Partial Collatz Proof : Why Loop Can’t Exist

∗
Yenya
October 2024

Abstract
The Collatz Conjecture states given any positve integer, the continues application of the rules
x
3x + 1 when x is odd, and when x is even will eventually reach 1.
2
C:Z→Z

 3x + 1 if x ≡ 1 (mod 2) ,

C(x) =
x
if x ≡ 0 (mod 2) ,


2

In this paper we show that a sequence of numbers produced from these rules that repeat
themselves which I call a loop, cannot exist. The only exception is the sequence 1, 4, 2, 1 . . .. To
do this, first I define the rules for the reverse Collatz Conjecture.

C:Z→Z
x·2 if x ≡ 1 (mod 2) ,






E

C(x) = x·2 if x ≡ 0 (mod 2) , − →



 x − 1 if x ≡ 0 (mod 2) , −

 O
→
3
Next, I establish that the smallest member of the loop has to be an odd number. Afterwards, I
show when working in reverse there are three distinct groups of odd numbers. One of these groups,
which I call a Type II Odd, is a multiple of 3. This group produces a sequence in the reverse
direction that can not be reached. The other 2 groups produce a pattern in the reverse that gives
us a Type II Odd one third of the time. This pattern is formalized with the sum
n
X 2k−1
3k
k=1

The above sum determines the percentage of branches that start with a Type II Odd. When we
evaluate this sum with a limit as n → ∞, we get
n
X 2k−1
lim =1
n→∞ 3k
k=1

As n → ∞, the number of branches that start with a Type II Odd goes to 1. We take this as
meaning that loops cannot form, as there is no way for a sequence to get back to its starting value.

∗ Inspired by Dad

1
1 Definitions
1.1 Collatz Conjecture

C:Z→Z

 3x + 1 if x ≡ 1 (mod 2) ,

C(x) = x
if x ≡ 0 (mod 2) ,


2
“The 3x + 1 problem, or Collatz problem, is to prove that starting from any positive integer, some
iterate of this function takes the value to 1” [1].

1.2 Reverse Collatz Rules

C:Z→Z


 x · 2 if x ≡ 1 (mod 2) ,




 E
C(x) = x · 2 if x ≡ 0 (mod 2) , − →


x−1


 O

 if x ≡ 0 (mod 2) , − →
3
The reverse rules unlike the normal Collatz rules have three conditions. One for odd numbers, and
two for even numbers. The arrows with the letters above them indicate whether or not an output is
Even(E) or Odd(O).

1.3 Collatz Loop Proof Definitions

• Collatz Sequence -A collection of numbers that follow from each other when the Collatz or
Reverse Collatz rules are applied

Figure 1: An example of a Figure 2: An example of a sequence for

sequence for The the Collatz The Reverse Collatz Conjecture.
Conjecture where x=7 This Differs from the other example
52 because some Evens(CERS) have 3 paths

1184
34 26 448 149
112
197 592
22 17 224 144

11 296
112 72

7 49 148 56 36

74 28 18

37 14 9

2
• Collatz Loop - This is a theoretical sequence such that when you start with a positive integer,
some iterate of the Collatz function takes the value back to the starting positve integer. Such a
sequence would disprove the Collatz Conjecture. This is the case because such a sequence would
if the rules of the Collatz Conjecture are applied to the numbers in the sequence would repeat
its self, and subsequently never reach 1.

• Collatz Odd Reachable (COR) - These are positive odd integers reached from applying the
RCR(1.2).
O
Example: 28 − →9
In the example above, 9 would be a COR. The arrow with the O above it helps differentiates
the two possible operations possible for even numbers when using the RCR(1.2). Of the two it
is the operation that produces an odd number.
We can see below that with the normal Collatz rules the opposite sequence can be attained.

9 → 28

• Collatz Even Reachable (CER) - These are positive even integers reached from applying the
RCR(1.2), and more specifically, positive even integers that have a valid COR.
E
Example: 7 → 14 −
→ 28

In the example above, 28 would be considered a CER, whereas 14 would not because it does not
13
have a valid COR. When applying the RCR(1.2), 14 produces which is not an integer. On
3
the Other hand, applying RCR(1.2) to 28 yields 9.
We can see below that with the normal Collatz rules the opposite sequence can be attained.

28 → 14 → 7

• Collatz Even Unreachable (CEU) - These are positive even integers reached from applying
the RCR(1.2) that do not have a valid COR.
E
Example: 11 → 22 −
→ 44

In the above example, 44 would be considered a CEU whereas 22 would not. The logic is the
same as in the example for CERs. We see, 22 would be considered a CER because its output 7
43
is a valid COR. Conversely, 44 has an output , which is not a valid COR because it is not an
3
integer. Therefore, 44 fits the definition of a CEU.
We can see below that with the normal Collatz rules the opposite sequence can be attained.

2811 → 44 → 22 → 7

• Branch - Branches exist only for the Reverse Collatz Conjecture. An example of this can be
found in Figure 2, for the sequence starting with 7. Immediate results would be considered to
be a part of the branch. So, CORS such as 9, 37, and 149 would be a part of the 7 branch.
Furthermore 9,37, and 149 are also considered to have their own branch.
a) Main Branch - From Figure 2, the branch starting with 7 would be considered the main
branch. Whereas the ones starting with 9,37, and 149 would not be considered Main
Branches. This is because the branch starting with 7 is the first branch we consider.

3
2 Basic Proof Criteria
1. We are looking for a Collatz sequence that fits the definition of a Collatz Loop. Such a sequence
would disprove the Collatz Conjecture as long as 1 is not a member of the sequence.
2. If such a sequence does exist the smallest member of said sequence has to be an odd number.
This follows from the definition of the conjecture itself. Even numbers produce a number smaller
than themselves. Whereas odd number produce a number greater than themselves.

3. There are 3 types of odd numbers in the Collatz world.

(a) Type I - Odds that satisfy the equation 1 + 6m, m ∈ Z+
0
(b) Type II - Odds that satisfy the equation 3 + 6m, m ∈ Z+
0
(c) Type III - Odds that satisfy the equation 5 + 6m, m ∈ Z+
0

4. When working backwards and applying the Reverse Collatz Rules(1.2) Type II Odds produce
a sequence that has zero CORs to be more precise, a sequence consisting of CEUs
(a) An Example of this is 3 → 6 → 12 . . .
(b) The starting number in this example of a Type II Odd is a multiple of 3. This fact comes
from the definition of Type II Odds. Proceeding numbers when applying the RCR(1.2) will
be even multiples of three. We can see that the x · 2 rule for evens produces an integer. This
x−1
is not the case when applying the rule. For even multiples of 3, this rule produces
3
non-integer values. CEUs are defined as even numbers that do not have valid CORs. We
can see from our description of the sequences of Type II Odds that all the members of the
sequences are even numbers with non-integer CORs, or in other words invalid CORs. This
is not the case for the other Type of Odds, because they don’t produce even multiples of
three when applying the RCR(1.2).

3 Equations for CORs, CERs, and DCORs

1. The general formulas for CORs and CERs are

2n (x + 6m) − 1 2n x − 1
COR = = + 2n+1 m
3 3

CER = 2n (x + 6m)

n ∈ Z+ , repersents the number of times the x · 2 rule has been applied.

m ∈ Z+ 0 , helps define our starting value.
x = 1, 3, 5 is used as a placeholder for the three starting odds.
2. The Different Types materialize CORs at different values of n. We show this by using the
equation for CORs, and by substituting the different values for x
(a) Type I :
x=1
n
2 −1
+ 2n+1 m
3
Valid CORs materialize at even values of n

n = 2, 4, 6, 8, . . .

n = 2k, k ∈ Z+

4
 (b) Type II :
x=3
1
2n + 2n+1 m −
3
This will never produce a valid COR. From the equation above, it is clear that we will
always produce a fractional value because the variables m and n are integers. Since CORs
have to be integers, values produced by this equation are not valid.
(c) Type III :
x=5
2n · 5 − 1
+ 2n+1 m
3
Valid CORs materialize at odd values of n

n = 1, 3, 5, 7, . . .

n = 2k + 1, k ∈ Z+
0

3. The Difference between subsequent CORs is equal to the value of the CER of the smaller COR.
2n (x + 6m) − 1
COR1 =
3

2n+2 (x + 6m) − 1
COR2 =
3
As we can see from the above equation and from the findings of 3.2.a and 3.2.c, valid CORs
materialize at n values that are +2 apart. This is observer in the pattern that emerges in
Figure 2.

2n+2 (x + 6m) − 1 2n (x + 6m) − 1

DCOR = COR2 − COR1 = − = 2n (x + 6m)
3 3
As we can see from the above equation, DCOR is equivalent to the smaller CER. Taking and
example from Figure 2 the difference between two valid CORs 9 and 37 is 28 which is the CER of
the smaller COR, 9. This pattern continues for the main branch, and for branches not starting
with a Type II Odd. This is shown with the branch starting with 37 which is not a Type II
Odd. The fact that Type II Odds do not produce this pattern is also shown in Figure 2. The
branch starting with 9 which is a Type II Odd has no valid CORs, and subsequently all of its
members are CEUs.

4 Odd Type Difference Categorization

1. The three groups of Odd Types have a difference of 2 and 4. Members of the same group have
a difference of 6. This can be formalized in the following way.
(a) Odd Type Difference 1(OTD I)

OT D I : Xp = 2 + 6p, p ∈ Z+
0 , Xp ≡ 2 (mod 6)

The members of this group when added to any of the odd Types will result in an odd that
is 1 Type away.
Example: 1(Type I) + 2(OTD I) = 3(Type II)
(b) Odd Type Difference 2(OTD II)

OT D II : Xp = 4 + 6p, p ∈ Z+
0 , Xp ≡ 4 (mod 6)

The members of this group when added to any of the odd Types will result in an odd that
is 2 Types away.
Example: 1(Type I) + 4(OTD II) = 5(Type III)

5
 (c) Odd Type Difference 3(OTD III)

OT D III : Xp = 6p, p ∈ Z+ , Xp ≡ 0 (mod 6)

The members of this group when added to any of the Odd Types will result in an odd of
the same Type.

Given the above equations we can group all even numbers into one of the OTD’s.
2. All valid DCORs are in the group OTD II

DCOR = 2n (x + 6m)

When determining an OTD group we know the most significant parts are essentially what is left
over after dividing by 6. So, we can rewrite the above equation as

2n x

(a) Type I :
x=1
n = 2k, k ∈ Z+
4k = 4 + 6p, p ∈ Z+
0

4k ≡ 4 (mod 6)
The above is equivalent to the definition of OTD II.
(b) Type II :
x=3
N o valid CORS
(c) Type III :
x=5
n = 2k + 1, k ∈ Z+
0

10 · 4k = (4 + 6)4k = 4K+1 + 6 · 4k
4k+1 = 4 + 6p, p ∈ Z+
0

4 + 6p + 6(4 + 6p) = 4 + 6p
10 · 4k ≡ 4 (mod 6)
The above is equivalent to the definition of OTD II.

3. Given that all DCORs are in group OTD II, and by the definition of OTD II, all subsequent
CORs are in Odd Types that are two apart ( TYPE I → TYPE III and TYPE III → TYPE II
and TYPE II → TYPE I).
This implies that given a starting odd number when we apply the RCR(1.2), subsequent CORs
will be of different Odd Type following a specified pattern.

6
5 Conclusions
1. From 2.3, we saw that there are 3 Types of Odds with Corresponding equations.
(a) Type I - Odds that satisfy the equation 1 + 6m, m ∈ Z+
0
(b) Type II - Odds that satisfy the equation 3 + 6m, m ∈ Z+0
(c) Type III - Odds that satisfy the equation 5 + 6m, m ∈ Z+ 0

2. In 2.4 and 3.2.b, I showed that Type II Odds create branches that are unreachable, or in other
words, branches consisting of CEUs.
3. In 3.3, I defined the equation for a DCOR
DCOR = 2n (x + 6m)
n ∈ Z+
m ∈ Z+ 0
x = 1, 3, 5

4. In 4.1, I defined the three groups of possible differences between the three Odd Types, which are
referred to as Odd Type Difference.
(a) Odd Type Difference 1(OTD I)
OT D I : Xp = 2 + 6p, p ∈ Z+
0 , Xp ≡ 2 (mod 6)

The members of this group, when added to any of the odd Types, will result in an odd that
is 1 Type away.

(b) Odd Type Difference 2(OTD II)

OT D II : Xp = 4 + 6p, p ∈ Z+
0 , Xp ≡ 4 (mod 6)

The members of this group, when added to any of the odd Types, will result in an odd that
is 2 Types away.

(c) Odd Type Difference 2(OTD III)

OT D III : Xp = 6p, p ∈ Z+ , Xp ≡ 0 (mod 6)
The members of this group, when added to any of the Odd Types, will result in an odd of
the same Type.
5. In 4.2, I showed that all valid DCORs are in Odd Type Difference 2(OTD II). This implies
that subsequent CORs are differnt Odd Types. The change from one Odd Type to another
follows a pattern, this pattern can be found below

( TYPE I → TYPE III and TYPE III → TYPE II and TYPE II → TYPE I)

6. Given that the COR Type changes every single time, a third of the Cors in the main branch are
observed to be of Type II, meaning that they are unreachable.
2
7. For the of the CORs that are not Type II Odds we know that those branches will also produce
3
1
the same pattern. Meaning of their CORs will be Type II Odds. We can formalize this with
3
the sum
n
X 2k−1
n ∈ Z+
3k
k=1
n represents the number of branches we are considering

The above sum gives the percentage of CORs that are Type II Odds.

7
8. The limit of this sum as n → ∞ is
n
X 2k−1
lim =1
n→∞ 3k
k=1

From this, we conclude that all branches will eventually terminate with a COR that is a Type
II Odd.

8
References
[1] Jeffrey C. Lagarias. The 3x+1 problem: An annotated bibliography (1963–1999) (sorted by author),
2011.