Slashdot Mirror
SETI Institute Is Looking For a Few Good Algorithms
blackbearnh writes "For years, people have been using SETI@Home to help search for signs of extraterrestrial life in radio telescope data. But Jill Tarter, director of the Center for SETI Research at the SETI Institute, wants to take things to the next level. Whereas SETI@Home basically used people's computers as part of a giant distributed network to run a fixed set of filters written by SETI researchers, Tarter thinks someone out there may have even better search algorithms that could be applied. She's teamed with a startup called Cloudant to make large volumes of raw data from the new Allen telescope available, and free Amazon EC2 processing time to crunch the data. According to Tarter: 'SETI@Home came on the scene a decade ago, and it was brilliant and revolutionary. It put distributed computing on the map with such a sexy application. But in the end, it's been service computing. You could execute the SETI searches that were made available to you, but you couldn't make them any better or change them. We'd like to take the next step and invite all of the smart people in the world who don't work for Berkeley or for the SETI Institute to use the new Allen Telescope. To look for signals that nobody's been able to look for before because we haven't had our own telescope; because we haven't had the computing power.'"
All you need is to hire the programmers who did the GW stuff. Tell then what results you want, and they'll supply the code and data to fit your needs. Well, they'll probably lose it for you before anyone else can review it.
Who would win this election: Andrew Weiner vs Andrew Weiner's weiner.
That's what she said!
Maybe it'll all be sorted out retrospectively following the singularity. There's a big crossover between AI and data mining/pattern recognition after all.
Might make a good plot for a novel... ;)
"I bless every day that I continue to live, for every day is pure profit."
print "WOW!"
Any sufficiently advanced technology is indistinguishable from noise. I remembered hearing this in school so I searched and found this paper.
As I understand SETI has always been searching for narrowband signals in the past. But our technology is moving toward spread spectrum signals for more efficient use of bandwidth, making our transmissions appear more like noise to anyone who doesn't know the encoding scheme. Aliens could be doing/have done the same. So good luck, scientists!
If they're already looking at the Shannon entropy of the Fourier Transform over a Markov chain of prime numbered frequencies then I've shot my wad.
Just wait and see what kinds of interesting "patterns" hordes of uninformed basement "researchers" can come up with given this huge dataset.
I predict hilarity.
I don't know about the specifics of analyzing radio waves to search for intelligible communication, but I can suggest one thing. The Kepler telescope set out to hunt for earth-like planets that may contain life. The SETI project as far as I can tell is scanning every part of the sky without discrimination. I suggest that Kepler and SETI team up so that when Kepler finds a planet that is capable of supporting intelligent life, SETI will point its telescopes to it and then run whatever algorithms you have to analyze the radio signals if there are any from that planet. This current method of scanning arbitrary locations in the sky sounds like it could be wasted effort when you don't know what it is you are pointing the radio telescopes at.
Ok, I don't know anything about this stuff but basically SETI is a big radio telescope array right? And we get data from all these stars using other radio telescopes right? So at some point we take the data of stuff we know (stars, whatever else) and take it out of the SETI picture right?
That leaves us with a lot of radio noise leftovers, but is that what SEIT is picking through? Do they target certain stars for a while hoping that a body in orbit happens to have been sending out signals?
Or is SETI listening to "everything" hoping to find anything?
I'd like to know!
crazy dynamite monkey
SETI is pretty much fumbling in the dark on where to look. The really interesting parts going on now in my opinion is the search for exoplanets, with better equipment we'll soon start having real targets to listen to. It's entirely possible that we've missed it simply because there's been no antenna pointing in the right direction long enough. After all, life as we know it takes millions of years to develop - it's not like they're going to ping us every five seconds "Is there life now?", at least not after the initial attempts at contact. The last could have been thousands of years ago, and you can't assume they're as interested in finding us as we are in finding them. They might know there's life on other planets already, and treat the discovery of earth with the same shrug we discover a new species in the amazons.
One thing I've wondered, if we assumed a mirror image of ourselves on the other side could we do it? Like, how huge an antenna, how narrow a beam, how high power would we need to make a point to point interstellar connection using known technology? Could we do it at all?
Live today, because you never know what tomorrow brings
So this is the beginning of the contest for better algorithms (ignoring how ee measure if they are better, since no one's found the data they are looking for in the first place), and then of course a new round of analyzing the data again.
- Issue call for better algorithms.
- Reprocess the data.
- Find nothing.
- Must be the algorithm.
- Repeat.
SETI will never die. It will just question its assumptions.
Feh.
deleting the extra space after periods so i can stay relevant, yeah.
I don't doubt that there is life elsewhere, there just has to be, right.
I also don't doubt that space travel is possible, though this may well be.
What I am beginning to doubt is if intelligent life really survives long enough. Seriously, the more intelligent we get, the more damage we do, and it seems that extinction is an inevitable consequence of any combination of freedom and destructive power, aka technology, over any long period of time.
One of these days someone will trip on a cord or spill their coffee, and we are all going to die!!!
1. Pizza ............. .............
2. Beer
3. Ppppaaaarrrrttttttaaaaayyyyy!!!
Yours In Novosibirsk,
K. Trout
if data contains alien_signal then
alert("Found Alien!!! Prepare for destruction!!!)
end if
The real Sig captains the Northwestern. This one captains
Some civilizations will, for a short period of time, use detectable radio as a means of communication, but I suspect that there are very few of these at the same point in their technological development as we are. It would make more sense to look for objects that are almost certainly artifacts. Geometrically placed stars moving in the same direction at the same speed. The infrared signature of Dyson spheres. Anything that's too geometrically perfect to be natural. Anything that's accelerating//decelerating relative to it's surroundings. In our own solar system, what would an asteroid mine tailing look like, and does anything look like that?
Please do not read this sig. Thank you.
"SETI Institute Is Looking For a Few Good Algorithms"
You misspelled "Aliens"
SETI as designed is incapable of even detecting and decoding something akin to the Arecibo message, so I'm always puzzled at how they think they're actually going to know when they have hot data for real. I applaud the effort but I've always felt it was more of a feel-good activity for people to join in on. Hmm....
You haven't heard? GW has been replaced by some Kenyan muslim.
But those of us who were around when the programme kicked off in 1999 got a bit peeved when it was found that we were processing the same workunits over and over again, and then there was the "problem" when it was announced that the clients weren't optimised for the scanning algorithms. A lot of people packed their bags and left the SETI@home project. Myself? I got a little annoyed when it was announced that new ideas for searching through the data were announced...and we'd have to start all over again. Lets hope that whatever is organised next time is a bit better, well, organised :-)
My web domain.
If aliens want to communicate with us we're just going to have to standardize a protocol.
Under the banner Easier Said Than Done:
SETI needs to launch a satellite into deep space that can record broad band emissions from EARTH!!! Then SETI will have a baseline data set on which to build algorithms. The idea of looking for a signal without knowing what you are looking for is just bad science. An experiment MUST be base-lined and to do that in this case, SETI needs some truth data. They can produce all of the synthetic data they like and amazingly enough their algorithms work every time. Clearly, it has been demonstrated that all of their assumptions about the makeup of the signal are incorrect (Or there is not signal to detect in the first place). Without truth data, these two possibilities can never be resolved.
I'd try to rewrite Brown's Dieharder Test Suite in CUDA/OpenCL:
http://www.phy.duke.edu/~rgb/General/dieharder.php
If the data passes the suite, it's probably not alien causative.
I tried decoding some of those signals. The only thing I got was this really big Mersenne prime.
I'm not going to try to use the right terminology, simply because I'm not an astrophysicist, but I wonder if attempts have been made to analyze the signal across different timeframes, as opposed to one continuous stream of radio-waves. Plato has done it in his works, so why couldn't the aliens?
From what I remember of SETI@Home many many years ago back in the days of the dec alpha they would keep running the same codes with different permentations for doppler shift that might then be recognized as a pattern once corrected for.
I'm way in over my head when it comes to signal processing but the gut feeling I've always had was that they were throwing cpu cycles at the problem to solve a problem that could more quickly be solved with a more sophisticated algorithm.
Sorry to rant a bit but why the hell would a civilization sent out a radio signal from their planet? Look at us, all we do is "listen" but we don't build any transmitters capable of transmitting a signal across a thousand light years. Transmit first, listen second.
But wait, what if they are capable of interstellar travel, they could send an invasion fleet... we are paranoid, why wouldn't another species be as well?
So what to do? One you don't send a signal from your planet. Two you design your signal to be easily found; found by another species not even listening randomly for a signal. Answer: you build a spacecraft and send it someplace interesting. Some place an astronomer would find interesting and you either transmit from there or somehow you modulate the natural phenomena to carry a signal for you.
You would have three types of signals. The first signal would be to get your attention and make you wait and listen for the second signal which would contain enough information (location, frequency, polarity, whatever) to direct you to a third signal that would actually contain an entire database worth of information.
For an example of a type one signal I don't think it's too far outside possibilities that in the future we might discover a way to generate gravity waves and while they might not travel very far they might be strong enough to influence a star, white dwarf, neutron star, nebula. Imagine one day an astronomer looks at a nebula only to think.. hmmm, that part there sure looks like an arrow...
Cloudant is a Y-Combinator company whose founders have more than 10 years of experience managing multi-petabyte datasets.
Y Combinator is a new kind of venture firm specializing in funding early stage startups. We help startups through what is for many the hardest step, from idea to company. We invest mostly in software and web services. And because we are ourselves technology people, we prefer groups with a lot of technical depth. We care more about how smart you are than how old you are, and more about the quality of your ideas than whether you have a formal business plan.
Ironic, if ET is discovered by the kids down at your local middle school who have yet to outgrow their sweet tooth for Reese's Pieces.
Orwell: "In a Time of Universal Deceit, telling the Truth is a Revolutionary Act"
Use whatever those guys are using to computer generate music and start listening for other planets' latest teen sensations.
It's like back when analog was around and trying to find the right position for the cheap antenna because you couldn't afford something better... except with SETI you can't be reasonably sure there's something to be found or the Intergalactic Space Government hasn't forced people to switch to digital yet.
"Most people, I think, don't even know what a rootkit is, so why should they care about it?"
What always amazes me (besides the argument that SETI project will never know if it actually has anything) is the fact that a lot of the data could be processed faster on the GPU than the CPU. Why not use some OpenCL or Cuda or DirectCompute to harness parallel computing power ON TOP of the current schema? Processing Perlin Noise on the CPU is ~5x slower than on the GPU these days so why not map the data to a texture and run the filters on the GPU instead?
I don't understand this. Okay... so we're listening for signals coming from outerspace. These signals travel at the speed of light. Say we hear something that originated from 300 light years away... that means that it was 300 years ago when they emitted that signal. So in essence, if WE send out a "Hello?"... we're likely to get back a "What??" 300 years later. I really don't understand the logic, but more power to ya.
A)bort, R)etry, I)nfluence with large hammer
bool hasIntelligentAlienSignal( data signalData )
{
return FALSE;
}
Most ofn the signal we sent out will have gone below noise floor after 1 light year, so by our nearest neighbor sun would be already not recognizable. That nicve picture *DO NOT* take into account the reverse square law. The only one we can pretend is MAYBE recognize a few light year away is the arecibo signal. And that was a *3 MINUTES* signal. Hint : it was not a "ping" to show us, it was only a technical demonstration that it can be done. Face it, using our own waste of signal into space from start of broadcasting to today, we would not be able to detect ourselves from alpha centasuri ! No, the only stuff we could detect is a constant willful strong signal from ET civilisation saying "we are here". And that is it.
C. Sagan : A demon haunted world:
http://www.amazon.com/gp/product/0345409469/
visit randi.org
What they care for is finding other civilisation that way. Already knowing *somebody* is existing would be an incredible knowledge and confirmation to our theory that "we are not alone". You might not find that great, but to some of us , this would be an incredible finding. That said, I don#t think SETI will ever come to anything valid.
C. Sagan : A demon haunted world:
http://www.amazon.com/gp/product/0345409469/
visit randi.org
Check your inbox. -- S.R. Hadden
I dream up, test and use signal analyses, but don't code (not since AppleSoft). I rely on others for that, but they have to have a grip on time series, especially oscillatory/pseudo-periodic signals. My source of study material in the brain, but I've readily adopted techniques from things like radio astronomy, and others have adopted some of mine.
I've got a set of algorithms in mind that'd detect interesting signals for later examination. Two of three already have open source variants. The third is mine, and isn't open source because it's not code, it's a set of instructions to follow when using most any commercially available research EEG software package. I would expect to make it open source once coded.
The set of tests would probably go:
1. continuous wavelet transform, for time/frequency analysis, finding signals in small frequency bins that persist or come and go within a time series
2. synchronization, testing for multiple signals in a narrow band that may shift in phase or near freq
3. blind source separation, pulls apart multiple signals convolved into a single one. Does it whether or not they're there, so s/n ratio is used to determine probability of being real
If needed: 2 recursive back to 1, 3 recursive back to 2 or 1, 2 or 3 may trade places
The middle one's mine. I've used all 3, so I know they work and what they can do. Others have used the first and third for various reasons. Each adds benefits where others have failings. Each can and probably would do more than I describe here. Statistical analyses can and would be developed from them. My paper of my technique stays mine until publication, any code developed would be open but frankly is easier to do as common steps built into existing packages, as far as my field is concerned. But that only means the validity is at least as good as the software I developed the process under, and it's very good.
I'm fairly free to work on this, maybe even travel if it'd help. If anyone wants to dig into the details of what's being made available by SETI to see if they can develop, and wants to try this, threesigma at rocketmail dot com
"I may be synthetic, but I'm not stupid." -- Bishop 341-B
SETI has been interesting only in that it has demonstrated, at least at our technology level, nobody is home and using electromagnetism to communicate. The next level of communication is utilizing "ansible" technology which leverages quantum entanglement, aka Einstein-Rosen Bridge. Quantum entanglement has been demonstrated recently at a distance of 10 miles, and there is no reason to believe any amount of "physical" distance is relevant for the "effect". Lots of luck detecting communication using this technique.
I just happen to have a serious and concrete suggestion. http://www.cs.ucr.edu/~eamonn/iSAX/iSAX.html The referenced research paper includes experiments in which significant yet subtle changes in large time series data, such as a full night of EKG recording, can be identified two orders of magnitude faster than previous methods. The approach is relatively simple (the paper isn't heavy on math), and should scale very nicely to a parallel processing attack on the SETI signal detection problem.
But only because you've blown my mind!
It is no longer uncommon to be uncommon.
EVERY broadcast signal that starts on this planet, or the Moon, or in orbit radiates in ALL directions. We have little control over their directions, so we've been broadcasting into space for nearly a century. Plus, I think the Skrull, from a Fantastic Four comic a few decades ago said it best, when one of the Skrull told Reed Richards "If you knew what was out there in the universe, you would be HIDING, not trying to contact them..."
Stone
SETI themselves.
FAQ distance detection
1.2.3 How far away could we detect radio transmissions? By Al Aburto and David Woolley Representative results are presented in Tables 1 and 2. The short answer is (1) Detection of broadband signals from Earth such as AM radio, FM radio, and television picture and sound would be extremely difficult even at a fraction of a light-year distant from the Sun. For example, a TV picture having 5 MHz of bandwidth and 5 MWatts of power could not be detected beyond the solar system even with a radio telescope with 100 times the sensitivity of the 305 meter diameter Arecibo telescope.
(SnIpped but basically only narrow band transmission would be detectable beyond 1 LY).
stupid junk filter
Source ; Frequency ; Bandwidth ; Tsys ; EIRP ; Detection
; Range ; (Br) ; (Kelvin) ; ; Range (R)
AM Radio ; 530-1605 kHz ; 10 kHz ; 68E6 ; 100 KW ; 0.007 AU ;
FM Radio ; 88-108 MHz ; 150 kHz ; 430 ; 5 MW ; 5.4 AU
UHF TV ; 470-806 MHz ; 6 MHz ; 50 ? ; 5 MW ; 2.5 AU
Picture
UHF TV ; 470-806 MHz ; 0.1 Hz ; 50 ? ; 5 MW ; 0.3 LY
Carrier
WSR-88D ; 2.8 GHz ; 0.63 MHz ; 40 ; 32 GW ; 0.01 LY
Weather Radar
Arecibo ; 2.380 GHz ; 0.1 Hz ; 40 ; 22 TW ; 720 LY|
S-Band (CW)
(SNIP)
Note : basically with 1 LY I was even generous as it seems TV and radio won't even go past *pluto*. Therefore my point is that the JPG with all range is misleading as those signal will go nowhere below noise signal. The only signal going outside and which might be recognizable is the arecibo one.
C. Sagan : A demon haunted world:
http://www.amazon.com/gp/product/0345409469/
visit randi.org
Do aliens exist? Yes. Have they found us yet? No. Why? Look at the age of the universe. Something like 13 billion years old. Billion.
Humans have been around for 200k years or so.
13 billion. 200k. We're not even a blip in the time line. We're a itty bitty teeny weenie fraction of a blip.
SETI is a total waste of time.
Well, whether I want to lend my algorithmic genius to this or not depends.
Lemme ask you, do you plan on keeping discoveries secret, reporting them to the White House first, until such time as the President can decide what to do?
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
we assume that all the noise we are receiving already contains data? We could simply assume it is encrypted on a very high level, and therefore run a number of decryption schemes on it.
This just might be crazy enough to work - or just crazy. You choose.
Yes but what about colonization waves. I recently read a good novel, Manifold: Space by Stephen Baxter. It provides one possible answer. His previous book actually is based on no other aliens existing. That said while you make some sense I'm not convinced. Certainly we know from SETI how hard it is. (real hard)