QUANTUM COMPUTER PERFORMS FIRST SUCCESSFUL FACTORING

[ciphergoth]

QUANTUM COMPUTER PERFORMS FIRST SUCCESSFUL FACTORING

IBM SCIENTISTS BUILD MACHINE THAT SOLVES MATHEMATICAL PROBLEMS WITH QUANTUM MECHANICS

NUMBER 15 FACTORED

FACTORS ARE 3 AND 5

http://www.research.ibm.com/resources/news/20011219_quantum.shtml

Comments

Gulp.

[babysimon]

Wow! Cool. I think.

But they appear to need a specific design of molecule to get 7 qubits - so this isn't going to scale. I like the quote "The first quantum computing applications would likely to be co-processors" -- as if people were thinking the market was about to be flooded with 1Gqb SIMMs.

But if they do manage to build a 16kqb machine (or however large a box one would need to brute-force RSA), this would lead to some serious asymmetry. We keep hearing about quantum crypto, but AFAIK that requires a special link between computers, and can only be done at the transport layer. Hardly something you can run over IP. Am I right? If so, I would like to wish everyone a happy goldfish bowl, like at the end of the Asimov story...

Re: Gulp.

[bootpunk.livejournal.com]

As per above, mibby I should leave a response to the expert in our midst, but afaik, it's the transmission of single photons twixt points - so an opto-connection would suffice. I'm sure ciphergoth will correct me if I'm wrong.

Re: Gulp.

[babysimon]

"an opto-connection would suffice"]

Last mile?

Re: Gulp.

[bootpunk.livejournal.com]

Again, IANAC(ryptofreak), but I seem to recall even our own BT saying that they could see thousand metre connections shortly/now, and much further soonish. The connections aren't the major issue, iirc. But I'm out of me depth here, and I can't remember the details clearly, so I'll STFU.

Re: Gulp.

[ciphergoth.livejournal.com]

For the record, I think quantum cryptography is silly. For provable security, you have to exchange secret keys in advance to get the authentication working, and if you're going to do that then I don't think the massive inconvenience buys you much over perfectly ordinary symmetric techniques.

Re: Gulp.

[ciphergoth.livejournal.com]

Symmetric crypto keeps working. Grover's algorithm means we have to double our key lengths, but that's no hardship, and really something we should be doing anyway; 256-bit keys are the Right Thing for a variety of reasons. So no need to use that impractical quantum crypto stuff.

But I think Shor has shown QC algorithms for pretty much all the important problems that lie at the heart of public-key stuff. We've got quite used to having public key crypto, it would be sort of weird to lose it again.

Despite this news, I would still be surprised if anyone ever gets, say, a 50-qubit machine going. The equivalent of Moore's Law for quantum computing at the moment seems to be that the machines grow by 1 qubit a year...

Re: Gulp.

[bootpunk.livejournal.com]

> Despite this news, I would still be surprised if anyone ever> gets, say, a 50-qubit machine going.
[my emphasis]

Don't you mean "in the near future"? Or do you think that quantum computing will prove to be too expensive or surpassed by a different technology before it ever becomes scaled sufficeintly?

Re: Gulp.

[ciphergoth.livejournal.com]

It's already surpassed by conventional techniques, and if both conventional and quantum techniques stick to their current rates of improvement then it always will be.

Also, it may turn out to be fundamentally impossible to run a 50-qubit machine without decoherence killing it...