Discussion Room : IDC101

Collatz conjecture

Re: Collatz conjecture

by Kapil Paranjape -
Number of replies: 0

A perhaps weaker question is whether this map eventually loops.

We can explain this better as follows. Let x_0=n and define x_{k+1} = x_{k}/2 if x_k is even and x_{k+1}=3x_k+1 if x_k is odd. We can now ask if this map "loops". In other words, is there always a pair of natural numbers p and q>0 so that x_{p+q}=x_p? It is not obvious that this question is the same as the Collatz sequence question.

In fact, we can define more general maps as follows. Fix a "base" b. Starting with x_0=n we put x_{k+1}=x_k/b if x is divisible by b; other wise, let r be the remainder of x_k on division by b and put x_{k+1}=(b+1)*x_k+(b-r). Note that in the second case x_{k+1} is divisible by b. So every two steps x_{k+2} is at most about (b+1)/b times x_k. We can ask the same question about this map.

Write a program to find "loops".