Date: Wed, 19 Jun 2002 12:03:30 -0400 From: Steve Smith <sgs at aginc.net> To: WSFA members <WSFAlist at keithlynch.net> Subject: [WSFA] Re: Goldbach's Conjecture Reply-To: WSFA members <WSFAlist at keithlynch.net> ronkean at juno.com wrote: > Continuing further by hand would be increasingly tedious, but presumably > a modern computer could extend the computations to some very large > number. But, since a computer could never count to infinity (in a finite > time), it seems that it would be impossible for a computer to prove the > conjecture for all even numbers. But, a human has allegedly done so. > Assuming that has been done, what is the difference between a human > mathematician and a computer which explains why the human can do what the > computer cannot? Because proofs are done by reasoning about the problem, not brute calculation. There are specialized "therom proving" computer programs, but I don't know how effective they are in the general case. My experience with similar programs is that it's almost as much work to code the problem into a form that the computer can solve as it is to solve it by hand. Simple algebraic proof: We want to prove that the sum of the first "n" numbers is given by the formula S = n(n+1)/2. First, we show that it works for one special case. If n=1, the sum is obviously 1. S = 1(1+1)/2 = 1. Okay, that works. Now we show that, if it works for "n", it will work for "n+1": S' = S + (n+1) = n(n+1)/2 + (n+1). Letting "y" = n+1 S' = (y-1)y/2 + y = [(y-1)/2 + 1 ]y = [(y-1) + 2]y/2 = [y+1]y/2 So if the formula works for "n", it will work for "n+1". We've shown that it works for n=1. Since it works for 1, it will work for 2. Since it works for 2, it will work for 3. Since ... So we've shown, in just a couple of lines, that our formula will work for *any* number. No calculation required; just a bit of algebra. -- Steve Smith sgs at aginc.net Agincourt Computing http://www.aginc.net "Truth is stranger than fiction because fiction has to make sense."