1) Start with a natural number, n0Note the conditional statement expressed in steps 2-3. The conjecture, as expressed in step 4, is that such sequences will always terminate in one, independently of the starting value, n0. The conjecture is unproven, but all sequences tested have ended in one.[2] The following plot shows the sequence starting at n = 983. Collatz sequence starting with n = 983. This interesting recursion was discovered in 1932 by Lothar Collatz, who was at the time a twenty year old mathematics student. (Plot via Gnumeric) As recalled in reference 3, when mathematician John Conway once visited Richard K. Guy, he spent some time in transit calculating subprime Fibonacci sequences, or Conway sequences as named in his honor. These sequences are built like the Fibonacci sequence, with two starting numbers. As in the Fibonacci sequence, the last two terms are summed to get the next term.[3] The next term of a Conway sequence is not used as calculated, but it's also subject to a conditional statement, just as in a Collatz sequence. If the term is a composite number, you divide by its smallest prime factor to get the next term. Because of the conditional statement, these sequences exhibit pseudo-random behavior, and the sequence repeats after a number of cycles. You can see a schematic of the 56-cycle sequence in the figure.[3]
2) If n is even, then ni+1 = ni/2
3) If n is odd, then ni+1 = 3ni + 1
4) Continue at step (2) until n = 1
The 56-cycle Conway sequence, from Richard K. Guy, Tanya Khovanova and Julian Salazar, "Conway's subprime Fibonacci sequences."[3] Shown are the nodes of the sequence, which are coprime odd integers not preceded by an odd term, and the number of terms between them. (Via arXiv Preprint Server)[3]. |
The 136-cycle Conway sequence. Unlike the Collatz sequences, which terminate in one, this sequence is periodic after a few initial terms. (Graph rendered by Gnumeric). |