IBM Quantum Computer

[Alan McDougall]

http://www.fastcompany.com/1821378/the-greatest-innovation-yet-making-computers-quantum-could-change-the-world

 

101010: That's the number 42 represented in binary, which is the mathematical way today's binary computers see every single piece of information flowing through them, whether it's a stock price, the latest Adele track, or a calculation to generate an MRI of a tumor. But now IBM believes it's made progress in developing quantum computers, which don't use binary coding. It is not overstating the matter to say this really may be the ultimate answer in computing machines. Quick, mop your brow and don't worry: The science isn't too hard to grasp and the revolution, when it comes, could rock the world. In a very good way.

 

First, a little background: Computers today, everything from the chip controlling your washing machine cycle to the screen you're reading this on, rely on binary math to work. This reduces the information in problems you ask a computer to a counting system based on just "1"s and "0"s. That translates beautifully into the electronics of a computer circuit: A "1" matches up with a little burst of electricity, a "0" means none. By shuttling trillions upon trillions of these pulses, called bits, through tiny silicon circuits and transistor gates that flip their direction or trigger an ongoing signal, the chip does math with these ones and zeros. It's a mind-bogglingly complex and very swift dance that ultimately results in Angry Birds playing on the screen of your iPad. Or, after kajillions of calculations more in a supercomputer, it results in a model predicting climate change.

 

Now, what if instead of simply being able to do math with ones and zeros, a computer chip could work with bits that included other numbers? You'd have to design more complex circuitry, for sure, but it means every single one of those tiny electronic calculations that's happening every millisecond could tackle more information at once, and would ultimately mean a more powerful computer that may calculate faster. Got that? Good. Now how about if instead of a one or a zero, your computer's "bits" could have any one of an infinite number of values?

 

That's quantum computing. Essentially this moves way beyond the well-known physics of electronics, and on into the weird and wonderful world of quantum physics--where bizarre twists of the laws of the universe mean a "bit" in a quantum computer could hold both a "1" and a "0" and any other value at the same time. That means the circuits of a quantum computer could carry out an incredibly huge number of calculations at the same time, handling more information at once than you can possibly imagine.

 

By using some other very strange physics (superconducting materials cooled to hundreds of degrees below freezing) IBM's research team is trying to build some of the core components of a quantum computer, and has made big progress. They're now saying they've made the quantum "bits" of information, also called qubits, live a lot longer before they essentially get scrambled. They've also worked out how to speed up the actual quantum computing circuit. IBM's progress is so impressive that they're now confident a quantum computer could be made sooner rather than later, perhaps as close as 15 years away.

 

Whenever it arrives, the world will change.

 

On a very simple level, this is because instead of asking a supercomputer to work with endless strings of "1"s and "0"s to calculate all the variables in, say, a global warming simulation (performing trillions of small math calculations one after the other to work out the dynamics of the climate over a period of hours or days) a quantum computer would be able to process much of the math at the same instant instead of sequentially. Which could reduce the compute time to a second or less. Which ultimately means better and more accurate models of the climate. Similar processing tricks could improve medical imaging, or maybe even simulations of your own particular disease's spread, which may improve treatment.

[Greg H.]

And let's not forget that quantum computing, when it does come, will render all existing forms of computer encryption not only obsolete, but completely useless.

 

"1024-bit encryption? This will just take a second."

[juanrga]

Am I allowed to be a bit skepctical? I am still waiting for promised AI (what happened finally with Kasparov accusations on IBM's Deep Blue?) and I am waiting for promised optical computers 1000 times faster than electronic ones.

 

Also I cannot see how "a quantum computer would be able to process much of the math at the same instant instead of sequentially" in nonlinear and/or non-parallel problems.

[Greg H.]

Am I allowed to be a bit skepctical? I am still waiting for promised AI (what happened finally with Kasparov accusations on IBM's Deep Blue?) and I am waiting for promised optical computers 1000 times faster than electronic ones.

 

Also I cannot see how "a quantum computer would be able to process much of the math at the same instant instead of sequentially" in nonlinear and/or non-parallel problems.

 

George Johnson's book A Shortcut Through Time: The Path to Quantum Computer (Johnathon Cape Publishers, London: 2003) offers some insight into how this works (in theory).

 

Each of the possible factors to test can be thought of as a quantum ripple, one piece of a complex wave packet. Computing the answer is equivalent to massaging the the wave with the laser so that all the possible answers interfere with each other, the least likely cancelling each other out and the most likely reinforcing each other. Finally the wave collapses to reveal the solution.

 

The book goes into a good deal more detail about the theory and the physics, as well as developments up to that point in time. One of the things that quantum computing can do very rapidly (in theory) is factor very large numbers into their relevant prime factors - which is the heart of modern computer encryption algorithms.

 

On page 89, Johnson goes on to point out why this is such a big deal to modern computer encryption. With a modern Strong RSA encryption, you get a 1,024 bit encryption scheme. With a 1,024 qubit computer, you could

...try out all the possible factors simultaneously, in superposition, then collapse to reveal the answer.

 

The reason long encryption codes work better is because modern computers, even supercomputers, are constrained by the laws of classical physics (Johnson, 89). With Shor's Algorithm (See Johnson, pp 67-82 or Shor's Algorithm on Wikipedia), and an appropriate quantum computer, you could compute the prime factors of an encryption string in far less time.

Edited July 14, 2012 by Greg H.

[imatfaal]

Scott Aaronson's blog Shtetl-optimizedis a great place for explanation of quantum algorithms, p v np etc

 

Here is his everytake on Shor's Algorithm http://www.scottaaronson.com/blog/?p=208

[Greg H.]

Scott Aaronson's blog Shtetl-optimizedis a great place for explanation of quantum algorithms, p v np etc

 

Here is his everytake on Shor's Algorithm http://www.scottaaro...com/blog/?p=208

 

Nice read. Thanks for the link!

[ivanv]

This would be particularly interesting for getting thrillions of possible specific AGI problem solutions by some standard proof searching method.

 

I just can't wait for the future