Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

COLLATZ EXPLANATION

1. Short Introduction
In this short note, we let x ∈ Z be an odd integer to which we associate the
corresponding Collatz sequence (xi )∞i=0 where x0 = x and
3xi + 1
(1) xi+1 =
2 pi
where pi is the largest power of 2 dividing 3xi + 1. Further, we have another
sequence associated to x defined as
2 pi − 4
ai = .
3
Using these definitions, equation (1) can be rearranged as
4
(2) xi = 1 + ai + (xi+1 − 1)(ai + ).
3
In particular, when i = 0 this equation becomes
4
x0 = 1 + a0 + (x1 − 1)(a0 + ).
3
Using equation (2) we can substitute the value for x1 to achieve the expansion
4 4
x0 = 1 + a0 + (a1 + (x2 − 1)(a1 + ))(a0 + )
3 3
4 4 4
= 1 + a0 + a1 (a0 + ) + (x2 − 1)(a1 + )(a0 + ).
3 3 3
Continuing this pattern inductively we have, for any i > 1, the (finite) expansion
i−1  k−1  i−1
X Y 4 Y 4
x = 1 + a0 + ak (aj + ) + (xi − 1) (aj + ).
j=0
3 j=1
3
k=1

To simplify this somewhat cumbersome notation, we introduce the sequence


(zi )∞
i=0 with
z0 = 1,
z1 = a0 ,
i−2
Y 4
zi = ai−1 (aj + ) for i > 1.
j=0
3
Furthermore, we set
i−1
Y 4
O(xi ) = (xi − 1) (aj + ).
j=1
3
The reason for the introduction of such terms is so that we have the nice expression
i
X
(3) x= zk + O(xi )
k=0
1
2 COLLATZ EXPLANATION

valid for all i > 1.


We now answer the question of whether or not we can write

X
x= zk
k=0
∞ X
i−1  k−1 
X Y 4
= 1 + a0 + ak (aj + )
i=2 k=1 j=0
3

2. Validity of Infinite Expansion


The crucial question to ask is whether or not we can extend equation (3) to an
infinite sum. Several sums in mathematics arise this way which can be extended
to the infinite case, e.g. the Taylor series. However, such expansions rely on the
following crucial observation.
Proposition 2.1. Let f : Z → R be a function. Suppose that y ∈ R is a number,
and (yi )∞
i=0 is a sequence such that
M
X
y= yi + f (M )
i=0

for all M ≫ 0 (the double inequality means “for all sufficiently large M ”). Then

X
y= yi
i=0

if and only if f (M ) → 0 as M → ∞.
The proof is quite easy so I won’t include it here. Now we compare the above
proposition with equation (3). We take f (M ) = O(xM ) and yi = zi . In order to
be able to write
X∞
x= zi ,
i=1
we need that O(xi ) → 0 as i → ∞. Do we have this? No, not necessarily. If the
Collatz conjecture is true, then in fact O(xi ) = 0 for i ≫ 0, but if it is not true,
then it is unclear whether or not O(xi ) converges to 0.
Consider the simple case of an odd number x > 1 whose Collatz sequence is
periodic with period n, i.e. x0 = xn , in which pi > 1 for each i ∈ {0, 1, ..., n − 1}.
Then
i−1
Y 4
O(xi ) = (xi − 1) (aj + ),
j=1
3
and it follows from periodicity that every factor in the above product is greater
than 1. In this case, O(xi ) → ∞ as i → ∞ and the infinite expansion is not valid,
meaning that

X
x ̸= zi .
i=1
We may even relax the assumption that pi > 1 for all i > 1 to pi > 1 for most of
the i > 1.
COLLATZ EXPLANATION 3

It may very well be the case that such an example does not exist, and indeed
it would not if the Collatz conjecture was true. But, to my knowledge there is no
proof that such an example is impossible.

You might also like