Slashdot Mirror

← Back to Stories (view on slashdot.org)

Distributed.net Starts New Project

Posted by ryuzaki0 on from the prove-to-the-world dept.

drydorn writes "Today, distributed.net will officially begin its next distributed computing project. Visit their Optimal Golomb Rulers project page for more details. Their first ruler length will be 24 marks, known in D.net lingo as OGR-24. " And, remember, your mantra: I must sign up for Slashdot Team. I must crack keys. You can grab your client here, which includes documentation on installation, what clients do, etc. etc.

19 of 86 comments (clear)

  1. Finally! by barzok · · Score: 2

    I've been getting sick of RC5-64 (over 2 years now, and no end in sight) and SETI just doesn't interest me. Finally, a contest I can sink my teeth into AND has real, usable results.

    1. Re:Finally! by jari · · Score: 2

      As an alternative, what about the Gamma Flux project over at dcypher.net ?

      This has a useful application to it - ray tracing for making safer containers to hold radioactive waste.

      Stats aren't quite as cool as distributed though ;)

    2. Re:Finally! by barzok · · Score: 2

      Ah, dammit, now I gotta pull out my script that I used to run OGR-23 12 hours/day and then switch to RC5-64 for 12 hours so I can toggle between OGR-24 and Gamma Flux. Looks pretty cool.

  2. At last! A *real* astronomical project. by szyzyg · · Score: 2

    After the huge publicity surrounding SETI@home and the various arguments for and against it we've now got a project which has *real* astronomical applications.

    OK OK I will concede that SETI@home has a small chance of finding alien life, but, the chance is still remote and the amount of data that they are processing is miniscule compared to what we really need to be doing to seriously have a chance.

    Large OGR's will of course help improve the sensitivity of Long Baseline arrays and other sensors, and therefore improve the quality of data produced. So... who knows - maybe running your CPU on this will help the search for extra terrestrial intelligence more than running seti@home.

    Personally I'm waiting for distributed.net to help the Spaceguard foundation save the world from cosmic hazards so that other alien races will have someone to talk to in the future.

  3. All np complete problems become shallow. by Poe · · Score: 2

    Perhaps annother way for companies to generate revenue would be to include sufficiently generic distributed computational software with their clients. This kind of app could also appear in web pages as a Java applet. (I would, of course, advocate an option to turn it off, and "nice"ing of all processing) Imagine what kind of stock market analysis or data mining one could do with ICQ or yahoo.com so enabled. Companies would pay good money for access to such a powerful processing force.

    --
    Thank you for not thinking.
  4. Mathematical masturbation by Sanity · · Score: 2
    I never understood why distributed.net always waste their time trying to solve these abstract mathematical problems that really aren't anything more than "my distributed penis is bigger than your non-distributed penis" competitions. We all know roughly how much computing power it takes to crack a key, or do this Golumb ruler thing, it is only a question of whether we have or haven't done it yet (and the world isn't really any better-off if we have). Nothing is proved by achieving it, we already know exactly how difficult and statistically how long it will take. On the other hand, genetic programming is an ideal application to be tackled in a distributed fashion, and could be used to evolve some really interesting stuff (like evolving sorting algorithms - or even creatures which learn to walk!). These guys have the right idea, but what is really needed is for someone like distributed.net to get in on the act.

    If you are interested in genetic programming take a look here for more info.

    --

    1. Re:Mathematical masturbation by Sanity · · Score: 2
      I am not saying that distributed computation is not a powerful tool, I am just saying that it is a shame that this tool has not been put to better use. Ok, so OGRs may have uses in some obscure field, but surely we should use this resource for something that will be of interest to more people. As for distributed.net proving the insecurity of encryption - this is meaningless. We can tell precicely how much computing power it will require to crack any given encryption algorithm (both worst case, and average case). We can even estimate how long a system such as distributed.net would take to do it, actually doing it adds nothing whatsoever to the debate, it is a waste of time and resources, and tells us nothing. Much more interesting to actually create something (such as a new sorting algorithm or A-life).

      --

    2. Re:Mathematical masturbation by Sanity · · Score: 2
      I don't question the value of astronomy, but this is dealing with such an obscure area that I really doubt it will build up much interest. Now evolving a walking algorithm - that is something where everyone could visibly see the results and would likely build up much more interest and support.

      --

    3. Re:Mathematical masturbation by gargle · · Score: 2

      I'm currently taking John Koza's (the inventor of GP) class on genetic programming at Stanford, and his company has built a 1000 node beowulf cluster to solve GP problems. Genetic programming is a powerful, generic, and highly parellizable method of solving difficult problems - GP has already managed to produce some patentable designs, and distributed.net could make potentially make a real contribution to humanity if they were to apply themselves to GP or GA.

  5. impostereous by LinuxGeek · · Score: 2

    Hmm... And all this time I thought that 'Hemos' was Jeff Bates, now you claim that 'Hemos.' is Jeff Bates. I, sir have met Jeff Bates, and you are no Jeff Bates. ;o)

    --

    Kindness is the language which the deaf can hear and the blind can see. - Mark Twain
  6. And then there's the Tierra project by grinder · · Score: 3

    The distributed computing project I'd most like to see get of the ground is The Tierra Project.

    This project is exploring digital evolution. Start off with a bunch of organisms and breed them with genetic algorithms. See how they fare.

    Then, and this is where it gets interesting, an organism can migrate from one host to another, possibly taking better advantage of the environement there. What kinds of digital ecologies will appear? What kinds of emergent behaviour will be encountered?

    It's actually much more complex than that. If you're curious, I recommend reading the introduction.

    1. Re:And then there's the Tierra project by Sanity · · Score: 2
      Tierra is somewhat interesting, but it has been around for ages, and while interesting at first, it seems to have stagnated. Tierra is interesting because it leaves evolution to itself, the organisms are given an environment in which they can exist, and they breed and mutate as a result of this environment. In most genetic programming experiments the breeding and mutation is much more contrived - somewhat akin to breeding horses where they are deliberately selected according to certain criteria (although normally this "breeding" is performed by a computer which blurs the distinction). The problem is that I don't think experiments like Tierra scale very well, after a while they converge to a few different types of organism and then things stay more or less the same at a macro level (much as has happened with our biosphere!). I think there is a place for the more contrived G.P experiments as they permit a much wider variety of stuff to be evolved.

      --

  7. Re:I was with you for the first 3.5 sentences by Sanity · · Score: 2
    At least Joe Public programmer can see why sorting algorithms are useful, I am not suggesting that a GA will come up with a better sorting algorithm, although it might come up with a sorting algorithm which is better in specific areas.

    --

  8. Its tomorrow now? by Tridus · · Score: 2

    I just checked the d.net homepage and it now says that they're starting on tuesday.

    Wonder what happened there, I guess I'll have to go onto irc and ask.

    --
    -- "So they told me that using the download page to download something was not something they anticipated." - Bill Gates
  9. Re:What's the Point? by Kris_J · · Score: 2
    Nothing really, but nobody questions spending money on music, movies etc.
    Indeed. I liked the comparison that Water World cost more than NASA's failed Mars Polar Lander project.
  10. Re:I was with you for the first 3.5 sentences by gargle · · Score: 2

    If all you can do is compare two keys at a time, to tell which one is bigger, then any sorting algorithm has a worst case of O(n lg n). And, since there are already algorithms that achieve this worst case, yes, we've found the best there is.

    Not really. Quicksort has the same worst case and "average" (from a analytical, random input standpoint) performance as mergesort, but yet quicksort performs better "in practice". Worst case or even average case analysis doesn't tell you everything.

  11. What's the approach? by Kris_J · · Score: 2

    Does anyone know if D.Net's OGR client uses some sort of eligant mathematical technique, or is it just another brute force effort?

  12. Re:I was with you for the first 3.5 sentences by Abigail-II · · Score: 2
    Not really. Quicksort has the same worst case and "average" (from a analytical, random input standpoint) performance as mergesort, but yet quicksort performs better "in practice".

    That's wrong. Mergesort is O (n log n) worst case. Worst case quicksort is Omega (n^2). And to make it worse, common implementations of picking the pivot element (first element, last element, median of the first three elements) have sorted inputs as their worst case (that is, they produce their own worst case input). Even if you pick a pivot at random, there is a non-zero chance you always pick an extreme, leading to quadratic behaviour.

    Now, you *can* find a median of a set in linear time, and using such a method to find the pivot leads to a worst-case O (n log n) sorting algorithm. However, the overhead is so much, the resulting algorithm will be slower, more complicated, and certainly less elegant than either mergesort or heapsort.

    References:
    Knuth, D.E: The Art of Computer Programming, Vol III, Sorting and Searching, 2nd edition, Addison-Wesley, 1998. ISBN 0-201-89685-0.
    Cormen, T. H., Leiserson, C. E. and Rivest, R. L.: Introduction to Algorithms MIT Press, 1990, ISBN 0-262-53091-0.
    Hoare, C.A.R.: "Algorithm 63, Partition; Algorithm 64, Quicksort" Communications of the ACM, Vol 4, 1961, p 321.

    -- Abigail

  13. Re:I was with you for the first 3.5 sentences by gargle · · Score: 2

    That's wrong. Mergesort is O (n log n) worst case. Worst case quicksort is Omega (n^2).

    Right, I forget. But the point remains that worst case or average case analysis doesn't tell you everything you need to know about the algorithm's performance - even though (as you point out) quicksort has an inferior worst case performance, it performs better 'in practice'.