists. Publication seems much more appropriate.
posted by mr_roboto at 4:36 PM on February 9, 2007

Which is not to say that I think this is phony; I'm in absolutely no position to judge that. It just all very strange.
posted by mr_roboto at 4:38 PM on February 9, 2007

thanks mr_roboto. My understanding though is that a patent can be pretty vague: "a novel way to wire the doohickies together while maintaining suspension in substances rich in silcon and myrrh". Whereas publication will require a more convincing exposition. I appreciate they could licence the idea, but, maybe their principle would lead a competitor to a better idea, and leave them with nothing.

But -- well, I know less about patents than about qubits. I just think that flying low can be understandable when developing something sufficiently novel -- it'd be good to know how many ways there might be to skin that cat in box before issuing knives to the competition.
posted by Rumple at 5:30 PM on February 9, 2007

"This will only be true and actual when Apple does it.

Product names, anyone?

My bet's on iQube."

iDunno.
posted by mr_crash_davis at 6:06 PM on February 9, 2007

5) I've never heard of these guys before.
they have been toiling away in vancouver for years. I attended a presentation by Geordie Rose about the technology back in 2002 when he presented to NewMIC. some impressive names from the local venture capital community in their management team
posted by seawallrunner at 6:24 PM on February 9, 2007

They're demoing it to a credulous public and performing the calculations remotely. "No really, there is a quantum computer at the other end of this cable, we promise!".

I don't believe it for a second.
posted by polyglot at 11:45 PM on February 9, 2007

I went to school (at UC-Berkeley) with one of the board members, Gene Dantsker, who got his degree under John Clarke, an expert on SQUIDs (superconducting quantum interference devices), which form the basis of this quantum computer.

I also worked in Hans Mooij's group in Delft, which was one of the first to demonstrate the flux qubit technology that appears to underlie this purported computer.

Based on these two pieces of information, plus the fact that a "public demonstration" is being used to demonstrate the technology, I can confidently say that this is crap of a particularly fragrant variety; it smells like money being burned.
posted by starkeffect at 2:41 AM on February 11, 2007

Back in 1994 or so, a friend of mine was trying to tell me that the next big thing was trinary memory...
posted by bugmuncher at 5:08 AM on February 11, 2007

Has anyone actually seen the machine at the other end of the remote interface? Heh..
posted by oberleit at 9:04 PM on February 17, 2007