Slashdot Mirror
Molecular Photography
med dev writes "An article at New Scientist discusses the latest in quantum computing - 1000 bits stored in the electron spins of a single polymer molecule. Add in a recent release of the how-to for the complete quantum computer, qubits that work, and it may not be much longer before Google is running on a server the size of a sugar cube."
it may not be much longer before Google is running on a server the size of a sugar cube
"Hey Johnny, where did the new $100,000 server go?"
"I don't know... I had it right here on the table!"
"Oh shit! I put it in my coffee! That's why it tasted kind of funny."
---
Hello, Slashdot user. My name is Dr. Sbaitso. I am here to help you.
Does anyone know if Synchrotrons, like the one in Saskatoon, SK, Canada play a part in researching molecular computers? The article mentions a magnetic imaging device. Is that like a synchrotron?
Saskboy's blog is good. 9 out of 10 dentists agree.
So the scientists have succeeded in encoding a tiny black and white picture on a polymer molecule. Hooray! Another tiny step for science, but a giant leap for mankind. However, realitically, I don't think Google will be running on a sugar-cube sized memory bank any day now. The money to move that kind of infrastructure onto a quantum computer is unthinkable.
So, a wonderful step forward....but there are still many many steps left.
Sincerely, your local cynic
"To make apple pie from scratch, you must first create the universe." -Carl Sagan
nuclear magnetic resonance (NMR) instrument.
I've done NMR, it takes ages. Preparing the sample takes about 30 minutes. Running the NMR takes between 1 and 20 minutes depending on what you're measuring. Analysing the results depends on how good you are.
I can't see google using this any time soon.
If they could just fit 24 more on there, it would be a much easier number to work with...
That's not a soda... it's a caffeine delivery device!
Sure, biochemists might need the massively paralell processing power to do molecular folding analysis, but regular joe bloes will, IMHO, be very comfortable with quad 2GHz HT Pentium 4s... for a decade at least.
I feel there will be a rift like there was in the old days when mainframe systems were few and expensive, and the rest were smaller systems.
Frankly, Quantum doesn't titillate me as much as a nice new nVidida chip at this point.
The other thing is that massively powerfull paralel processing isn't always a Good Thing. It's just A Thing. Take for example early Pentium Pros which had 16 stage pipelines. Nice in concept, but unless you use it properly, it's not really usefull. Many problems aren't massively parallel... The brain for example, is massively parallel, but not in the sense that many mean: all of your brain isn't adding two million 4 bit integers at the same time. It's doing millions of different tasks...
Sunday night... must sleep... must shadap.
"I think there is a world market for maybe five computers." --Thomas Watson, chairman of IBM, 1943
There is no reason anyone would want a computer in their home." --Ken Olson, president, chairman and founder of Digital Equipment Corp., 1977
Just because you don't see the possibilities inherent in something does not mean that the thing has no value or is not relevant.
Besides, with the way things are moving, I can imagine the possibility of a computer that needs no clumsy interface cables, no removable media, and such... We're moving closer to being able to make systems that truly have no moving parts.
After all, there was a time when computers were built around the size and heat of vacuum tubes. Someday, probably not all that long, the interface mechanisms, storage devices and display systems we use today will be as quaint as a vacuum-tube driven computer programmed by hard-wiring it seems to us now.
Apparently quantum computers "borrow" exponential scratch space from some Hilbert space during computation. As far as I know, there's no one living in Hilbert spaces that it could upset, but I could be wrong. So at worst it's a victimless crime, like punching someone in the dark.
Moreover, the peculiarities that make quantum computing interesting (e.g. the ability to factorize in polynomial time) also make it completely inappropriate for mundane tasks. So please stop the "google in a cube" shit.
The Raven
regular joe bloes will, IMHO, be very comfortable with quad 2GHz HT Pentium 4s... for a decade at least
The entire history of consumer electronics belies this statement. People demonstrably don't by a system because it's sufficient for their needs, they buy it because it's the most powerful one available.
If they make it, they'll buy it. Whether or not there's a good reason for them to need that kind of power. All that will be required is for the manufacturers be able to make it affordable enough or sell it well enough to make people see it as affordable enough.
After all, my cell phone (and maybe my calculator) has more raw memory and computing power than the computer used by the men who flew to the Moon.
To everyone who has so far commented: so what?
My mother was born in 1947. The transistor was also invented in 1947, by Shockley. 55 years later, I got her a new computer for Christmas.
What will I see when I turn 55? I can't wait to find out.
Will quantum computing make using database table indexes obsolete? ie. will the time saved by using an index be small enough that it's not worth the effort to create/maintain one (for most uses)?
Sounds like "what-if" analysis will be taken to a new extreme, big time.
so we can store information on a molecule, but how big was the machine that created the spins? And how long did it take to process the 1's and 0's on the molecule?
Sure, we could store information on molecules, but the speed and the size of the machines involved would put us back to working with punch cards...
What needs to be done simultaneously is to improve the method in which we induce and read the spin in molecules, or those sugar cube sized computers will just be expensive and slow RAM inside a computer the size of a room...
That's the dumbest thing I've ever heard. If I can fit four images on a single floppy disc, what if one of those images was of the floppy drive itself... Woah
Send lawyers, guns, and money!
I can argue of a future where the emphasis is on the Mobo that can house up to 32 CPUs. and the new AMD Thunderfolts that are so small you can actually fit 32 of them in a mini ATX case... With very low power consumption, and low heat emissions. And big hdd capacity, and loads of RAM, and high bandwidth, and this and that...
People will have many gimicks to market before they run out of ideas and turn back to the speed issue of a CPU.
Once again, IMHO.
I have to wonder what type of redundancy and error correction will have to be built into quantum computing.
Lots and lots. In 1995 Peter Shor (the factoring guy) and Robert Calderbank devised that possiblethe first error correcting code for quantum computers. Many others have been designed, including proposals for some that operate as a natural consequence of the system being used. Here is a good survey of the field.
It has been shown that if the error rate is below a certain threshold (currently estimated to be one error per 103 operations for optimists, and one per 106 per pessimists) then efficient error corrected quantum computation is possible. The pessimistic estimate is well above what is currently possible experimentally in quantum systems but the problem seems to be an engineering one, not a fundamental one. It should eventually be possible with clever implementations of qubits, shielding and cooling to near absolute zero.
:wq
You are incorrect. Classical computers can search an indexed database in log(n) time. Grover's algorithm allows quantum searches to be much faster, perhaps even in constant time. Search engines could benefit immensely from quantum computing.
Lots of information can be found on Lov Grover's quantum search algorithm. Do a search for it on Google. Dr. Dobb's even analyzed the quantum source code for the algorithm. Pretty cool stuff.
Been there, done that; reading core was destructive, so you had to copy back what you just read. Admittedly, it means that there's no read-only version.
Such variances are common and expected in quantum computing; hence the field of Quantum Error Correction. (Google for more...)
The researchers fired an electromagnetic pulse containing 1024 different radio frequencies close to 400 megahertz at the molecule
Gee...if it takes that many requencies to read 1024 bits, imagine how many you'd need to access the memory space of the average desktop PC. You'd need the whole damn electromagnetic spectrum! I wonder if the FCC will grant them a license for that?
-ted
Blah blah blah
...and other neobabble.
The quantum states of phosphorus atoms are particularly long-lived,
The article tells us basically nothing real, except the names of a few people and that they're working on something called "quantum" computing.
So here's how it should work (off the top of my head):
An atom or molecule (a collection of particles) has a set of wave-equation solutions. Each of solutions corresponds to a single point in a lattice, whose coordinates are the quantum numbers; or a single value of an n-tuple whose indices are the quantum numbers; or a single member of a set of n-tuples each of which is identified by a unique combination of quantum numbers...however you want to express it. These quantum numbers are inserted into the wave equation and out pops a solution--a wave-function--that does not diverge or otherwise go kaput.
If the atom, molecule, collection of particles, etc., is in one state (one combination of quantum numbers; one wavefunction), it's just a matter of applying energy in the right way to push it into another state. The quantum numbers move to a new point in the lattice, you change the n-tuple indices, whatever. You really cause the wavefunction to change, and the spatial arrangement available to the particles moving in the system changes. A spherical shell becomes a dumb-bell shape (not really, but it's a simpler visual than what really happens, so go with it).
Now you have a binary memory system. Most systems have way more than two states, but only a few will be stable (metastable, actually) enough to be useful for computation. But trinary, quaternary, etc. are certainly not out of the question; though the question is a lot easier if you can still use all this software expertise that has binary math running through its veins.
Quantum calculations are a lot harder to grok than quantum memory. Something has to work so that the state of the memory actuates another part of the system to undergo a change on a quantum level from one stable state (n-tuple value/wavefunction) to another.
The Heisenberg Uncertainty Principle would get involved, so the family of states you use would have to be pretty special to keep the particles in knowable states. I think that's what the reporter was really getting at when talking about the phosphorus thing.
if you want some details on how it's done, read my other post.
Um... the other posts by username "Anonymous Coward" all involve a website called goatse, whatever that is.
We can neither love nor pity nor forgive. If you make a slip in handling us you die!
Does anyone know if Synchrotrons, like the one in Saskatoon, SK, Canada play a part in researching molecular computers?
No, not at all.
The article mentions a magnetic imaging device.
Is that like a synchrotron?
No, not at all.
Syncrotrons produce gamma/X-rays. Expose a polymer to some of those, and it won't stay a polymer for long..
NMR instruments (and MRI devices) use radio waves. Much longer wavelength, much lower energy.
The only similarity I can think of is that both use big magnetic fields, but for different reasons.
(syncrotrons use them to accelerate particles, NMR machines use them to split the spin energy levels)
Let me explain more clearly, because it seems that some of the moderators didn't understand my comment.
Think about a photon, which has a linear polarization: up-down, left-right, slantwise, or at whatever angle you want. You can in principle put in an arbitrary amount of information in setting the polarization angle of a photon. You could divide a circle into as many parts as you want, and set the polarization to an angle corresponding to the value you want to send. This is like how they pack 1024 bits into a 19 nuclei molecule.
Now, the problem is reading the data back out. If you have only one photon in a particular polarization state, you can't determine that state with any accuracy. You can in fact only get one bit of data out of that photon. You can pass it through a polarizer and either it makes it, or it does not. This gives you information about the polarization state but it destroys that state in the process. You can put lots of information into a single photon, but you can't read it back out.
Now let's imagine that we have lots of photons, in a laser beam for example. We can set them all to the same polarization state. Now we can read the polarization quite exactly, by using large numbers of photons and turning our polarizing detector until we get a peak in the output.
Even though all the photons are in the same state (like in the NMR molecule experiment), it is because there are large numbers of them that we can read the state back out accurately. We would NOT be able to read back the data from a single photon, and in the same way we would NOT be able to read back the data from a single molecule.
Hopefully that explains my comment above. A qubit, whether photon polarization or nuclear spin, holds only a limited amount of information, and you can't read more out than it holds. There's no way you can get 1024 bits into 19 nuclei, and no one should try to "spin" the results of this experiment that way.