Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
×
Math

52nd Known Mersenne Prime Found (mersenne.org) 61

chalsall writes: After more than six years of work since the last discovery, the Great Internet Mersenne Prime Search (GIMPS) has found the 52nd known Mersenne Prime number. This is also the largest prime number known to humans.

The number is 2^136,279,841-1, which is 41,024,320 decimal digits long.

Luke Durant, a researcher from San Jose, CA, found it after contributing a fantastic amount of compute to the GIMPS project.

This discussion has been archived. No new comments can be posted.

52nd Known Mersenne Prime Found

Comments Filter:
  • But why? (Score:1, Interesting)

    by raburton ( 1281780 )

    > Luke Durant, a researcher from San Jose, CA, found it after contributing a fantastic amount of compute to the GIMPS project.

    > (From TFA) It is only the 52nd known Mersenne prime ever discovered, each increasingly more difficult to find.

    So it's like mining bitcoin, but much less useful. At least you can sell the bitcoin for real money.

    • From what I can tell they're only useful for finding perfect numbers, which fascinate some people.

      I don't care too much but I don't want to hear the same people complain about poor people taking hot showers because "the planet".

      Maybe some day we'll find higher meaning in perfect numbers - I dunno.

    • Re: But why? (Score:4, Insightful)

      by gijsterbeek ( 1735024 ) on Monday October 21, 2024 @11:17AM (#64881381)
      Ah, useful science! If we science would only be measured against what is today deemed useful, and useful would only be measured as what today can be sold for money, mankind would never have had electricity, the wheel, fire, and such.
      • I think you're propagating a vacuous Subject, though I rather like your comment.

        However what your comment and the story both reminded me of is a recent report about a new breakthrough in recognizing primes. The clickbait intro described it as the first major advance in 25 years--but when I clicked it seemed to be a scientific website that would not show me more unless I subscribed. From the visible bits I couldn't even figure out what to search for to find a public version of the result. (But now it's an ex

    • God damn that's a heckuva definition of useful.

    • So it's like mining bitcoin, but much less useful. At least you can sell the bitcoin for real money.

      The ability to find value in things other than their utilitarian or monetary worth is one of the things that distinguishes us from lower forms of life.

    • Slashdot bitches and moans about the power consumption needed to run LLMs that people actually use in their daily lives, but 6 years of computation to find a number solely because it tickles mathematicians' fancies and they don't say peep.

    • Because it is there! (George Mallory [wikipedia.org])
    • For science! Not everything in life is about money. If you're going to waste electricity, better to do it on science which is more useful to the world than the scam that is bitcoin.

  • Decimal? (Score:4, Funny)

    by rossdee ( 243626 ) on Monday October 21, 2024 @11:16AM (#64881379)

    "The number is 2^136,279,841-1, which is 41,024,320 decimal digits long."

    Wouldn't it be easier to express in binary, as every digit would be 1

  • Isn't 2^2-1 a Mersenne prime?

    • Re:Just 52? (Score:5, Informative)

      by pjt33 ( 739471 ) on Monday October 21, 2024 @11:22AM (#64881413)

      Yes, that's the first Mersenne prime. Small primes have better chances of working as Mersenne exponents because there aren't many candidates to divide 2^p - 1. The first 48 such exponents are listed in OEIS A000043 [oeis.org]; I'm not sure why no-one has added the 49th and onwards to the B-list.

      • Indeed. The first prime exponent that is not prime is 2^11 - 1 = 23 * 89.

        The next is 2^23 - 1 = 47 x 178,481.

        There are 10 Mersene primes below 100. After that, they get sparse.

        • "Indeed. The first prime exponent that is not prime is 2^11 - 1 = 23 * 89."

          It should be noted that only prime exponents are of interest. It is not required by the definition of a Mersenne prime that the exponent be prime, but it has been shown that any Mersenne prime will have a prime exponent.

          • Re:Just 52? (Score:5, Informative)

            by TeknoHog ( 164938 ) on Monday October 21, 2024 @04:06PM (#64882355) Homepage Journal

            it has been shown that any Mersenne prime will have a prime exponent.

            This is easy to explain in binary, where for example 2^5 - 1 = 11111_2 and in general 2^N - 1 = N ones. If N is composite such as 6 = 2*3, you can split the binary number 111111 into groups of 2*3 ones as in 11 11 11, which can be written as the product 010101 * 11. So every composite exponent makes the resulting Mersenne number composite.

            • Oh, wow. That method is really slick.

              I already commented about factoring a "Mersenne composite" that is equivalent, but your method is more intuitive.

              My example was 3*5, which would be:

              2^15 - 1 = 111111111111111 = 11111 * 10000100001 = 31 * 1057

              So, same answer, but more intuitive in binary.

              • I think your solution makes a much better math proof, as it doesn't rely on visual handwaving, and clearly works for any integer m and n. But I think we need both kinds of explanations to get a better feel for the math.

                It's also interesting that by swapping the m and n in these proofs, you get a different factorization if m != n. For example 111111 = 010101 * 11 = 001001 * 111. So one of the two factors must be also composite, and the Mersenne number has at least 3 factors.

          • it has been shown that any Mersenne prime will have a prime exponent.

            This is very easy to prove.

            If the exponent is composite, then by definition, it can be written as the product of two integers, n*m.

            Then, it can be factored as:

            2^(n*m) - 1 = (2^n - 1) * (1 + 2^n + 2^2n + 2^3n + 2^4n + ... + 2^((m - 1)*n))

            Which is obviously composite.

            QED.

            Example: n = 5, m = 3.

            2^(5*3) - 1 = (2^5 - 1) * (1 + 2^5 + 2^(2*5))
            2^15 - 1 = (31) * (1 + 32 + 1024)
            32767 = 31 * 1057

            Of course, the inverse is not true. A prime exponent is necessary but not sufficient, with 11 being the first example.

    • by HiThere ( 15173 )

      No. For some reason 1 doesn't count as a prime number.

      • Eh?

        2^2-1=3

      • by jonadab ( 583620 )
        The original and most common definitions of "prime" and "composite" break down for n<2. Sometimes when old definitions only cover a subset of the domain you're interested in, like this, it's useful to create an extended defintion that works on a larger set, so that you can do things like raise numbers to fractional and/or negative powers, take factorials of numbers that aren't integers, and so on. But for the definitions of "prime" and "composite", nobody to my knowledge has found a useful way to exten
  • Forgot the link (Score:4, Informative)

    by jmccue ( 834797 ) on Monday October 21, 2024 @11:28AM (#64881431) Homepage

    This is the link: https://www.mersenne.org/ [mersenne.org]

    In honor of this, I kicked it off again, but I need to use "Throttle=10" on my laptop, otherwise it will catch fire :) I won't find anything, but it will help with somekind of verification.

    With that throttle, and my BOINC settings, I can now have World Community Grid going along side mprime while keeping the Laptop CPU Temp under 35F (55C).

    And yes, running mprime is fun for me and with my settings, it uses less electricity than watching a youtube video.

    • I also have BOINC running on the two desktops I have at home but neither have GPUs so sadly I cannot participate in this project.
      • I also have BOINC running on the two desktops I have at home but neither have GPUs so sadly I cannot participate in this project.

        The official GIMPS software runs on CPUs only. GPU software for finding primes has been around for a long time, but their use with GIMPS remains a little experimental. Generally, the GPU worker doesn't deal with the network for connecting to GIMPS servers, so you need another piece of software for that. Currently GIMPS recommends [mersenne.org] AutoPrimeNet [github.com] (which is based on my earlier Primetools [github.com]).

    • "while keeping the Laptop CPU Temp under 35F (55C)."

      That doesn't seem right, maybe 135F would be closer to 55C

  • If a computer does the calculation, can we truly be certain that the output is correct?
    • If computers did not reliably do their calculations correctly, you could not have posted that.

    • Sure, you can verify it via a second calculation, if you like. It'll be much easier to confirm (or deny) than it was to find in the first place.

    • Certainly we can trust it more than if a human did this by hand. Which a human likely cannot do given the observed statistical limits of human lifespans. Once we've got this number it can be validated over and over (with a lot of time but not as much as it took to find it), with different algorithms or implementations of algorithms.

  • by Dan Posluns ( 794424 ) on Monday October 21, 2024 @11:53AM (#64881513) Homepage

    I don't know what this tangibly accomplishes for math or science - maybe nothing - but there is something insane and remarkable to think that a number that unimaginably large has no divisors other than 1 and itself, and something monumental about the fact that we were able to identify it.

    • I'm in the opposite camp where it almost seems unremarkable. I realize the way exponents work and that is a VERY large number, but intuitively computers seem so fast and we have so many of them that the fact we've only been able to check 136,279,841 of them is surprising.
      • I'm not sure you can say you understand how exponents work and that intuitively you think it's surprising that it's hard to check this number. Checking large numbers is a problem of exponential complexity increase. That's kind of the problem with exponents, and this level of complexity increase is what we rely on ensuring we have security mechanisms which computers can't easily crack. Just in this case it's applied to calculating primes.

        • by narcc ( 412956 )

          It's not something I would have expected from a 6-digit uid, but the attitude isn't uncommon. For the last 20+ years we've been telling kids that memory is infinite and that performance issues are best addressed by adding hardware. It's not hard to find a CS grad that doesn't understand, or care about, time complexity.

          If he were to try implementing a function that takes a positive integer n and checks to see if 2^n-1 is prime, I expect that he'd rather quickly develop an appreciation for exponential growth

          • I am a CS graduate and understand the tyranny of confirming that something is prime, but if you'd asked me to guess how long the longest known prime number was I think i'd have guessed more digits than that.
            • by narcc ( 412956 )

              ... you would have guessed the largest known prime had more than 41 million digits, requiring more than 17 megabytes to store in memory?

              I understand the tyranny of confirming that something is prime

              You may want to reevaluate that belief.

              Remember that GIMPS [mersenne.org] is exclusively looking for Mersenne primes, which dramatically cuts down on the search space, and even with massive amounts of computational power behind it, it still took 6 years to find after the previous record.

  • Damnit. (Score:4, Funny)

    by UnknowingFool ( 672806 ) on Monday October 21, 2024 @12:04PM (#64881533)
    Now I have to change my luggage combination again. Also the passcode to the air shield. Damn you mathematicians!
  • The last five decimal digits of the number are 71551, or TISSI which is Finnish for "titty". This is easy to check e.g. in Python using

    pow(2, 136279841, 10**5) - 1

    (FWIW, I'm the author of Primetools [github.com] which connects GPU workers to the GIMPS project. It has later been forked/extended into different projects such as AutoPrimeNet, which is currently recommended [mersenne.org] by GIMPS.)

  • The number is 2^136,279,841-1, which is 41,024,320 decimal digits long.

    Hey⦠Thatâ(TM)s my luggage combination!

  • by PPH ( 736903 )

    Congratulations! Not a trivial task to do that level of math work while wearing latex hoods, dog collars and leashes.

  • I was certain I'd dropped it somewhere in the garage, I must be thinking of my O(log(n)) travelling salesman algorithm instead.

/earth: file system full.

Working...