I really, really hate it when science journalists perpetuate the myth that quantum computers are like superpowerful general computers. Please, journalists, stop saying "it has the processing power of a computer the size of the known universe" and start saying "it can perform one specific algorithm faster than a computer the size of the known universe". I'm reminded of an article I once saw on how soap bubbles where able to calculate something like "the minimum surface multipartitioning of a volume" much faster than a supercomputer could solve the problem. There's a big difference between being able to do one thing, and being programmable.

Decivre wrote:

Smokeskin wrote:

I really, really hate it when science journalists perpetuate the myth that quantum computers are like superpowerful general computers. Please, journalists, stop saying "it has the processing power of a computer the size of the known universe" and start saying "it can perform one specific algorithm faster than a computer the size of the known universe".

Agreed. Even if you don't take into account Turing-completeness, this "computer" is completely non-programmable, and designed for a specific formula. It is more akin to a "quantum sliderule" than any sort of computer. Still, it showcases that quantum computing is potentially feasible. Or at least feasible in this one scenario. In which case, hopefully this scenario is [i]extremely vital[/i]. :P

Do you share your forum account with the guy that wrote this in another thread?

Decivre wrote:

http://www.eclipsephase.com/no-limit-limited-technology-synthmorph-infom... Quantum computers have one major advantage over normal computers... the handling of massive numbers. [...] And when it comes to floating-point integers, quantum computers theoretically handle them incalculably better than a normal computer. [...] For example, let's say you had a quantum computer that had 64-qubit processor... meaning it could natively process values up to 64 qubits long. In order to have a traditional computer capable of handling the same size value, it would need to be a [i]16 exabyte processor[/i]. Then you have to think about storage space (64 qubits of space on quantum RAM, as opposed to 16 exabytes of space on traditional RAM... [i]for a single value[/i]), and the speed at which it can calculate through that data. Yeah, quantum computing is probably necessary for actual simulspaces. I imagine that classic digital computing is now relegated to low-end servers and personal computing (ectos and mesh inserts).

Smokeskin wrote:

Do you share your forum account with the guy that wrote this in another thread?

Sure do. But quantum computers are still largely theory. Even so, I know the basics on how they are suppose to work. But the one presented in the article isn't the same as a quantum computer within the setting. Quantum computers are theoretically capable of running programs. This one isn't... it's a crystal that calculates one thing. Two different concepts. However, one major point brought up in this article is crucial. Large-value algorithms work vastly better in theory when utilizing qubits. If an actual Turing-complete quantum computer is successfully built, the consequences could be revolutionary for data processing... potentially more so than the very invention of the original computer. I suppose all of that would be irrelevant if you are claiming that all quantum computers in Eclipse Phase are limited to calculating Ising spin states though.

Smokeskin wrote:

Where is that brought up? "Large-value algorithms" in general doesn't run any better on a quantum computer as far as I'm aware. Some very specific ones, like prime factorization, can be made to run much faster because certain steps can be handled very easily by a quantum computer.

Actually, a quantum computer's ability to handle large values is the key reason that scientists want to create them. Qubits can represent multiple values simultaneously, and exponentially more values as the qubit word size increases. That potential theoretically puts all standard computing to shame... it makes quantum computers near-infinitely superior in parallel processing.

Smokeskin wrote:

As far as I'm aware, when people talk of a quantum computer, they're really talking about a "regular computer that runs regular programs the regular way, but has a quantum module it can use to quickly perform certain operations". We're not talking about writing software that runs entirely on "qubit processors".

Very true. A quantum computer can only utilize its full potential when it is working with a quantum algorithm. And since it's likely that small calculations have no need for the formulation of quantum algorithms, it would make sense to hybridize them. But it would still be an insane breakthrough. The massive simulations we run on supercomputers today, as well as the processing that data farms utilize, could potentially be handled by much smaller systems with quantum processing.

Smokeskin wrote:

No, I believe they're limited to performing a few more tasks, which they do exceptionally well like performing prime factorization in order to break public key cryptography fairly easily. They're listed under Covert Tech and are mentioned as codebreaking machines. The Mesh section of the rules only mention them for codebreaking. If quantum computers had a more general use, don't you think they'd be mentioned as such? Like on page 247 for example?

It's a bit more than a few more tasks. Quantum computers make superior simulators for nanoscopic (or smaller) processes by quantum simulation (which potentially means they are crucial for organic simulation before we reduce the necessary data down to its essentials). They sort through and search unordered data quadratically faster by use of Grover's algorithm (making for the potential of insanely powerful search engines and data sifters). And do note that these are the algorithms that have been formulated [i]before a feasible computer has been produced[/i]. There's no telling how many other quantum algorithms can be designed while we are merely in pre-production. As for why they aren't mentioned in the books, I'd say that it's for the same reason that there isn't much info on servers, data farms, space ships, or other elements of the setting. It could also be because of the fact that Shor's algorithm is one of the most ubiquitous quantum algorithms, and perhaps the developers only knew of it's cryptographic potential.

Arenamontanus wrote:

However, they have an interesting drawback. When you end the superposition you can only get one value out. So while your N qubit system calculates all 2^N possible states, you will only get one answer. In a normal parallel computer you can get all states. Writing algorithms that do something useful despite this restriction is tricky. We have a fairly small library of quantum algorithms today, but there are enough really useful possibilities there to make quantum co-processors promising if they could be easily and cheaply built. Even expensive quantum processors would be useful for high-value computations like cracking codes, doing massive quantum physics simulations (important for nanotech) and some searches.

But the dearth of quantum algorithms might also be chalked up to the fact that quantum computing is still mostly just an on-paper concept. We've had breakthroughs in the field, but nothing even remotely close to a Turing-complete quantum processor. I imagine that the creation of algorithms will pick up nicely once programmers get their hands on one. Hell, even I'm sort of avoiding the quantum computing papers out there until I actually get to see a quantum CPU work its magic.

Arenamontanus wrote:

Fork opinion divergence is a big problem when you are widely distributed :-)

It is difficult enough when one distributes one's cognition across a cluster. One DDoS attack and suddenly the right parietal lobe goes away...