Quantum Computing Explained! (Well, Sorta)

An anonymous reader writes "Valiant effort to 'explain' quantum computing over on silicon.com — covering the difference between classical computers and quantum machines."

Re:chatty narrative

[Anonymous Coward]

Agreed. The chattiness simplifies some stuff towards poor wording or misrepresentation.
For example:
"Factoring massive numbers is what internet and banking security depends on, so security, financial services and military applications for quantum computers are easy to envisage."
RSA doesn't depend on factoring massive numbers. RSA depends on the fact that factoring massive numbers is computationally difficult.

Could be better.

Highly recommended book

[AdmiralXyz]

Have to agree with the comments above, that article is pretty useless.

Coincidentally, though, at a university book sale a few weeks ago, I picked up a copy of N. David Mermin's Quantum Computer Science: An Introduction, for just $5 (seems to be about $30 on Amazon) and I can't recommend it highly enough. It's an intro to quantum computing textbook, about 200 pages, written specifically for people who have CS or math (as opposed to physics) backgrounds, and while it's almost impossible to get into the nitty-gritty of why quantum computing works without a lot of quantum mechanics esoterica, this book does a great job of explaining how it works (which is plenty complicated on it's own).

It's not a light read (it's a textbook, after all), and contains some serious math, but it's nothing someone with a college education can't handle and it really helped me understand this whole mess better than any popular news article.

Re:Highly recommended book

[blueg3]

Incidentally, from Mermin's website [cornell.edu], you can download his lecture notes [cornell.edu] at no cost. The book is directly based on the lecture notes and, as far as I recall, the notes are pretty good. I took the class while he was working on the book, so all we had to work with was the lecture notes (which have since undergone some revisions), which were essentially a beta version of the book's text.

It should be reasonably understandable to someone with a good CS and mathematical background but limited physics background. (Likewise, it should be reasonably understandable to someone with a good physics background but relatively little CS.) The course was designed to be taken by both CS and physics students. I think it was fairly challenging for the Cornell CS undergrads that were in the course, but your mileage may vary.

Oh no, not again!

[maxwell demon]

From the article:

"This shared state means that a change applied to one entangled object is instantly reflected by its correlated fellows"

Why, oh why, is this nonsense repeated again and again. If you change one entangled particle, you do not change the other. For example, if you have two spins entangled in a way so that if one is measured "up" the other is measured "down" and vice versa, and you turn the one spin around (without measuring it) then you'll have an entangled state where if you measure the first spin "up" the other one is also measured "up", just as you'd expect. As long as you don't measure, there's no "spooky action at a distance" but only local changes. The "spooky action at a distance" happens at measurement (which BTW destroys the entanglement), and it's all but a given that there's indeed an action at a distance (you only need it if you want a certain type of interpretation, where basically "under the hood" the system behaves completely classical, but we don't see it because there are so-called hidden variables which we cannot determine). The point is that in an entangled state the correlation is all which is defined, and the result of local measurements are completely undefined (OK, strictly speaking this is only true for maximally entangled states, for others there's less correlation and more local information; it's basically a trade-off between the two). Now when you measure the spin of one of the particles, the value of the spin gets a defined (but random) value (up or down, in the direction you measure), and also the value of the spin of the other particle gets a defined value, which is determined by the entangled state and the result you got for the first particle, i.e. if the entangled state was "both particles have opposite, but otherwise undefined spin" then after measurement, the particles will have opposite, well-defined spin. However, since the result is random, if you have the other particle, you cannot see any difference whether the first particle has been measured or not; he will get a random result either way. Only if he gets told the measurement result of the first particle, he can predict (or, if he already measured, compare) his measurement result.

Oh, and yes, I'm working in the field of entanglement, so I know what I'm speaking about.

But I absolutely like the following statement from the article:

"Hang on, what's quantum entanglement when it's at home?
I was afraid you were going to ask."

I hope the above explanation is understandable ...

Re:Weather Prediction?

[Comrade Ogilvy]

Won't happen. While better computers help to some degree and building better models helps even more, weather is fundamentally a chaotic system. There is a hard upper limit based on the quantity and accuracy of the data. An exponential increase in overall quality of data yields a linear increase in the quality of the prediction.

To make up some numbers to illustrate the point...So if we have 10^4 weather stations to have 48 hours of good accuracy, it might take 10^5 weather stations to achieve 60 hours of good accuracy, assuming that you have all the computing power you could possible want in the first place.

Re:Horrible

[noidentity]

Yeah, with a 350-pixel-wide web page (yes, the entire page), and an opening like this, I can't imagine why nobody would read any further:

Time machines - oh, boy!
Steady on Sam, I love science fiction as much as the next geek but I'm not talking about Quantum Leap here. This is even more exciting than time travel. OK, so what is this quantum computing lark then? Quantum computing and quantum information processing are research efforts that seek to exploit quantum mechanical phenomena to perform tasks such as massively parallel computing. The quantum research field also encompasses quantum cryptography, which utilises quantum phenomena to guarantee secure communications.

What are these quantum phenomena you talk of?
Tsk! Clearly weren't paying attention in physics class were you? [...]

Tip to new writers: you aren't witty, you aren't funny, you aren't entertaining. Leave your antics out of the writing and cover the subject matter so well that its inherent nature will be interesting to the reader.

Re:Are quantum computers Turing machines?

[maxwell demon]

The set of problems you can in principle solve with a quantum computer is exactly the same as you can solve with classical computers. The best proof of this is that you can simulate a quantum computer with a classical computer (and vice versa). However, as far as we know you cannot simulate a quantum computer on a classical computer in polynomial time.

Re:Horrible

[bradgoodman]

So here's the way I understand it - or rather I probably misunderstand it:

.

Quantum Computing - because it stores superpositions of bits, which can represent all values, can work work on data with not just a single value, but with all possible values, thus doing stuff effectively in parallel

Before you stop reading - it gets better...

This is good for solving "what-if" problems - (which I think is technically described as "NP-Hard" - but I'm not positive) - or problems which can only be solved be trying any/every/a-shitload-of permutations, rather than figuring it out through an exact formula - like the "Traveling Salesman Problem" - or cryptographic analysis.

So for a crappy example: Suppose you were trying to figure out the square-root of 11 (and didn't know how, like me, and could only do it) by using successive approximation.

You would try "3" - which yields "9" - and then "4" which yields "16". So knowing it's between the two, you'd try "3.5" which yields "12.25" - then maybe "3.4" which yeilds "11.56", etc.

So if you were testing each variable "A", instead of setting "A" to be "3" or "4" or "3.5", "A" was a superposition of a bunch of numbers, you could somehow easily deduce which "A" was valid from a formula like "A*A=11".

I obviously don't know exactly what the mechanism is - and a better example would probibly be one which used a few binary digits (solved by using qubits) - which would have a finite range of possible solutions - rather than a floating-point number like I did in my bad example.

So - if you follow my train of thought - you could see how a cryptographic analysis problem which would require a lot of "what-if" quessing could be solved if all of the potential states of all of the bits could be instantly tested in parallel

Could anyone who understands this stuff chime in?