Australian Researchers Demo Random Access Quantum Optical Memory 74
nuur writes "Researchers at the Australian National University have developed a new form of optical memory that allows random access to stored optical quantum information. Pulses of light are stored on a kind of 'optical conveyor-belt' that is controlled with a magnetic field. By manipulating the magnetic field, the conveyor-belt can be moved, allowing the recall of any part of the stored optical information. The research is published in Nature." You'll probably know after reading the abstract linked whether you'd be in the market to pay for the whole thing.
All I know (Score:5, Funny)
All I know is that my head hurts.
Re: (Score:1, Offtopic)
Yeah, well (Score:5, Funny)
For my storage requirements I need something more reliable than "random" access. Sheesh.
Re: (Score:1)
Hmmmm...... You run w/ no RAM?
Re: (Score:2)
I have random access to all of your data, one bit at a time. Here's a little example to prove it (sorry if this freaks you out):
0
Re: (Score:1)
Re: (Score:2)
Nah, it's encrypted using an algorithm that results in all 1's. You need to learn the key as well, so you know which bits need to be flipped to 0. But once you know the key, the actual data can be decrypted using the simple bitwise AND operation between the data (all 1's) and the key.
Re: (Score:1)
Re: (Score:2)
For my storage requirements I need something more reliable than "random" access. Sheesh.
Oh it's very reliable. You just don't know what your data will be until you read it.
And then it's gone.
Re: (Score:1)
RAM?? (Score:5, Insightful)
Doesn't really sound like 'RAM". Sounds more like a tape storage device. The term 'RAM' was coined to indicate any spot in memory could be accessed in generally equal time. Tapes have to rewind, move.
Re: (Score:3, Interesting)
But it could be fast enough that it doesn't matter. Perhaps someone that read the brief could chime in?
Re:RAM?? (Score:4, Funny)
Re: (Score:1)
Being fast != similar access speeds.
Re: (Score:1)
RAM is not about O(1) read latency. Quantum information can be read only once, so _on-demand retrieval of arbitrary quantum states of light_ may prove useful in quantum computation.
Re: (Score:1)
Sounds like someone is about to drop out of CS and head towards a liberal arts degree.
Read up on what RAM was/is and get back to us. It'll do you well to improve your HS Freshman midterm exams in basic CS theory.
Of course it is not O(1)! (Score:2)
It is (under physical constrains in out Universe, speed of light and such) is O(log(N)) in the best case (for N being number of addressable locations).
And "conveyor-belt" would imply O(N) access time, which, in my book, is not RAM, but more like tape or HDD (possibly flying by at the speed of light, but still linear, not logarithmic!).
But the experiment itself might be cool, everyone who have seen an optical table before should check out "A top view of the experiment" http://photonics.anu.edu.au/qoptics/ALE [anu.edu.au]
Re: (Score:2)
Re: (Score:1)
How are you going to figure out *which* cell you're going to read from the RAM without looking at each of the O(log N) bits of the address? The decision to read from cell N depends on all O(log N) bits of N. It's not immediately obvious to me that *must* require O(log N) depth in your control circuitry, but it certainly isn't constant unless you allow unlimited fanin/fanout.
Re: (Score:2)
In practice, memory addresses on a given architecture have a fixed size (usually 32 or 64 bits), and the hardware can look at all the address bits in parallel, thus allowing O(1) access to any location in RAM. It's the same reason that you can add two integers in constant time.
Re: (Score:2)
Of course O(log(N)) for *fixed* N is the same as O(1)! the dangers of O-notation is that in different contexts people can to disagree what relevant N is. Yes, you can sort N fixed-length integers in NlogN time, but what if integer length is growing?
As to memory access, one can argue that there is also O(sqrt(N)) term in it for conventional semiconductor RAM, organized as squarish matrix: one can access it only as often as it takes to charge word line (with capacitance proportional to sqrt(N), where N is num
The term you're looking for... (Score:2)
is Bubble Memory [wikipedia.org]
Re: (Score:2)
If you want to be pedantic about it, all those little electron-y bits and whatnot move around in RAM too.
So I guess I have four gigs of tape drives plugged into my motherboard right now.
Cool. (Score:5, Funny)
It's more than cool! (Score:1)
Sorry I peeked (Score:5, Funny)
Re: (Score:2)
It is a superposition of useful and unusable data, until you read it.
Then Thors hammer slams down on the laws of nature, and amid lightning and a mad guitar riff, Murphy's Law and wave form collapse combine into always unusable data.
meh (Score:3, Informative)
Re: (Score:2, Funny)
You practically have to shovel cats into the system...
I used to work at a Chinese buffet, so that shouldn't pose a problem.
Re: (Score:2)
You practically have to shovel cats into the system
Schrodinger's? I'd rather use beer [angryflower.com] than cats.
Self destruction? (Score:5, Funny)
Re: (Score:1, Funny)
Re: (Score:1, Funny)
cool! so that means I can download 1 porn movie and have them all?
Re: (Score:2)
Re: (Score:2)
Sort of like using NT Backup.
Re: (Score:2)
No, this is a special kind of quantum memory. The data is still there, it's only the storage mechanism that is annihilated [wikipedia.org], so you'll want to stand well clear of it.
Interesting... (Score:1)
Not quantum addressable (Score:4, Interesting)
Unfortunately from the description it would appear that the memory is not quantum addressable ... that is, you can't use a set of qubits as the address of which qubit to read. For a fully general-purpose quantum computer, we will probably need quantum addressable memory.
Why this could be useful: (Score:5, Interesting)
Now, the computer geeks out there probably heard "qubits" and immediately thought "OooOOOooo... Quantum Computers!". Not so fast. Photonic qubits are generally too quick to decohere (even when stored in memory such as this) and difficult to interact with to be good candidates for quantum computing. It's certainly not impossible, and perhaps even probable in the long-run, but atomic qubits are currently more promising and more widely being looked at for quantum computing. What a photonic quantum memory is immediately useful for is communications. i.e. Quantum cryptography. More specifically, building quantum repeater networks.
If you know a little about computer networks, you know that signals traveling over long distances have to be boosted by repeaters every so often or loss humps your data. Optical networks are exactly the same. After a few hundred kilometers of fiber you have a lot of loss. Unfortunately, unlike classical bits, which can simply be copied, qubits cannot be reliably copied. (Google the "no cloning" theorem if you care.) The work around is a little complex to explain (it's essentially a daisy chain of entanglement swapping), but requires quantum memory to work.
The short of it is, this sort of quantum memory will allow us to build longer distance quantum encryption networks than currently exist. (Quantum crypto is currently being used by some European banks.) At first, this might allow banks in North America to jump on the Quantum bandwagon. It's hideously expensive at the moment, naturally, and probably less economical than running volkswagen's full of hard-drives with one-time-pads on them back and forth, but in principle nothing about this tech is any more expensive than the repeaters the internet currently runs on. Economy of scale should eventually kick in, and these quantum crypto networks will be pretty handy if quantum computers manage to toast public key encryption. (Authentication, of course, is another issue entirely...)
Now, I haven't had a chance to read the Nature paper yet. I've read this groups past papers though, and they really are world leaders in experimental CRIB implementation. Last I checked, they still didn't have adequate efficiency to make their tech useable (must be greater than 50% recall to be practical). Still, CRIB is one of the more promising methods out there.
Re: (Score:3, Interesting)
Unfortunately, unlike classical bits, which can simply be copied, qubits cannot be reliably copied. (Google the "no cloning" theorem if you care.) The work around is a little complex to explain (it's essentially a daisy chain of entanglement swapping), but requires quantum memory to work.
Correct me if I'm wrong here, but wouldn't that same work-around also allow someone to functionally tap into a quantum communications network, thereby invalidating the cryptographic utility of quantum communications?
Re:Why this could be useful: (Score:5, Informative)
Here's the principle on which quantum repeater networks will operate:
Alice (----- Entangled Photon Pair Source -----) Bell State Measurement (------ Entangled Photon Pair Source -----) Bob
What we want is for Alice and Bob to each wind up holding half of an entangled pair of photons. The two sources create two pairs of entangled photons and send the halves in opposite directions. Alice and Bob initially receive photons that have nothing to do with each other. However, when the other halves of Alice and Bob's pairs are annihilated together in the Bell State Measurement in the middle, the entanglement of the annihilated photons is swapped to Alice and Bob's photons such that they wind up being entangled together. The nice thing about this is that Alice and Bob can verify that they share entangled pairs and there's no way for anyone in the middle to fool them, provided Alice and Bob authenticate each other and there are no real-world deficiencies in their apparatus. In essence, Alice and Bob don't have to trust the man in the middle even though he's handling their photons.
To build a quantum repeater network, you just expand this out in a giant daisy chain with many many steps. Quantum memory is necessary for caching photons at each node in the chain so that you can wait for all nodes to be ready before proceeding with the bell state measurements. Caching is necessary because the probability of photons reaching each of the stations in the network simultaneously is no better than the probability of one photon going from end-to-end. i.e. Not bloody likely over long distances.
P.S. Funny aside: The first BB84 system built by Bennett and Brassard (the first quantum crypto system ever built), had some rather noisy pockel cell's controlling measurement bases such that you could tell what basis Alice was measuring in from the sound of the cell. Additionally, Alice and Bob were on the same lab bench, so an eavesdropper in between them would necessarily be inside the room. It was therefore famously joked that the first quantum crypto system was only secure if any potential eavesdropper was stone deaf! This is an example of a side-channel attack that can occur when reality doesn't quite live up to theory, and is the sort of thing people building any kind of crypto system, quantum or otherwise, have to worry about.
Re: (Score:1)
Ahh... Magnetic fields... (Score:1, Funny)
Quantum Clippy (Score:3, Funny)
Can you just imagine it...
"It looks like you want to save your last hour's edits to disk. Well, maybe they're in Quantum RAM, maybe they're not. Do you want me to have a look?"
[YES] [NO] [BOTH]
Re: (Score:2)
Heh,
Reminds me of one of my fav lolcats:
http://icanhascheezburger.com/2007/06/02/im-in-ur-quantum-box/ [icanhascheezburger.com]
Quantum Leap? (Score:2)
"Ziggy says you have to manipulate the magnetic field first!"
I tried this new-fangled Auddie storage mechanism (Score:1)
I would love to tag it but, ... (Score:2)
I would love to tag it but, the new /. code doesn't allow Firefox to interact with the main page. For those of you using IE, I suggest you tag it with "RAQOM".