Slashdot Mirror

You're right- Optical Camouflage doesn't hold a candle to this story. Here's another linky.

2003-02-05 21:06:34 Optical Camouflage a Reality(articles,tech) (rejected)

Modern programming languages can perform some of the safety features that operating systems provide. Many functional languages "can't go wrong". (meaning the compiler can prove a program does not access memory outside of its own space). Some languags like value Microsoft's Vault programming language can make sure an API is used properly. Check out Kernel Mode Linux for an example of how a "safe" scheme program can be speed up by executing in kernel space.

Another factor is the great disparity between actual processing power (often measured in FLOPS etc) and the number of transistors on a chip. For a while, transistors numbers were doubling every 12 months, but computing power was only doubling every 24 months. Why? The need for pipelining and data management meant more and more of the chip had to be dedicated to pre- and post-processing of the actual calculation, along with intelligent caching and the related works of predictive streams.

An alternative approach has been to build specialized hardware to put all those transistors to use, at the expense of turning your general purpose computer into a very special purpose machine. This has been used, sometimes to great effect, in for example N-body calculations (GRAPE 1-6), yielding 50 or more TFlops of performance for the general computer cost of a 500 GFlop machine. It provides yet another example of the misappropriation of Moore's law.

We had guy visit our lab from Japan a few months ago- he's been working on a similar project there to take laser scans of huge Buddah statues and temples. IIRC there were a couple of reasons for doing this- the obvious one being to preserve their cultural heritage. I think one reason was a ban on military research due to WWII, so they have to find ways to apply neat tech which don't involve blowing shit up. (don't quote me on that). I believe they also did a computer reconstruction of a temple which used to be around one of the statues but was destroyed in a tsunami, so you could do a virtual walkthrough of a nonexistent temple, with an accurate virtual statue inside.
He also talked about some of the neat texture mapping they're working on to map the images back onto the laser scanned models.

We used a pinhole as the projector's iris in order to obtain a perfectly focused image. Furthermore, the projected image through the small aperture on the normal surface is too dim to be perceived by human eyes.

However the light coming out from the projector is reflected on the half mirror then on the screen and goes straight back in the eye to form the image, which is about ten or hundred times brighter than the image on the normal surface. Therefore the image only appears on the retroreflective material so that the viewer can observe as if the images projected on the retroreflective material are occluded by the object which exists in front of the screen.

I guess it might be sort of like a dim flashlight hitting a bicycle reflector at night, you only see the reflector lit up.

Until Apple submits SPECCPU benchmark results, it is hard to escape the conclusion that they are not cost effective machines for building scientific computing clusters.

Of course the benchmarks might make that conclusion inescapable.

Mac fans are welcome to do the benchmarking to prove my suspicions incorrect. Or you could translate this page from Japanese. It seems to say that a G4 at 1GHz is about 1/6 the speed of a 2.8GHz P4 on the floating point benchmark.

Yes, they would be rockin fast if they used IBM Power4s. But they don't.

No operating system can have userspace drivers without context switches.

By that I mean no operating system with memory protection for userspace apps. It's possible to give up this benefit (even selectively) for improved userspace performance; see Kernel Mode Linux for a kernel patch that does just that (and which might be applicable to this fellow's situation).

This paper describes ways of doing the lighting underwater.

Once upon a time, high-end boxes came pre-modded. But, then the cold war ended and high-end box manufacturers could no longer afford extravagances like that. Of those machines of the last generation of blinken-lights, the Convex Exemplar SPP-1000 was the most kickass looking computer system ever designed.

Check it out
(As you can see, they were so kick-ass they not only walked on water, they hovered above it!)

Those yellow-green light-bars that go up the front, over the top and down the back are actually fully programmable individual one-inch lights. These boxes came with code to do all kinds of fancy effects with the blinken-lights, such as a ping-pong effect, or racing dots that went at different speeds depending on the load of the machine.

Although the pictures only show the base metallic-purple skins, you could order them with one of 20 different color schemes. The Scripps Research Institute got theirs in a very bright red, as you can see.

Ultimately, Convex got bought by HP and all future designs from that group were exceedingly dull-looking, until finally, just last week, HP laid off a boat-load of the Convex engineers because HP doesn't need technical expertise anymore - they are Microsoft's largest partner!

This device, called TWISTER, was at the Siggraph 2002 - it consists of a drum made of of panels of LEDs that spin around the viewer standing in the middle. It was created by Kenji Tanaka, et al at Tachi Lab, University of Tokyo. I would imagine such a device could even be built to do full 3D, perhaps by using shutter glasses of some sort synched to the scanning of the LEDs. What would also be cool is to add a head tracker that could tell which direction you are looking in, and only activate an "arc" of panels such that the view went beyond your peripheral vision, but didn't wrap around, lessening the load on the computer driving the system (why display what you can't see?)...

Anyhow, this image was taken by Jerry Isdale, a long-time graphics/VR researcher, who attended the show (sadly, I was unable to attend - can't afford it).

The rest of his report is also interesting, showcasing other 3D and VR technologies presented...

Also, the $6000 price tag is not particularly unreasonable for a commercial wearable computer, eg. Xybernaut's stuff isn't much cheaper. Last year I had the job of purchasing a wearable for our lab -- we almost went with the earlier model of CharmIT, but in the end decided that we needed a bit more power and expandability, so we rolled our own. Had the Crusoe version existed then, we quite probably would have chosen it.

Cheers,
-j.

Another computer you may be interested in is Grape-6 which is a 48 Tflop accelerator for gravitational calculations, developed at U. Tokyo for astrophysics. The creator won the Gordon Bell Award a couple years ago.

Ok, look here. There was also something called the Digital Orrery but it was less analogous to a coprocessor.

It might be of interest to geeks that TRON also stands for The Realtime Operating system Nucleus(sucky acronym, I know), a project at the prestegious University of Tokyo that develops software for embedded systems. Their universal character encoding format is also interesting because it isn't Unicode.

It might be of interest to geeks that TRON also stands for The Realtime Operating system Nucleus(sucky acronym, I know), a project at the prestegious University of Tokyo that develops software for embedded systems. Their universal character encoding format is also interesting because it isn't Unicode.

It may be just my opinion (as a former chemist turned physicist), but I think that chemists are rather limited. They're (in general) not very well versed in technological issues and the hard science -- I've found that they're usually an "end-user" of other disciplines' accomplishments.

It's actually rather unlike that they'll miss nuetrino events because of such a change. I've had the oppurtunity to look at individual event plots and raw data, and the Cerenkov light from a single event actually registers in a considerable fraction of the tank. IIRC, typically 5-30% of detectors see each event.

This isn't entirely true. It depends a lot on the type of event. The pictures you probably saw were either of an atmospheric neutrino event, a cosmic ray background event, or one of the K2K events. In all of these cases, the particle in the detector will have somewhere between 100 and 5000 MeV of energy (and in some cases more).

As a particle travels through matter, it loses energy as it goes. The more energy it has to start with, the longer it will go before it stops. A general rule of thumb is that a muon travelling through water loses 2MeV for every centimeter travelled. So a 100MeV muon produced by a neutrino interaction would travel for 50cm. The higher the energy, the longer the track, the longer the track, the more light produced, the more light produced, the easier it is to see.

Very high energy cosmic ray muons will produce so much light that every tube in the detector will register a hit. A typical high energy event picture is here. Would you see the pattern with half as many pixels? Of course.

The problem is that "solar neutrinos", neutrinos which come from the sun, typically have much lower energies. (Only 1-10 MeV) So low, that even before the accident, SK would miss most of them (anything below 5MeV) because they just didn't produce enough light to be distinguishable from random noise in the dector or the decay of stray radon particles. If you look at pictures of these events, normally you can't see anything by eye. There's just a few photons (5-10) which are recorded which can only be identified as a real event by their timing because you can triangulate back to a single point based on their arrival time at the PMTs. Solar neutrino physicists rarely post event display photos because there's so little to see in them. Even then, it's hard to distinguish solar neutrino events from noise. In fact, it's not possible to identify an individual event as a solar neutrino event and not a radon event that looks like a solar neutrino event. It can only be done statistically. (Radon events don't point in any particular direction, while solar events all come from the sun, so you can compare the number of events coming from the sun and the number of events apparently coming from other directions and do a background subtraction.)

It's these types of events which will be hurt most by the loss of the extra tubes.

Just what is the SuperK?

A detector for neutrinos. Have a look at their web page.

I attended a talk last night by one of the scientists from the Sudbury neutrino detector. One of their Big Issues at the moment is figuring out why all the best neutrino detectors only pick up a fraction of the neutrinos predicted by all the best theories on the innards of stars.

...laura

Actually, the neutrino detectors weren't built to study the "Solar Neutrino Problem" either, although that has been a pleasant side effect.

check this description of the photomultiplier tubes

Actually, it looks like the alternate site has both H6 and H7 information here:

H6 and H7 information page

~c

Since the H7 page is already Slashdotted, perhaps some information on its direct ancestor, the H6 would shed some light.

Humanoid Robot H6 page

~chris

Since the H7 page is already Slashdotted, perhaps some information on its direct ancestor, the H6 would shed some light.

Humanoid Robot H6 page

~chris

No, wrong. A lot of gamers, like me, use Q3 benchmarks to roughyl measure how well a piece of hardware performs on games in general, not just on Q3.

I hope you meant that you use Q3 benchmarks as well as others, because if you don't plan on running Q3 you won't pay attention solely to Q3 benchmarks, and you definitly won't "use Q3 benchmarks to measure how a piece of hardware performs in general." Remember how Unreal ran really, really well on 3dfx cards, and on that one specific "benchmark" it would outperform competition that would spank it silly in every other performance assesment? Well, people didn't (or, people with any sense didn't) buy a Voodoo3/5 if their game of choice was Q3, because it obviously didn't perform as well then. True, you can't judge a video card's long-term (what, 6 months? :P) value, especially off of one benchmark. It seems that if one's buying decisions are so upset by a single benchmark, then one probably plans on extensively using the hardware for that one function (e.g. the voodoos for UT, or one of these cards for Q3). That, or one is just really gullible, and as I said before in this case should be greatly confused by all of the largely differing numbers that they put in big print on the boxes.

I agree that it is a shady practice, I just don't think that one single and moderately specalized benchmark should influence buying decisions so much. Thats like reading this article and deciding that because its so adept at performing one function that its great for anything. Perhaps its not exactly the same, but IMHO it is just as silly.

Seriously. I have colleagues that work on this type of thing:

"Sound Symbolism in Conversational Grunts in English"
"The Challenge of Non-lexical Speech Sounds"
"Issues in the Transcription of English Conversational Grunts"

http://www.sanpo.t.u-tokyo.ac.jp/~nigel/publicatio ns.html

All of the various national character sets and vendor character sets are subsets of Unicode, so if you want to write something today you have little practical alternative.

Not true. TRON, for example, has defined its own character set that includes characters not available in Unicode (130,000 in total, currently). See here and here for more info.

All of the various national character sets and vendor character sets are subsets of Unicode, so if you want to write something today you have little practical alternative.

Not true. TRON, for example, has defined its own character set that includes characters not available in Unicode (130,000 in total, currently). See here and here for more info.

Note that SSH2 is NOT compatible with SSH1. If you'd like further info: Click here. But it can be made compatible (see caveats).

Admins, read this PDF document.

Brief product spec page from Matsui

Fuller details from U of Tokyo. Huge amounts of technical detail, but a January 1995 article (ie before the sea trials). Should answer most of the calls for "but how does it work?".

Paper describing and appraising the sea trials. Less detail on the CCDE, but a better overview (and written after they've tested the thing for real!).

Brief product spec page from Matsui

Fuller details from U of Tokyo. Huge amounts of technical detail, but a January 1995 article (ie before the sea trials). Should answer most of the calls for "but how does it work?".

Paper describing and appraising the sea trials. Less detail on the CCDE, but a better overview (and written after they've tested the thing for real!).

little robot? 8.2m long and 4 ton weight ain't little!

Though I grant you it does consume less than a car. This article describes the power output as 5kW. This compares to 56kW on my Rover 214 (weight approx 1 tonne)

TFlops not GFlops. Read the link.

Well, you got one thing right...

However, this isn't the case in Asia. The common example is Animation and Japan. For some reason, they see Animation as a very important part of their culture. People hold parades to look like their favorite anime characters in Japan! Anime is for all ages, as you can see by the wide selection of everything from the super sappy to the hard core violent and sexual scenes one can only see in "adult" anime.

Sorry to burst your bubble, but in Japan, anime otaku are every bit as marginal and despised a phenomenon as in the US. Kiddie comics are read mostly by young kids (surprise surprise), adult comics are read by the same demographic that reads Playboy in the US. Yes, there is a parade of people who dress up as anime characters (and anything else they want to) every Sunday in Tokyo -- half the people there are giggling tourists and amateur photographers like myself who like watching the freakshow.

Try another generalization instead: online gaming is fun. I have no doubt that in a few years, well over 5% of the entire industrialized world will be playing them in one form or another.

Cheers,
-j. (in Tokyo)

Along a similar line, you might want to take a look at Takeo Igarashi's 'Teddy' applet, a real-time cartoon-rendered modeling utility. It's still a research prototype, but it has a lot of potential!

P.S. Sorry about the atrocious spelling of my last post... Too much coffee!

I've seen a cool little toy/tool from the folks at Carnegie Mellon. It's called Alice, and it's a pretty interesting 3d graphics package for the web. Oddly enough I learned about this from a multimedia class. I'm a little more taken with Teddy2 and major props to Takeo Igarashi at the University of Tokyo. Teddy2 is a spiffy little tool for building objects. And as long as I'm plugging 3d graphics tools I like to putter around with, check out Texture Weapons. It is certainly worth checking out some of the demos. This isn't just another way to make 3d widgits...ok well maybe it is, but they are really cool widgits, and sometimes more.

Just have a look at this image from the construction of the Superkamiokande Neutrino Detector. The photomultiplier tubes ("mushrooms") used there are very much similar to those used for the AMANDA detector. You can see two of the AMANDA sensors here, together with the glass pressure globes they're put in before deployment.

I know this - have been working for the AMANDA group once, when we were calibrating the first PMT's for AMANDA back in 1995. It's done at Desy Zeuthen near Berlin. And we were using Linux boxes in the lab for data aquisition purposes ;-)

The nifty thing about AMANDA aren't the PMT tubes but the pressure globes they are put in (1500m of solid ice do exert some force ...). I've got one of the predecessors (used for the BAIKAL experiment) at home, it's cool telling people at a party that the salad bowl has once been at 1500m depth in Lake Baikal.

By the way, did someone notice that the AMANDA logo is a Penguin ?

How much not very often?

Something like one in ten to the thirty-odd neutrinos will interact.

So, that's enough to see a few of if you if you build a really big detector - Super-Kamiokande in Japan or MINOS in Minnnesota being the experiments I work on.

But, even if you are a fifty-kiloton person and a few neutrinos interact in your chest, that's not enough interactions to pose a danger to you.

Note that this is why these experiments are all deep underground - the cosmic rays bombarding the surface of the earth produce many times more interactions than do the neutrinos. Going underground shields you from a lot of the CR's, so you can actually see the neutrinos.

I wonder what else they'll be able to find out about neutrinos with this detector. I remember the Super-Kamiokande Detector at the University of Tokyo Institute for Cosmic Ray Research. They detected the first neutrino oscillations with it back in 1998 and did an experiment a couple of years ago with an atrificial neutrino beam that further supports the hypothesis that neutrinos oscillate and therefore possess a small amount of mass. I guess this Canadian detector ought to support the theory further.

It will probably be many years before the SNO can produce any kind of useful experimental results, though. Neutrino interactions are of extremely low probability...

Neutrinos are the least studied elementary particles because of they interact very, very rarely. It's no joke that they can "pass through matter like smoke", as the story said. The typical neutrino can pass through several light-years of lead without interacting once. The only reason they can be detected at all is that a tremendous number of them pass through the Earth every second. I forget the exact number, but it's something like trillions per square meter per second. Even so, a decector the size of SNO will only see a few hundred events per second. On the other hand, this is also why neutrino experiments like SNO or Super-K are so exciting for astrophysicists. The light that we see from the sun has all come from the surface, photons produced in the core can't make it through the sun to get to the earth. Neutrinos produced in the core can easily penetrate the whole of the sun and reach the earth. As a result, a very good neutrino telescope can look directly into the core of the sun. There are a berzerk number of other reasons to be excited about neutrino experiments, see the Particle Adventure for more.

Oh, and if you thought the SNO picture was cool, check out some of the photos on the Super-K, they've pretty much won the best-looking physics experiment ever contest.

Neutrinos are the least studied elementary particles because of they interact very, very rarely. It's no joke that they can "pass through matter like smoke", as the story said. The typical neutrino can pass through several light-years of lead without interacting once. The only reason they can be detected at all is that a tremendous number of them pass through the Earth every second. I forget the exact number, but it's something like trillions per square meter per second. Even so, a decector the size of SNO will only see a few hundred events per second. On the other hand, this is also why neutrino experiments like SNO or Super-K are so exciting for astrophysicists. The light that we see from the sun has all come from the surface, photons produced in the core can't make it through the sun to get to the earth. Neutrinos produced in the core can easily penetrate the whole of the sun and reach the earth. As a result, a very good neutrino telescope can look directly into the core of the sun. There are a berzerk number of other reasons to be excited about neutrino experiments, see the Particle Adventure for more.

Oh, and if you thought the SNO picture was cool, check out some of the photos on the Super-K, they've pretty much won the best-looking physics experiment ever contest.

According to the Cryptography FAQ , differential cryptanalysis was first discovered by the NSA, then rediscovered by Shamir. Quoting from the FAQ:

IBM has classified the notes containing the selection criteria at the request of the NSA.... `The NSA told us we had inadvertently reinvented some of the deep secrets it uses to make its own algorithms,' explains Tuchman.

search 10 million digits of pi
download up to 51 billion digits of pi
there's only 4.2 billion digits available for public download, but up to 51 billion can be downloaded by request (if you get them to email 51 billion digits to you, cc me ;>)

Tiny! On this ftp server here they claim to have 4,200,000,000!! But only 200,000,000 for public download. BTW, they're at 6,442,450,000 PI digits now...

What I don't get is: We know it is an irrational number so this is will go on forever. Is there any practical use to knowing PI with such precision or is it just a pissing contest among mathematicians?

http://www.is.s.u-t okyo.ac.jp/tech-reports/TR94-02-letter.ps.gz

There was a recent discussion on NetBSD's mailing list about implementing this high-performance thread architecture for NetBSD. You can read about it below, under the ``upcalls'' thread.

http://mail-index.netbsd.org/tech-k ern/1999/12/

Tru64, Irix, and Solaris 2.6 all use scheduler activations. Linux does not. And IBM is not suggesting it. The Masuda Lab implementaiton is slightly better than scheduler activations and far more elegant. I do not think commercial OS's have adopted the Masuda & Inohara architecture.

Naive many-to-many without scheduler activations is simply not an efficient enough threading model. Note that all this research I have cited is over five years old, and yet Linus still wants kernel threads like NT uses (does he?). In my opinion, it is very important to keep current with a diverse array of research, because I've found that when I only pay attention to white papers and press releases from Intel, Sun, IBM, Ars Technica, whatever--I miss a lot of highly relevant work that later turns out to be every bit as essential as the usual wise softspoken few knew it was going to be from the beginning.

If you liked the IBM paper, I highly reccommend reading the ones I've referred you to. You might look at some academic papers on scheduler activations or threads in general from your neighborhood university library--they're surprisingly accessible even to people like me who aren't research studs.

• The Top Four really are superb, but the people who say "choose based on field" are also right. If any of CMU, MIT, Berkeley, or Stanford cover your field and you can get in, move heaven and earth to go to one of them.
• If the Top Four don't take you, try for a school that has several people working in the field you're interested in. Check out their recent publications and see if they're cool; if possible talk to them to find out whether you like them or they're assholes. Talk to their current grad students. Try for a school that has a reputation (easy measure: they consistently get papers into the top conference in the field). Avoid a school that has only one prof in your field: if you hate him/her, or if the research s/he's doing two years from now isn't fun, you're screwed.
• An experienced, understanding advisor at your current school is invaluable.
• Know thyself. Why do you want a graduate degree? The person who said "go straight to work" was partly right. An MS is quick and easy to get, and it will pay off in industry (lots of people are impressed). A Ph.D. is a very specialized degree, and not tremendously useful unless you want to go into research or academia. (Exception: in some consulting positions the prestige factor helps.)
• It is very hard for most people to return to school after time outside. I'm not talking about forgetting how to study, I'm talking about having a life, kids, and car payments. Most people never try, and of those who try, most never finish.
• A Master's usually takes two years. I did it in one under abnormal circumstances; I know a guy who took eleven (full-time!). A Ph.D. in CS usually takes from 4 to 7 years depending on the school and advisor. I know of a guy who did it in 3 (and regrets going so fast) and one who took 13.
• When you're looking for a job after getting a Ph.D., many things matter. Some of the important ones are the quality of your dissertation, the number of publications you have, the name of your school, the names of your references, and the content of your reference letters. All of those are affected to some degree by your choice of school. Employers also care about you, of course (make sure your interview is great!), but the above items are harder to fix late in the game.
• Early in your graduate career, it can be good to do internships at industrial research labs. This approach gives you good dissertation ideas, and also gives you a wider base to draw on when it's time to get reference letters.

Enough random advice. Here are some books and URLs:

• Tomorrow's Professor: Preparing for Academic Careers in Science and Engineering, by Richard M. Reis. This book contains absolutely essential advice, starting with how to pick a graduate school and ending with advice on surviving your first year as a professor. If you are thinking about grad school, or in it, this book is a MUST! I only wish it had been written before the last year of my doctorate. Even so, it made a huge difference in my eventual success. I owe my current job to many people, but Dr. Reis is unquestionably one of them.
• A Ph.D. is Not Enough: A Guide to Survival in Science, by Peter J. Feibelman. This book has some very realistic, sometimes cynical advice for prospective scientists.
• How to be a Professor: Some Good Books is a Web site devoted to helping people adjust to academia. For a prospective grad student, it can also serve as an introduction to what to expect.
• Rank PhD Programs in Computer Science from CRA gives graduate-program rankings, though they're somewhat dated. (Take all rankings with a grain of salt, though.) The Computing Research Assocation is a useful resource in general (check out their salary survey).
• The US News rankings are also useful.

As usual, I've run on and on, so I'll close with a wish for your success and one last thought: grad school was the most fun thing I ever did!