April 19th, 2003, 04:32 AM
A relatively simple device that sends individual photons cycling through a fiber-optic loop could provide the memory needed to make ultra powerful computers that use the quantum states of light as bits.
Quantum computers are potentially powerful enough to solve problems that are beyond the most powerful classical computers, including cracking the strongest secret codes and quickly searching huge databases.
Several research teams have shown that it is possible to carry out logic operations using the traits of individual photons -- the fleeting particles of light -- as quantum bits that represent the 1s and 0s of computing. Computers must also be able to briefly store the outcomes of logic operations.
Scientists at Johns Hopkins University have come up with a method for capturing photonic qubits for tiny fractions of a second, which enables them to briefly store information about the state of a quantum particle. The memory device consists of a storage loop and a switch that directs photons into and out of the loop.
The memory device stores a qubit by switching a photon into the loop, where it flies around at the speed of light, said James D. Franson, a physicist at Johns Hopkins University's Applied Physics Laboratory. A short time later, the state of the qubit can be read by switching the photon back out of the loop, he said.
The memory stores binary information that is based on the polarization of photons. A photon is polarized when its electric field vibrates in one of four directions: horizontal, vertical and the two diagonals. The directions are paired, and one of each pair can represent 1 and the other 0.
The researchers used a polarizing beam splitter, which is transparent to one polarization and acts like a mirror to the other, to shunt photons into and out of the loop. The beam splitter separates the two polarization components of the photon, causing one to loop in one direction and the other to loop in the opposite direction. "You can envision these components as traveling in counterpropagating directions through the device," said Franson.
It is only possible to split the polarization components of a photon when the photon is in the weird state of superposition, meaning it is in some mix of the two polarizations at the same time. Quantum particles like photons enter superposition when they are unobserved and otherwise isolated from their environments.
When the photon in the loop passes the opening, it goes through a switch. When the switch is closed, it continuously flips the values of the photon's polarization components, turning horizontal polarization to vertical and vice versa. This causes both parts of the photon to hit the mirror portion of the beam splitter, which keeps the photon inside the loop. When the switch is opened, it no longer changes the polarizations and the photon passes through the beam splitter and exits the loop in the same superposition state as when it entered.
A photon takes 13 nanoseconds, or billionths of a second, to make one round-trip through the memory device, said Franson.
Optical quantum computers are likely to employ laser pulse trains, or pulses of laser light fired at regular intervals. "These pulse trains provide a natural clock cycle for the various quantum logic operations [and] memory readouts," said Franson. The cyclical nature of the memory device fits well with this type of architecture, he said.
In principle, the researchers' device is resistant to errors caused by light-phase shifts, said Franson. As a photon makes multiple passes through the storage device, its wave can gradually stretch or compress at different rates depending on polarization. These changes are neutralized, however, because the storage device repeatedly flips the polarizations, said Franson. "These phase shifts essentially factor out of the final state and may, in some applications, not affect subsequent computations using the stored qubits," he said.
Although researchers have known for a long time that optical fibers can store photons, "this might be the first demonstration," said Eli Yablonovitch, a professor of electrical engineering at the University of California at Los Angeles.
The researchers' device "is a very cute way to provide a limited amount memory" for linear optical quantum computing, said Jonathan Dowling, a principal scientist and supervisor of the quantum computing technologies group of at NASA's Jet Propulsion Laboratory. Its potential uses are limited because "it likely cannot robustly hold the qubits for very long periods of time required for... quantum communication applications such as quantum optical repeaters," he said. Repeaters boost fading signals along communications lines.
The researchers' current prototype cannot store information long because it suffers from photon loss, said Franson. "We estimated about 19 percent loss per cycle, which means we really couldn't store the qubits for very long," he said. In principle, the loss can be overcome by a better design, custom optics and possibly new types of fiber optic components, he said.
Scientists are exploring other means of storing optical qubits, including trapping photons in special semiconductor devices and transferring quantum information from photons to groups of atoms. "Many of these techniques rely on very clever manipulations of fascinating physics," said Franson. The researchers' method is less interesting for basic physics, "but may have some technical advantages for certain applications in the near term," he said. The devices are relatively simple and their timing corresponds to the repetition rate of commercially available lasers commonly used in optical quantum computing experiments, he said.
The researchers are now working on storing a pair of entangled qubits in a pair of synchronized cyclical memory devices, said Franson. Controlling entangled qubits is key to unleashing the power of quantum computing.
If two particles in superposition come into contact, one or more of their properties, like polarization, can become linked, or entangled. If two photons have their polarizations entangled, when one of the photons is measured and leaves superposition, the other photon leaves superposition in the same instant and assumes the opposite polarization regardless of the distance between them.
A sufficiently long string of qubits in superposition can represent every possible solution to a particular problem. Entanglement allows a quantum computer to check all possible solutions with one set of operations. Ordinary computers are much slower because they have to check answers one at a time.
The cyclical memory device could be used in practical applications in five to ten years, said Franson. Researchers generally agree that full-scale quantum computers are 20 years away.
April 19th, 2003, 12:28 PM
Dammit! I want one NOW!!
I first read about quantum computing when I read The Code Book by Simon Singh and learned about Quantum Cryptography. The possibility of these computers will be fantastic. Now, instead of TRUE/FALSE we'll have 3 possibilities: TRUE, FALSE, and sorta TRUE AND sorta FALSE. A trinary numeric system (*UGH*) and a whole world of possibilities.
BTW, when I was looking for a definition of qubit, I found this: http://www.qubit.org/ .. check out the Tutorials there.
April 19th, 2003, 01:55 PM
Read Etching in the stone by Dannie Hillis (guy who founded pixar) Great computer book and it gose over why quantum computers will not be that great of a setp. Basicly A computer is a finite state machine, as such it can emulate any type of finite machine...ie a analog compter can emulate a digital computer , can emulate a quantum computer (he show an example of a tinrker toy computer). If you realy want to see waht quantum computeing then check out the quantum perl modual that emulates quantum computeing right now.
April 19th, 2003, 02:06 PM
April 19th, 2003, 06:08 PM
i've been to three lectures on the topic of quantum computers by researchers from the u of michigan, who have actually demonstrated some of the concepts. their site is http://www.umich.edu/~newsinfo/Relea...02/061202.html
i'm not going to claim to understand half of what they've accomplished, it's pretty amazing. my biggest questions is, who really needs these computers? everyone seems so hyped up on whether or not it can be done, none of them have offered any applications for these mulitmillion dollar machines. they kind of gloss over encryption and cryptanalyis, but these results we get from the computers are only accurate like half the time. it just seems like a bigger, faster why to get wrong answers.
i think my point is that even when moore's law does finally fail, we're not all going to be running out getting quantum computers.
U suk at teh intuhnet1!!1!1one
April 21st, 2003, 12:28 AM
Encryption is one of the biggest applications for quantum computers. But not the only one. Another application could be statistics (or prediction of statistics to be more precise).
What do I mean by statistics. Take horse racing for example. You can easily see with 7 horses how many have each has won or lost. Its easy to see what the chances are of each horse winning. But now betters are getting into things like, what food did they eat in the last week, who put on the last set of horse shoes, and how much oat bran did the jockey eat the night before. Okay, so now lets take a race with 100 horses, and you want to test it with each of the 4 statistics mentioned. Not toooo hard, but taking more time. So what if we had 1000 horses, and we had 1000 statistics? And how much credibility does each stat get? If a horse has a broken leg then even if it was #1 in every other stat, it shouldn't be weighed the heaviest. It can be done with normal computers, but would take a while. With Quantum computers, every combination can be checked at the same time.
Ok, so that is a genral idea, but still kind of a waste of time for a super duper computer like the quantum computer. But lets take another "statistic" The weather. There are sooo many variables and different thinigs that can affect changes in the weather that it is extreamly difficult to predict. The largest/fastest computer in the world (I think its in japan right now, but can't remember) has a hard time with the weather, and that is the only thing it does. It accurately predicts something like 6 hours away, longer then that, and its still just guessing. But with a quantum computer (obviously still a pretty large one) it could be done a lot more accurately. Because multiple possilble changes in the weather can all be tested at once, it could speed up the results. I haven't seen any data on how fast it could go, but it only makes sence that it would be a lot faster then the current machines.
\"Ignorance is bliss....
but only for your enemy\"