Slashdot Mirror
A Working Quantum Computer in 3 Years?
prostoalex writes "Vancouver, BC-based D-Wave Systems got $17.5 mln from Draper Fisher Jurvetson to work on a preliminary version of a quantum computer, Technology Review reports. Delivery date? Within three years: 'It won't be a fully functional quantum computer of the sort long envisioned; but D-Wave is on track to produce a special-purpose, "noisy" piece of quantum hardware that could solve many of the physical-simulation problems that stump today's computers, says David Meyer, a mathematician working on quantum algorithms at the University of California, San Diego.'"
Yeah, but will it play Duke Nukem Forever??
The truth shall always be free: Boris Floricic is Tron.
All your possible answers are belong to us!
The whole mania behind this technology is that somehow we will be able to pull correct data out of thin air using the magical properties of quantum units. Somehow eigenvalues will just instantaneously pop into existence by the careful selection of input parameters.
Too bad that's not how it works. These computers will still have to process data the same as any other processor and all the threat behind magically decoding 128-bit encryption is pure fluff. We are talking about another way of computing, for sure, but it is just another step in the evolution of computing systems rather than a brand new magic bullet for encryption maniacs.
It is also unclear why people want to build a "quantum computer" when it seems that simply putting it on a peripheral board and using it as a separate calculation machine seems to be a much more straightforward application of the device than trying to cram a whole computer with these chips.
The 2006, 2007, 2008 Vaporware Award goes to D-Wave Systems.
Wow, a Quantum Computer that only exist in a "Powerpoint Universe ©".
Grundgesetz * 23. Mai 1949 - 30. November 2007 - http://www.vorratsdatenspeicherung.de/
... but it's not a proper quantum computer. It's based on tunneling, not entanglement. The latter is what everyone understands by the term 'quantum computer'. Their computer just requires knowledge of quantum theory to build it. Well, so do conventional computers...
GHz has no meaning with Quantum computers. Sorry. Visualizing QC in terms on the Pentium in your computer is invalid.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
People have been building quantum computers for years now. The biggest ones these days (around 14-qubits) are NMR quantum computers, although that technique appears to have scalability issues.
Seems to me that this is only news since they plan on selling quantum-CPU time.
There are two kinds of "quantum computers":
;-) which basically makes use of quantum effects to implement smaller/faster/better transistors. that's all what this one boils down to: make better transistors and build the very same computers we made so far (of course, while trying to improve things like speed, energy usage, size, costs...)
;-)
;-) they're probably going to be available as extension cards for "classical" computers (similar to of 3D accelerator cards today...)
the one is a dererministic computing device (call it "pentium" or similar...
the other is a whole new kind of devices. these are devices where bits of information are not represented by small elecrtronic components meaning either '0' or '1', but by quantum mechanical systems (say: atoms, molecules or even photons) that are both '0' and '1' at the same time (each of them with a certain probability).
the very moment you try to find out in what state a given quantum bit (say: qubit) is, it "decides" whether it wants to be '0' or '1'. but until then, it is _both_ (it's not like it's either one or the other, but you just don't know... it's really _both_ of 0 and 1 at the same time!)
so the big advantage of the latter is that instead of, for example, multiplying two numbers, then multiplying other tho numbere, than others and so on, you can really multiply _all_ numbers with _all_ numbers in a single computation step (ok, that's a very simplified description, but that basically is it).
thus, it reduces the computation time for certain numbers (like cracking RSA-based encription keys) from "exponential" to "constant", or to say it in numbers: from "1000 times the age of the universe" to "5 seconds"
but all this only with a given probability -- a quantum computer is not a deterministic device, so don't imagine firing up mozilla on your brand new QC
This is somewhat offtopic, but I ran across it a few months ago and it's really interesting. QCL allows you to write and run quantum algorithms. Runs on Linux and OS X with some tweaking.
The documentation that comes with it is really interesting, and gives some good insights into how quantum computing works and how to write programs for a quantum computer.
Famous Last Words: "hmm...wikipedia says it's edible"
A little searching on arXiv.org brought up:
Quantum Algorithm for SAT Problem and Quantum Mutual Entropy
So at least the first half of that title relates to your question.
Bit of a problem that one. As soon as you know the speed of your quantum computer you're unable to find it...
One of these days I'm moving to Theory - everything works there
GHz has no meaning with Quantum computers. Sorry.
Clock speeds still do mean something in quantum computers. Arguably they're even more important than in classical computers, since in quantum computers you need to get operations done at least 10^4 times faster than the system's decoherence time for quantum error correction to be robust. Decoherence times can be as short as microseconds, meaning that multiGHz operations could be important. Of course, if you're building a quantum computer, you want to work with a system with as long a decoherence time as possible....
It will, however, be ideally suited to solving problems like the infamous traveling-salesman problem . . . D-Wave's chip performs exactly this type of calculation automatically, in seconds.
How many seconds?
Are they claiming that the travelling salesmen problem can be solved in polynomial time? This would be the biggest news to come out of the computer industry since the invention of the transistor. As far as I know, no quantum algorithms exist for solving NP complete problems such as the travelling salesmen problem. Can anyone here enlighten me?
In other news, CompTIA have released a working draft for their new Q+ exam - it's suitable for any engineer with 6 months' hands-on experience of Quantum Mechanics and GR. The pass mark is 80% and all 20 questions on the exam must be answered simultaneously.
AT&ROFLMAO
One or two bit at a time quantum computers - sure, we can build those. My hunch, however, is that to build an N bit quantum computer is exp(N) hard. I expect we will eventually have non-trivial quantum computers, but unfortunately the amount of effort to make them will be as much as the effort to build a classical machine that can simulate them. This isn't just nay-saying, unlike the claims that driving at over 30mph would kill humans, my claims are backed up by many physicists, in particular those that don't have a financial interest in quantum computers.
On the other hand, quantum computer science is very interesting as a branch of mathematics and Shor's algorithm for factoring, for example, is a thing of beauty. So I don't blame people bluffing in order to get grant money. And I suppose I don't really hold it against researchers trying to get money out of venture capitalists this way either. Just as long as that money isn't coming out of any funds I'm investing in...
Doesn't it make you feel good to know that our freedoms are protected by politicans, lawyers and journalists.
We'll, it's kinda cheating. The algorithm is STILL NP, but in quantum computing we can run all paths in parallel so we solve all possible combinaisons at once, which becomes polynomial. However, we have no way of finding the good answer at 100%.
See, the problem in quantum computing is that you can have multiple states in parallel, but you can only 'read' one and lose all other states. This is like having a book with 400 pages, but when you open it, it selects (with a certain probability) a specific page and the whole book becomes that page, you lose all other pages.
We need to make the system converge/interfere in a meaningful way to the correct solution, and in its own way, this is the challenge of QC. In the end, if our algorithm works, we will be able to get the answer to the travelling salesman problem with a probability (depending how good our convergence is). Just like our book above, we need to increase the chance of opening the book on the page with the correct solution. This is non-trivial.
The thing is, the 'weight' of that convergence/meaningful interference, in problems like the travelling salesman, is usually as high as the time it takes to run the normal algorithm in classical computing. We end up not having much gains, it's not that fast. So, yes, if they are that good, we can solve the travelling salesman dilema in seconds... with a certain, probably very low %. Probably even a meaningless %.
However, in problems like finding if a function is unanimous(f(x)=0 or f(x)=1 for all x) or balanced (f(x)=0 for exactly half of x and f(x)=1 for exactly the other half of x) could be done in quantum computer with no errors and very fast, while in classical computing you'd have to try each value of x. If you however allow a certain % of error, the classical way with a stochastic computer would work best (test only a certain pool of value).
Well...I don't think trotting out Einstein's example everytime a theorist makes a surprising claim is very productive. What the parent post was pointing out is that there isn't an absolute correspondence between our mathematical formalisms of physical laws and physical reality itself. Surprising things happen when our experimental limits are pushed...the mathematical model holds or sometimes it breaks. Afterall, Einstein wasn't a science celebrity after the publication of his first papers. It took the startling physical realization of his predictions, namely, the anamoly in Mercury's orbit.
By the way, it's extremely false myth that Einstein was bad at math.