Slashdot Mirror
Quantum Computers
joecool12321 writes: "Although Richard Feynman spoke about quantum computers in 1981, technology is only now starting to catch up. This article at Scientific American discusses recent developments towards the goal of 'infinite computing,' and research is showing that scalibility may not be far away, and thus scalable qbits."
A reliable /dev/random! /dev/null that can hook up to a USB port, everything would be fine.
Now if only I could get a
Hopefully I didn't put any [] around my words.
Actually let me clarify a bit of this. First of all an example of what it means to be reversible. The best example of this is that you can't clear memory / registers. Setting something to zero is a destructive, non-reversible process. Basically any "program" run on a quantum computer would be runnable backwards and using the all the outputs you could find all the inputs. Even a simple program like C = A | B would have to keep another bit of data, a "D" that would enable you to reconstruct A and B using C.
Now the types of things that quantum computers would in theory do really well take advantage of being able to use an input state that is a superposition of all possible inputs. The prime example is factoring huge numbers. The number to be factored is entered as one input to the process, and the second input to the process is a superposition of every number from zero to the number to be factored.
The quantum computer then divides the number to be factored by this input vector, and retains the remainder, which is a superposition of all possible remainders from that division.
Now there will be patterns in the remainders from the division, and if you take a fourier transform of those remainders you will get big peaks that correspond to the factors.
At this point your calculation is done so you measure the output. Remember that everything that has happened so far has been happening internally to the "quantum computer" and has not been observed. Your observation of the output collapses the probability and you get one output point, but if you repeat this operation a hundred times or so, most of your output points will be somewhere in this peak.
The cool thing about this process is that it takes advantage of the fact that you can do a fourier transform in the intermediate step before you collapse the probability. To get enough points to do a Fourier Transform in the intermediate state in a traditional computer you'd need to get thousands input vectors, but the quantum computer only needs one.
(btw, IANLAPBITEPIS (I am no longer a physicist but I took engineering physics in school), so if I messed up somewhere here and someone can correct me please do)
Quantum computers will likely mark the end of Moore's law. They don't scale the same way as conventional computers. Think of a quantum computer as being able to process 2^n values (where n is the bits per item being processed) for every single value processed in a conventional computer. If we can keep the same kind of mojo going for quantum computers that we've kept going for classical computers, then we might rougly experience computing power squaring every 18 months or so. (Of course, there are a whole new set of challenges facing us in the field of quantum computing that we've never seen in classical computers.)
What got me about Penrose is that he goes through the gyrations of showing the limitations of computers. GEB did a better job of that IMHO. But then he fails to prove to my satisfaction that humans have a special insight into math that computers cannot have.
If tits were wings it'd be flying around.
In the news today, a local overclocker and avid QuakeIII player died in an accident when his AMD Megatron 100,000,000 GHz suddenly overheated and self destructed.
Fellow gamer, who goes by the name "1337", was in a game at the time that the accident occured. "The last thing I remember he was up about 15,000 frags, and boasting about how he was getting 2,000,000 FPS on his machine." 1337 said. "Then suddenly he disappeared. Served him right, he had an unfair advantage."
Experts who have been investigating the accident have managed to peice together a probable chain of events. "From what we can figure, a pump failed in the cooling tower that he had in the back yard, and caused the chip to overheat" claims Cho Man Foo, a scientist at Los Alamos National Laboratory, one of the many expert quantum phyiscs experts called in to reconstruct the chain of events. "He may have survived the incident," states Mr. Foo, "except when the chip detonated, it started a chain reaction that caused the magnetic shiels around his Segate black hole quantum hard drive to collapse, imediatly consuming everything within 500 feet."
Pictures of the aftermath can be found here.
In related events, LAN party co-ordinators have been advised to postpone all further gatherings untill a meathod of confirming that no overclocked machines be used in game meetings. AMD and motherboard maker ASUS will be taking steps to insure that further incidents will not happen. In the next version of their products, the clock multipliers will be locked at the factory, hopefully preventing further injuries from their products. Also, tobacco giant Phillip Morris is suing AMD, claiming that the chips cause danger to their large customer base, and they could potentially loose a large number of customers if there was to be a major accident.
Related articles:
Overclocker Creats Rift in Space-Time Continuum
"Everything that can be invented has been invented."
--I assume full responsibility for my actions, except the ones that are someone else's fault.
One thing I've never seen explained is whether there's a way to link together multiple QC widgets (No, I didn't say a Beowulf Cluster of them! :-) or whether you're limited to the resolution of one widget, which Heisenberg limits to a value around Planck's constant (~10**-46 = ~140 bits.) If you could do the physics and precision construction to get this resolution, it would be lots of fun, but it doesn't fundamentally change cryptography, because it doesn't get you unlimited exponential growth - you can always add another 140 bits to your key length.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
so its not exactly the sort of room-temperature you'd want to have in your cube - unless you happen to be Mr. Freeze.
There are a thousand forms of subversion, but few can equal the convenience and immediacy of a cream pie -Noel Godin
will we see the dawn of Quantom Computing? I don't see why not ... but the sheer power, programming that would go in to it, and the understanding (hell it's still a thereom), will make this a reality further in the future ..
still is fun to dream ...
Ignore the "p2p is theft" trolls, they're just uninformed
Sometimes I wonder if AI and Quantum Computers wouldn't compliment each other nicely.
I mean, I've always suspected that true, self aware computers might only be possible in a Quantum form.
Sure, we can do some excellent AI with faster Digital Computers, but for a system to be both intelligent and diverse it needs to be able to store a lot of data and process all of it quickly.
With today's computers, one can assume that the the more complex the information an AI is dealing with, and the more it "Learns" the more it has to process. Theoretically, this wouldn't be a problem for a Quantum Computer.
"Good morning computer."
"What's so good about it? You're just going to ask me to check your e-mail, read you the news at Slashdot, and give you the stock report. Then you're going to drink your coffee and head off to work, leaving me here alone as ussual. Good morning indeed."
"Everything you know is wrong. (And stupid.)"
"Everything you know is wrong. (And stupid.)"
Moderation Totals: Wrong=2, Stupid=3, Total=5.
I remember seeing something about atom trapping. I was able to find a tone down version of the Science magazine article here: www.academicpress.com/inscight/06022000/graphb.htm
Schmiedmayer, who's mentioned in the parent story, is also in this story from mid-last year.
A recent slashdot article that I submitted also concerns the aspect of using silicon buckyballs as cages for qubits.
The crux of the matter still remains unsolved in this SciAm article, and I have yet to see any explanation on how to solve it in any of the scientific journals that I read: that is, we don't use pure quantum states to preserve the very fickle quantum condition. When we can do that - there have already been enough postulation on what a qubit can consist of - then we can seriously consider quantum computing in the future.
---
---
I'm just an ordinary man with nothing to lose.
Michael Crichton's latest book, Timeline, has a pretty interesting spin on Feynman's work and references it a bit if anyone is interested in his work (It also happens to be a damned good book, I might have to say it's his best work).
Fuck Ajit Pai
If you're interested in quantum computing, try to check out Damian Conway's talk on the Perl module Quantum::Superpositions. It's very funny and actually quite useful!
Damian is travelling around the world talking to perl user groups. Check out his schedule to see if he's due to talk near you.
This is (was?) Roger Penrose's postulate as to why true AI was not possible -- because the human brain has truly random processes, and computers do not. His theory, not mine. I think it's hogwash.
You might need a quantum computer to find the information necessary to build one. No wonder it's so difficult.
In what sense is quantum computing, "infinite computing"? Computing is inherently finite. See typed lambda-calculus, intuitionistic logic, and Curry-Howard isomorphism, if you don't understand why computing means finite.
Now, I anticipating someone bringing up non-terminating Turing machines and untyped recursive lambda-functions as counter examples to my claim... Those don't represent computing! They are nothing more than short-comings of our formal systems. They have no logical meaning. Mathematicians, logicians, and computer scientists have put great effort towards gaining an understanding of computing that weeds out such nonsensical non-constructs.
An effective procedure that doesn't terminate with a result is not an effective procedure. Now, if by "infinite", you mean unbounded, then you have done nothing more than abuse terminology. "Infinity" does not mean the same thing as "unbounded", in a computational sense of the words.
A good place to start would be this article. With quantum encryption you can determine whether there's an eavesdropper due to quantum entangled particles. Any eavesdropping will create a detectable effect on the message. There's relevant slashdot articles here and here.
For those who don't already know, an editor put together Feynman's lecture notes into a great book, which he used during his famous 1980s lectures. It's worth the $30, if you are interested in computers, computation theory, physics, or you just really like Feynman's stuff. This book is one of those classics that everyone has on their shelf.
From the article:
"In five years we will know if it's an interesting physics problem or if it's really something that we can use"
Unfortunately, this was the general opinion 5 years ago, and it will probably be the general opinion 5 years from now. It's like with AI, we're always on the cusp of a breakthrough, but that breakthrough never seems to come.
Ah well, someday...
He who joyfully marches in rank and file has already earned my contempt. - "Big Al" Einstein
If Moore's law has anything to say about it, these will become practical immediately after the 1 atom transistor CPU becomes commonplace. After that, the limitations of silicon will have been reached, and we will need to use a new technology.
The question is, who will have this technology? The early computers were the sole preserve of governments, but transistorised computers were more commonly seen in big businesses.
Early silicon based designs were bought solely by home hobbiest. Later silicon monsters once again went ot business, but this time those of all sizes. Will we see these more widely spread than the current breed, or more selectively targeted?
First, I'd like to point out that quantum computation and quantum encryption are two almost completely separate concepts. Quantum encryption is based on the fact that quantum states cannot be measured without altering. The most common example is the polarization of a photon, but it will work for any quantum state, so long as there exist, effectively, two unique states that can transmit the data.
Quantum computation, however, is much more complex and much more interesting. Quantum computers are based on the concept of quantum entanglement, the ability of a quantum state to exist in a superposition of all of its mutually exclusive states: It's a 1 and a 0. However, this is not as easy to use as one might think. While it's true that if you have n quantum logic gates you have the ability to input 2^n data values simultaneously (as opposed to only 1 piece of data if you have n digital logic gates), this is not going to be the end of classical computing for a few reasons. First, quantum computers have to be perfectly reversible. That means for every output there's an input and vice versa. And there has to be no way of knowing the initial states of the data. You don't process data, you process probabilities in a quantum computer; if you know exactly what any one value is throughout the computation, you can find out all of the values: the superposition ends and you're stuck with a useless chunk of machinery. This means YOU CAN ONLY GET ONE RESULT FROM ANY QUANTUM COMPUTATION, THE END RESULT. You can't see what the data in the middle is or the computer becomes useless. (Landauer's principle makes heat loss data loss. When your processor gets hot, it's losing data. If the same thing happened to a quantum computer, it wouldn't be quantum anymore.) Decoherence is what happens when you randomly lose data to the environment by design, not by choice, and the superposition ends. This is bad for Q.C. Oh, and quantum computers can only do *some* things faster, like prime factorization and discrete logarithms. Not multiplication or addition. Plus, the circuits that would do basic arithmetic would be bigger and slower than what you've currently got.
So what does this all mean? It means that quantum computers are going to provide some advantages (real quick big number factorization), and some disadvantages (that whole RSA standard). The most realistic initial use of quantum computers will be as add-ons to existing super-computers to resolve certain types of NP-Complete headaches that regular math can't simplify yet. At best they will someday be an add-on to your PC; but they will never replace the digital computer.~
If you want more info, check out ahttp://www.qubit.org, it's got some decent tutorials.
Papers, 1933-1988. Feynman's correspondence, course and lecture notes, talks, speeches, publications, manuscripts, working notes and calculations and commentary on the work of others are all included in this extensive collection.
Collection size: 91 boxes, 39 linear ft.
That's a big twinkie.
--- Hot Shot City is particularly good.