[net.math] Number theory question

jwp@sdchema.UUCP (John Pierce) (10/05/83)

The alogrithm:	1)  Select an integer 'X'
		2)  If X is even, X = X/2 else X = 3X + 1
		3)  Iterate step 2 with the new value of X

The conjecture is that for all integers the alogrithm will eventually
generate the sequence "4 2 1" (which cycles).

Examples:  X = 6, 3, 10, 5, 16, 8, 4, 2, 1, 4, 2, 1 ...
	   X = 11, 34, 17, 52, 26, 13, 40, 20, 10 ... (as above from 10)
	   X = 51, 154, 77, 232, 116, 58, 29, 88, 44, 22, 11 ... (as above)

Does anyone know the name of this conjecture (I believe it was first proposed
by a Canadian), and whether or not it has been (dis)proven?

			John Pierce, Chemistry, UC San Diego
			ucbvax!sdcsvax!sdchema!jwp

leichter@yale-com.UUCP (Jerry Leichter) (10/10/83)

I believe this problem is due to Stanislaw [?] Ulam; the last I heard, it was
not known whether the algorithm settles into a cycle.  (There is a small
number - 41 or somewhere near there - that produces a VERY long run
before settling down.  Come to think of it, it may not even be know if
the algorithm EVER settles down, starting on that number.)

Ulam's conjecture has been discussed more than once in the Scientific
American Mathematical Games section.
							-- Jerry
					decvax!yale-comix!leichter leichter@yale

emjej@uokvax.UUCP (10/14/83)

#R:sdchema:-88400:uokvax:3300001:000:485
uokvax!emjej    Oct 12 09:21:00 1983

Funny you should mention that. Dr. Richard V. Andree here at the
University of Oklahoma came up with it one fine day to give a student
an example of a recurrence relation, and discovered that it was *far*
more interesting than it at first appeared. I recall hearing that every
starting value up to 10**40 either terminates or loops. A book by
Nievergelt et al., I think called *Computer-Assisted Problem Solving*,
gives further analysis of this recurrence relation.

						James Jones