31 Dec '04 08:30>
Originally posted by BigDoggProblemFalse alarm - these unusual cases (n1 and n2 both very close to 100) only revealed a minor programming bug. The cases of (98, 99) and (99, 99) almost lead to alternate solutions. Both can be summed only one way, because of the limit on n1 and n2. Therefore they lead to only one acceptable sum - but Mr. SUM would know n1 and n2 in advance because it's TOO narrowed down. I think the creator of the problem got a bit lucky here! (Especially if he assumed that all acceptable sums must be odd numbers....)
I just realized that I have not run the program since I set it to exclude sums with n1 or n2 greater than 100. Now it claims that (96, 99) and (97, 99) are solutions! I don't have time to analyze them by hand right now, but this is an interesting development nonetheless.
Could it be that the problem is unsound??
So, my earlier conclusions have not changed. I believe (4, 13) is the solution to the original problem (even though you can't assume all acceptable sums are odd, making the solution a bit less elegant).
I also still hold that the problem becomes unsound once you allows (437, 437) or higher.