Slashdot Mirror
MIT, Nanovation to Partner on Photonic Research
Tirisfal writes "The Massachusetts Institute of Technology and Nanovation Technologies Inc. today announced plans to establish a world-class center dedicated to the research and prototyping of photonic technologies, a 21st-century field that will make communications hundreds of times faster. Check out the press release here."
Lehigh University, my alma mater, and Princeton have been researching photonics for quite some time. Lehigh, being the engineering school that it is, has been focusing on the material considerations of actually building photonics systems and developing any new materials that will make it work properly. Princeton has an entire web site devoted to their research in photonics.
The field has been around for quite some time, so there's a lot of information on the web about it. Certainly, MIT's partnership will help push things along, but it is only a very small piece of the research puzzle.
Dragon218 dun said:
Well, unless and until someone comes up with a Theory of Everything that both meshes up with quantum mechanics and Einstein, and also allows FTL travel...I think we might be stuck with c as the speed limit for the observable universe. :P
If memory serves, it HAS been proven that change in one "linked" particle can simultaneously cause change instantaneously in its "twinned" particle, but if memory serves the scientists who discovered this doubt very much it will ever be useful for communication. (For one, it probably won't work across lightyears, and for two, probably the most complex method of communication you could do with it would be Morse-code type on-off communication.)
As far as tachyons go...as someone noted, firstly, assuming they exist at all nobody has any earthly idea on how to create them. (This is, in part, because nobody really knows how to make matter with negative mass--which, at least according to our understanding right now, would require something with negative mass because once one hits c unless you're massless or have negative mass you have infinite mass--wanna birth a universe, anyone? ;) Also, Einstein's formulae for the theory of relativity, at least for mass and time dilation, go REAL funky once the magic barrier of c is passed--I've played about with stuff over the value of c in the equasions for shits and giggles, and you get odd answers like, oh, imaginary time and imaginary mass...maybe you really DO end up spawning a baby universe :).
In fact, if memory serves tachyons (at least the predicted existence of them) are what ended up doing in one of the first superstring theories (which had a solution requiring the universe to hae 26 dimensions); most of the newer superstring theory flavours (including M-theory, which is sort of a "superset of sets of superstring solutions" and factors in an 11-dimensional universe of which there are six solution sets involving 10-dimensional solutions) do not predict tachyons (weird stuff like photinos and quarkinos (supersymmetric "twins" of quarks and photons, only the photinos are the "mass" particles and quarkinos carry force), sure--weird enough stuff is predicted that at least one fellow wrote a novel called "Moonseed" of which the major part of the plot line involves VERY funky subatomic particles predicted in some flavours of superstring theory--but no tachyons) and in fact the fact the 26-dimensional flavour of superstring theory required tachyons is considered to be a fatal flaw in the theory.
Now, if you can find a way to create negative mass, I think we can maybe lick that whole FTL-communication thing. Not to mention find a way to make stable wormholes, invent FTL travel, and find out whether black holes really DO become baby universes if they don't evaporate away due to Hawking radiation :) I'm more than certain you'd win a Nobel Prize at the least, not to mention give science-fiction writers wet dreams for the next millennium :) Till we do, or we find a better theory in which FTL travel works without breaking time or mass, we might be screwed though :P
-Windigo The Feral (NYAR!)
What you describe is one of the standard proposals for FTL communications. It is indeed interesting; however, if you glance at the later parts of the article, you can see that they're talking about more mundane applications of photons:
I hope this was of interest. Purely optical computing is neat, but would be less useful now than it would have been a few years ago. More on this in another message.
The ramification of those circuits is that we could essentially have CPU's 10-50x faster than current chips, with much lower energy consumption as well.
Bear in mind that your computing element size will be limited by the wavelength of the light you're using, though. While waveguide effects might let you push this a bit, remember that the feature size of current chips is already into the "extreme ultravoilet" wavelength range. The wavelength of an electron (at normal energies) is much shorter, making the feature size limits of electrical devices much smaller than those of light-based devices.
This doesn't mean light-based devices are useless; on the contrary, as was pointed out, they tend to dissipate considerably less heat than (conventional) electronic devices. It may also turn out that it is easier to build three-dimensional optical devices than it is to build three-dimensional integrated circuits (both have been done; ICs are just very difficult with current processes). However, I'm skeptical of claims that optical devices will _definitely_ be faster than even the best electrical devices.
Oh yes - before you suggest just using a smaller wavelength of light, that runs into two problems, both due to the fact that your photons wind up having very high energies.
In order to carry a signal, within a given sample period you need to have on the order of n^2 photons, if you are trying to measure n signal levels. This is due to the statistics of measurement errors. The least painful case uses a binary signal, with only two levels (on and off). However, you still need to send between two and four photons per clock to be reasonably sure of detecting "1"s. The problem is that as you reduce feature size, you are both increasing the clock rate and increasing the energy of the photons used. Power dissipation per communications stream goes up as the inverse square of the wavelength. As you will be packing more communications lines on to the chip, your actual power dissipation will be even worse than that. Thus, you rapidly run in to energy limits when reducing the wavelength.
Visible light is about as energetic as you can get without damaging chemical bonds in at least some materials. Many materials can resist higher-energy photons, but many can't (think of plastics that turn yellow in the UV light from the sun and fluorescent light bulbs). When you start moving into hard UV and soft X-rays, the problem rapidly gets worse. Your energy per photon is considerably higher than the energy stored in the chemical bonds in your material. Thus, you will get fairly frequent interactions where chemical bonds are broken or rearranged. Your material will degrade over time, probably quite quickly with the brighness you'd need (see first point).
Like I said, optical computation is a neat idea, and is very useful for many things, but is unlikely to completely replace electrical computation.
Anyone notice that nowhere in the press release did they bother to mention exactly what constitutes "photonics"?
:)
(Flame retardant: yes, I've conducted a search and found it out. I'm merely pointing out that the PR rants for pages and pages about an undefined word. Maybe it's just my innate dislike for the phony buzzword-heavy writing style. Ah well.)
Anyway, this is really interesting. Obviously there are plenty of "but"'s, as other posters have no doubt pointed out by now, but this is a real step forward in research. And it seems rather close to my primary field of interest (self-replicating artificial molecules) - maybe it will give me a chance to actually do some official work on it. Really makes MIT look better and better to me. Hmmmm, I can see it already... "Rafael Kaufmann, Ph. D."... oh yeah.
By the way, I'm still waiting for my nanites!
(P.S.: <spelling-nazi>The correct form is "revolutionise".</spelling-nazi>)
To the editors: your English is as bad as your Perl. Please go back to grade school.
The ramification of those circuits is that we could essentially have CPU's 10-50x faster than current chips, with much lower energy consumption as well.
Anyway, this announcement strikes me as good karma for both Nanovation and MIT. One of the aspects which struck me as being highly positive is that MIT researchers will be free to publish their findings, although Nanovation will have the right to patent the devices.
The question I haven't answered for myself from the press release is whether or not the publishing of those results would allow others to develop technology without breaking the patents... Comments anyone?
...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
This is a truly fascinating area of research. The sheer wierdness of it makes any papers you can find worth reading. It takes a while to let the idea of thinking of light in individual chunks sink in.
Imagine having stateless logic gates that you can send a thousand signals through simultaneously, or having a device as tiny as a pair of glasses that beams a perfect 3D image at higher resolution than your eyes can distinguish directly onto your retinas.
This stuff is da bomb.
Where's the Natalie Portman?
natalie portman
is naked and petrified,
happy is the troll
Note that I'm not the natalie portman troll guy, and I apologize to him if he's offended by this haiku.
I attended a lecture several years ago on photonic research. It was mostly conceptual, but here's how it works, from what I remember. Quantum physicists have found that certain quantum particles, such as photons, can be "linked" to other photons, regardless of physical distance (relative to three dimensions, anyway.) A change in state in one photon results in the same change of state in the other photon, instantaneously, even if they are light-years apart. I believe this ability has to do with higher spatial dimensions, i.e., although they are far apart in three-dimensional perception, if a three dimensional universe is curved in the direction of another dimension, the two photons may be right next to each other. Anyhoo, if you want to set up a communications link, get a photon and monitor its state. The person on the other end can change the state of his photon, thus sending you a message. Pretty nifty stuff....I hope I got it right. Please tell me if I got a part of it wrong, as IANAQP and it's been a while since that lecture.
A quantum channel can also be designed to untappable, (search for quantum crytography to find out more).
Meanwhile quantum computers can bring the same exponiential boost to some mathmatical problems, one of which is factorizing prime numbers (goodbye RSA and https).
Researches are current bizzy developinf the basics for quantum computation and quantum telecommunications. Theory is still decades ahead of practice. E.g. Physists have already designed quantum telephone exchanges in theory, (using entanglement swapping) to distribute entangled pairs of particles, but it will be years before you can buy a quantum router.
What kind of opportunities will this bring to an undergrad at MIT? How sensational would an undergrad have to be to have an important place in the project?
From my experience, this is likely to give a lot of undergrads a spot on the research team, but in terms of "important places" -- highly unlikely. It is very rare that an undergrad obtains a top role in a major research project, let alone a widely-publicized one such as this.
The article says that discoveries made solely by MIT affiliates (or whatever) are owned by MIT. What does this mean in practice? What exactly does MIT do with a patent it posesses?
In practice, you will find that all universities retain IP rights to IP created by their professors -- otherwise why would they be paying them? (No, teaching is far too minor to be the primary source of income). A patent held by a university usually only means that those in the industry who want to actually implement the idea have to pay royalties (otherwise the world's scientific community would just be a public-use R & D team). A university would not, generally, put restrictions on a patent that would prevent the work from being at the basis of future studies.
// zyqqh
I don't believe orbital satellite latency is due to the speed of light. Light travels at ~186,282 miles per second. That means you get 1ms of latency for every 186.282 miles. Orbital satellites are not high enough to have substantial latency due to distance.
While it's true that satellites tend to have slower processors, latency due to the speed of light is very real. Think about your own calculation - for a satellite in Low Earth Orbit, about 300 km up (about 186 miles), you have a 2 ms latency round-trip. And that assumes that the satellite is directly overhead.
In practice, the situation tends to be much worse than this. Viewing at an angle can easily add a factor of two or three here, but that's for LEO; many satellites are instead in geosynchronous orbit, at about 40,000 km. At this altitude, they have an orbital period of 24 hours, which means that you don't have to keep adjusting your satellite dish to track them. However, it also means that you'll be getting about 130 ms delay _each_way_ to the satellite. Round-trip from one point on earth to another, and you start to see why you get latency.
Even fiber over the surface of the earth will give you latency. Per thousand km, you get about 3.3 ms latency each way (ping of 6.7 ms). The farthest point from you is about 20,000 km away. That's almost (but not quite) as bad as geosynchronus orbit.