[samiam]

BM scientists said Tuesday that they have made several breakthroughs in quantum computing that put them "on the cusp of building systems that will take computing to a whole new level."

The IBM researchers said they have established three new records "for retaining the integrity of quantum mechanical properties in quantum bits, or qubits, and reducing errors in elementary computations." Those advances, presented at this week's annual American Physical Society meeting, get the team "close to the minimum requirements for a full-scale quantum computing system as determined by the world-wide research community."

nerdgasm

"The special properties of qubits will allow quantum computers to work on millions of computations at once, while desktop PCs can typically handle minimal simultaneous computations," the IBM researchers said. "For example, a single 250-qubit state contains more bits of information than there are atoms in the universe.

"These properties will have wide-spread implications foremost for the field of data encryption and other possible applications such as searching databases of unstructured information, performing a range of optimization tasks and solving previously unsolvable mathematical problems," the team added, further elaborating on the possibilities of quantum computing in the video below.

Enjoy being unable to secure your passwords by this time next year! Also Skynet.

[redsniper]

Wait, I could have sworn that IBM or someone was already selling quantum computers, as in they already had some ready to go, or nearly ready at least. I guess they were farther off than I thought. Maybe they were just looking for potential customers. Ah well, cool stuff either way. FSO 3.7 is going to support quantum CPUs right?

[Mars]

Wait, I could have sworn that IBM or someone was already selling quantum computers, as in they already had some ready to go, or nearly ready at least. I guess they were farther off than I thought. Maybe they were just looking for potential customers. Ah well, cool stuff either way. FSO 3.7 is going to support quantum CPUs right?

I suspect you're probably thinking about quantum encryption.

[redsniper]

Nah, I'm pretty sure it was honest-to-god quantum computing they were talking about. I bet I just misread the story or something. Maybe they're taking preorders...

[Mars]

I just realized that this will probably result in machines smarter than us.

[redsniper]

These will be the last computers designed by humans.

[SpardaSon21]

I for one welcome our new robot overlords!

[Bobboau]

just set the computer's goal on enhancing human cognitive abilities, rather than on making better computers and we will be the new robot overlords.

[KyadCK]

First they make a 155GHz CPU.

Then they can fit one bit on 12 atoms for an HDD.

Now Quantum Computing.

IBM has some cool ****.

[Klaustrophobia]

this is all well and good, but until i can play crysis on one i don't care.

[Sarafan]

Can someone explain to me what is a quantum computer?

[samiam]

It's a computer with really tiny bits that uses quantum physics to compute exponentially faster with each additional bit you throw in.

You can read the article, where it explains it, or just think of it as a really, really fast computer from Star Trek or The Matrix.

What surprises me is how they're coming out with this about 20 years earlier than I thought they would.

[General Battuta]

Quantum computers are very interesting because (at least from the theory I know) they are absolutely incredible at some forms of computation that digital computers are awful at, and absolutely awful at some forms of computation that digital computers are good at.

I may be completely wrong though

[Grizzly]

Isn't the problem with quantum computers that observing the results alters the result?

[General Battuta]

Isn't the problem with quantum computers that observing the results alters the result?

No.

e: I should clarify. Handling quantum decoherence is a huge problem in designing quantum computers, yes. But it's very easy to be misled by popular discussion of quantum physics.

[samiam]

Quantum computers are very interesting because (at least from the theory I know) they are absolutely incredible at some forms of computation that digital computers are awful at, and absolutely awful at some forms of computation that digital computers are good at.

I may be completely wrong though

No they're good at everything. They're just really hard to make.

Where's Hera?

[General Battuta]

No they're good at everything. They're just really hard to make.

Where's Hera?

Herra is an enthusiastic amateur about many things, and while the enthusiasm is laudable, you should always keep in mind the amateur.

I'm not great at this domain of computational physics but I've reached out to someone who is. Moment.

[General Battuta]

No they're good at everything. They're just really hard to make.

Here's the verdict as I understand it: while yes, in theory, a universal quantum computer can do anything a universal classical computer can do, assuming Turing-style infinite memory, in practice decoherence protection becomes escalatingly difficult as the number of qbits in the quantum computer scales. Even with optical lattices, topological computing, what have you, this is still an issue. (IBM has doubtless made enormous strides in this field if they're claiming en-cuspedness.)

By contrast, classical computers can be built with an essentially arbitrary number of bits and the cost doesn't scale too sharply (linear or log function, I believe).

In practice what's likely to happen in the near future is that quantum systems will serve as specialized processors, loosely analogous to GPUs, that target specific operations. (When one of those operations is something like Shor's algorithm this is still a huge deal.)

Again, though, I am prepared to be totally wrong.

[Nuke]

Isn't the problem with quantum computers that observing the results alters the result?

back in the olden days of computing when they used core memory, simply reading the data destroyed the data, so after reading a bit of data you would have to re-write it. they actually designed the computer architecture around these limits, it also altered the programming paradigm. you would do a lot of inline operations, read a var, do computations on it and then store it back into memory at the same location. if you needed to free up some ram, just read a bunch of vars you werent using anymore.

core memory is awesome though because it was not volatile. you could power it off at the end of the work day and come back next morning and resume where you left off without having to load any punch cards. the **** was durable as **** too. some core memory recovered from the space shuttle challenger (it was used because of its immunity to radiation) was still in tact and readable, they were able to do a core dump and analyze the data. yes the phrase core dump comes from the core memory days. nothing to do with quantum computing but its cool ****.

[samiam]

It's good that you're ready to embarrass yourself, I am too.  They are hard to build, because of decoherence as you mentioned. They also are more suited to parallel processing than sequential processing, which is what I think you're getting at. Quantum computers solve BQP problems efficiently while classical computers solve BPP problems efficiently, if you want to sound smart.

Quantum computers use the phenomenon of quantum entanglement. It's possible to entangle two quantum particles so that certain states on one particle (or qubit), such as spin, affect the state of the other qubit.

You know how logic gates work, right? The NOT SEXP and the AND SEXP? Well let's say two qubits are entangled so that qubit A always has the opposite spin of qubit B. So if qubit B is manipulated to have an up spin, A will have a down spin. We've just made a NOT logic gate.

Now say that qubit C is entangled with both in such a way that it will only display an up spin if both D and E display an up spin. That's an AND logic gate. And so on, recreating the entire series of logic gates that exist in any other processor.

The great thing about quantum computing is that each qubit may be entangled with every other qubit in the processor at the same time, so you can perform exponentially greater operations if you can increase the number of qubits entangled with each other. Basically, infinite money. Exponential versus linear. Although everything's log, really, if you take it far enough.

However, if you were to write a program for the quantum computer, it would have to wait for each gate to complete its current operation before proceeding to the next step. So one should always try to construct the problem in a way that allows parallel processing to get the most juice out of each step when writing code for these machines. But the high speed of data transmission between qubits (near instantaneous) and the exponentially increasing parallel capabilities with each additional qubit makes these computers a whole lot more effective than classical computers for most problems if the programmer isn't an idiot.