5
$\begingroup$

How would I go about proving the following:

For any odd positive integer $s$, there exists a sequence of nonnegative integers $( a_0, a_1, \cdots, a_{n-1})$ and a nonnegative integer $m$ such that,

$$ 3^ns + \sum_{k=0}^{n-1} 3^{n-k-1}2^{a_k}=2^m.$$

I am really stuck. I was thinking of using the ring $\mathbb{Z}_{2^m}$ in a proof by contradiction, but I cannot even get started reducing the LHS to something simpler.

Note 1: I have reason to believe there exists such a sequence where

  1. $a_0=0$
  2. $\{a_n\}$ is strictly monotonically increasing

Note 2: I think an example might help.

$$ n=3 \quad \{a_n\} = [0, 2g-1, 2g+3] \quad s = \frac{5\times 4^g -2}{6} \quad m= 2g+5 \quad g >0$$

Note 3: I originally asked this on the Mathematics Stack Exchange, but it seems to be a question more suited for this exchange.

Note 4: The full example for $n=3$.

If $s$ is an odd positive integer such that its orbit contains exactly $3$ odd integers including $s$ and $1$ then $s$ has exactly one of two forms:

$$S = \frac{2^{6j+2g-3} - 2^{2g-1} -3}{9} \quad \text{ with } \quad a_{g,j}= \{ 0, 2g-1, 6j+2g-3 \} \quad g,j > 0$$

or

$$S = \frac{2^{6j+2g+2} - 2^{2g} -3}{9} \quad \text{ with } \quad a_{g,j}= \{ 0, 2g, 6j+2g+2 \} \quad g,j > 0.$$

Its possible to get the exact forms for bigger orbits in the sense that the orbits contain exactly $k$ odd numbers including $s$ and $1$, it just gets harder and more tedious. Also, this thing needs a proof.

Note 5: Equivalently, the above can be stated as follows:

If $s$ is an odd positive integer such that its orbit contains exactly $3$ odd integers including $s$ and $1$ then $s$ has exactly the following form:

$$S = \frac{2^{j+g} - 2^{g} -3}{9} \qquad a_{g,j}= \{ 0, g, g+j\} \qquad g,j > 0 $$

where

$$ 2^{j+g} - 2^{g} -3 \mod 9 = 0 $$

$\endgroup$
12
  • 6
    $\begingroup$ Isn't this a rephrasing of the 3n+1 problem ? (It really sounds like the type of expression you get after applying the 3n+1 iteration to s until you get 1... ) $\endgroup$ Oct 30, 2019 at 18:19
  • $\begingroup$ The 3n+1 problem has a condition on $a_n$ which is not required here. $\endgroup$ Oct 30, 2019 at 18:34
  • 2
    $\begingroup$ Reminds me of a puzzle due to Erdos, from Mathematics Magazine, February 1994: prove that any positive integer can be written as the sum of terms of the form $2^a3^b$ where no summand divides another. See also Richard Blecksmith, Michael McCallum and J. L. Selfridge, 3-Smooth Representations of Integers, The American Mathematical Monthly Vol. 105, No. 6 (Jun. - Jul., 1998), pp. 529-543. $\endgroup$ Oct 30, 2019 at 22:06
  • $\begingroup$ At a short view it seems it is required that $m$ and $a_{n-1}$ must be such that $3 | 2^m - 2^{a_{n-1}} $ ... $\endgroup$ Oct 30, 2019 at 22:17
  • 1
    $\begingroup$ @SimonHenry: in the formulation $ {3^ns + ... \over 2^m }=1$ where $m>a_{n+1}$ it is the codification of the orbit of the odd positive number $s$ downto $1$ by the iterated Collatz-transformation (in the "Syracuse"-notation of the problem). Don't see the different restrictions on $a_k$ here against that of the Collatz-transformation, btw. ($a_0=0$, $a_{k+1}\gt a_k$ ). I think that formula is also in wikipedia. $\endgroup$ Oct 30, 2019 at 22:29

2 Answers 2

4
$\begingroup$

The expression you've coined reflects the orbit of one initial number $s$ towards $1$ by (the Syracuse-notation of) the Collatz-transformation. A perhaps better expression for this is $$ a_{N+1} = \small {3^N a_1 + (3^{N-1} + 3^{N-2} \cdot 2^{A_1} + 3^{N-3} \cdot 2^{A_1+A_2} + ... + 3 \cdot 2^{A_1+A_2+...+A_{N-2} } + 2^{A_1+A_2+...+A_{N-2}+A_{N-1} } )\over 2^{A_1+A_2+...+A_{N-2}+A_{N-1} + A_N } } $$ where the denominator is your expression $2^m$ .
Let's make this monster-expression shorter; express the parenthese by the shortform $$ Q([A_1,A_2,...,A_N])=3^{N-1}+3^{N-2}\cdot 2^{A_1} + \cdots + 2^{A_1 +...+ A_{N-1}} $$ and $S = \sum_{k=1}^N A_k$ (I like the capital letters $N$(-umber-of-steps/-exponents) and $S$(-um-of-exponents) and $A_k>0$ for the terms in exponents instead of small letters like $m$ as you use it here).

Then we have in general

$$ a_{N+1} = a_1 \cdot {3^N \over 2^S} + {Q([A_1,...,A_N])\over 2^S} $$

Your first question is to prove, that for all odd $a_1$ there is a vector $E(a_1)=[A_1,A_2,...,A_N]$ such that $a_1 \mapsto 1$ by finitely many $N$ steps.
As it is well known, nobody has a proof so far and thus the Collatz-problem remains an open problem until now.

Your other observation is that of properties of three-step orbits ending at $1$. For this I propose to revert the notation: which numbers $a_3$ can be reached by the inverse Collatz-transformation of $N=3$ steps.
We can write it this way: $$ a_{k-1} = {a_k 2^{A_k}-1\over 3} \qquad \text{where } a_k \equiv 2^{-A_k} \pmod 3$$ and rewriting $$ a_{N+1} = a_1 \cdot {3^N \over 2^S} + {Q([A_1,...,A_N])\over 2^S} \\\ a_4 = a_1 \cdot {3^3 \over 2^S} + {Q([A_1,A_2,A_3])\over 2^S} \\\ 1 = a_1 \cdot {3^3 \over 2^S} + {Q([A_1,A_2,A_3])\over 2^S} \\\ 2^S = a_1 \cdot {3^3 } + {Q([A_1,A_2,A_3])} \\\ 2^S - {Q([A_1,A_2,A_3])} = a_1 \cdot {3^3 } \\\ {2^S - Q([A_1,A_2,A_3]) \over 3^3} = a_1 \\\ $$ Of course, ${1\cdot 2^{A_3}-1\over 3}$ being integer means $A_3=2k_3$ is even, and given your demand that $a_4=1$ gives $a_3={4^{k_3}-1\over3} = \{1,5,21,85,...\}$
Along that line the possible values for $a_2$ and then for $a_1$ can be determined.

In an older webpage I've drawn a tree where you can identify the possible $a_1$ starting at $a_4=1$ applying $3$-times the reverse odd steps . The vector of $[A_1,A_2,A_3]$ here is surely identical to what you have found yourself, but, well, there's not much to prove here: just to determine the possible values due to simple modular conditions (such that for instance $A_3$ must be even).

The following picture is a graph excerpted from one manuscript. Numbers $a_k$ on one horizontal row transfer to the same number by one transformation. Reading the tree backwards (in direction of the arrows!) you can get an image, which numbers can be created by an inverse 3-step-transformation...

newer version of the tree

$\endgroup$
2
  • $\begingroup$ So...finding orbits with 2 odd integers is trivial. Just take $b_n=\frac{4^n-1}{3}$, which is exactly what you have listed there. It is also easy to prove that this is the only form there is for 2 odd integers in an orbit containing s and 1.The interesting part is taking it beyond that using explicit formulas without using recursive constructs. That is much harder. $\endgroup$ Nov 2, 2019 at 15:08
  • $\begingroup$ @ReverseFlow - yes, that is much harder. I've left this route when I was more engaged in the analysis of the Collatz-problem, and looked into the question of cycles instead, and to have some analyzable material discussed few-step cycles and also the so-called "1-cycle". There is much untractable there ... sigh... :) $\endgroup$ Nov 2, 2019 at 17:06
0
$\begingroup$

I found it interesting to consider the problem with the minus rather than plus sign in front of the sum: $$(\star)\qquad 3^n s - \sum_{k=0}^{n-1} 3^{n-1-k} 2^{a_k} = 2^m.$$

Consider the function $$f(x):=3x-2^{\lfloor \log_2(3x)\rfloor}.$$

It can be easily seen that iterations of $f(x)$ have 1-cycles at powers of $2$, and 2-cycles of the form $(5\cdot 2^t,7\cdot 2^t)$. I conjecture that there exist no other cycles. Under this conjecture, the question on representation $(\star)$ has an easy answer.

First, assume that iterations of $f$ starting at $x=s$ end at a power of $2$. Let $2^m$ denote this power, $n$ denote the number of iterations to reach $2^m$, and $s_0:=s\to s_1\to s_2\to \dots \to s_n:=2^m$ be the sequence of iterated values of $f$. Then $$3^n s - \sum_{k=0}^{n-1} 3^{n-1-k} (3s_k - s_{k+1}) = 2^m$$ has the form $(\star)$ since $3s_k - s_{k+1}$ are powers of $2$.

For example, for $s=456$, iterations of $f$ give $456\to 344\to 8$ gives the following identity: $$3^2 456-3\cdot 2^{10} - 2^{10} = 2^3.$$

Second, if the iterations of $f$ lead to the cycle $(5\cdot 2^t,7\cdot 2^t)$, we artificially replace it with $5\cdot 2^t\to 11\cdot 2^t\to 2^t$, and proceed as above.

For example, for $s=120$, iterations of $f$ give $120\to 104\to 56\to 40\to 56\to\dots$. We replace it with $120\to 104\to 56\to 40\to 88\to 8$, which corresponds to the identity: $$3^5 120 - 3^4 2^8 - 3^3 2^8 -3^2 2^7 - 3\cdot 2^5 - 2^8 = 2^3.$$

$\endgroup$
4
  • $\begingroup$ I think you have a typo, did you mean $2^{a_k}$ instead of $a^{a_k}$? $\endgroup$ Nov 10, 2019 at 15:45
  • $\begingroup$ @ReverseFlow: Fixed, thanks! $\endgroup$ Nov 10, 2019 at 15:47
  • $\begingroup$ Hmm, I didn't yet work through this, but isn't this just the cycle in the negative integers? (Another would be the $17,...,17$-cycle of I think seven or eleven steps, see wikipedia). Or did I overlook something trivial? $\endgroup$ Nov 17, 2019 at 20:29
  • $\begingroup$ @GottfriedHelms: I do not quite follow your comment. Could you please elaborate? $\endgroup$ Nov 17, 2019 at 21:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.