From Science News:
Here’s a number to savor: 243,112,609-1.
Its size is mind-boggling. With nearly 13 million digits, it makes the number of atoms in the known universe seem negligible, a mere 80 digits.
And its form is tidy and lovely: 2n-1.
But its true beauty is far grander: It is a prime number. Indeed, it is the largest prime number ever found.
The Great Internet Mersenne Prime Search, or GIMPS, a computing project that uses volunteers’ computers to hunt for primes, found the prime and just confirmed the discovery. It can now claim a $100,000 prize from the Electronic Frontier Foundation for being the first to find a prime number that has more than 10 million digits.
Don’t worry prime hunters, there are prizes still to be claimed:
The Electronic Frontier Foundation became interested in prime hunting because it makes an excellent challenge problem for cooperative, distributed computing. “The award is an incentive to stretch the computational ability of the Internet,” says Landon Noll of Cisco Systems Inc., one of the judges for the Electronic Frontier Foundation prize and a discoverer of a former biggest known prime. More prizes remain to be claimed: a $150,000 award for a prime with 100 million digits, and a $250,000 award for one with a billion digits.
In case you are wondering why I’m posting this here on my blog, I do have some personal historical trivia that makes the issue of large primes sentimental for me.
The first job I ever had writing software was an unpaid high school internship at NASA Ames Research Center, here in Mountain View. My project was to build a simulation model to evaluate error rates for different fluid dynamics algorithms. In order to do the project, which was executed on a Cray X-MP supercomputer, I had to learn Fortran.
The sample project I chose to do to learn the language was a simple program to take as input a Mersenne Prime, and then generate the actual digits for the number in a large output file.
As a side note, this was the first time I also ever became familiar with the operating costs of these type of high end systems… I remember being fairly shocked when the scientist I was working with explained to me that my program had taken several hours of Cray time, which was billed at about $2,000 per hour.
Of course, I’m fairly certain that my new 8-core Mac Pro is significantly faster than those old Cray supercomputers… 🙂