Matrix-Based Representation of the Weak Collatz Conjecture Sequence

Mohlomi Cliff Makhetha

May 2024

Abstract
The Collatz conjecture posits that any sequence generated by a specific iterative transformation on a positive integer will ultimately
converge to 1. The weak Collatz conjecture modifies this rule to ensure the sequence terminates when it falls below its initial value.
This paper introduces a novel matrix-based representation of the weak Collatz sequence, providing a structured analytical method.
By translating sequence transformations into matrix operations, we offer a visual and mathematical framework for understanding
the conjecture’s implications, demonstrating its potential validity through finite sequence analysis.

1
1 Introduction
The Collatz conjecture, also known as the 3x + 1 conjecture, explores the behavior of a transformation applied iteratively to positive
integers. For any positive integer Ti , the sequence progresses according to the following rule:

3Ti + 1 Ti is odd
Ti+1 = (1)
Ti /2 Ti is even
This conjecture asserts that regardless of the initial value T0 the sequence will invariably reach 1, subsequently entering a perpetual
cycle of 4, 2, 1, ....

1.1 Example 1
Starting with T0 = 52 we find the sequence:

52, 26, 13, 40, 20, 10, 5, 16, 8, 4, 2, 1, 4, 2, 1, . . .

While starting with T0 = 27 we find the sequence:

27, 82, 41, 124, 62, 31, 94, 47, 142, 71, 214, 107, 322, 161, 484, 242, 121, 364, 182, 91, 274, 137, 412, 206, 103,

310, 155, 466, 233, 700, 350, 175, 526, 263, 790, 395, 1186, 593, 1780, 890, 445, 1336, 668, 334, 167, 502, 251, 754,
377, 1132, 566, 283, 850, 425, 1276, 638, 319, 958, 479, 1438, 719, 2158, 1079, 3238, 1619, 4858, 2429, 7288, 3644,
1822, 911, 2734, 1367, 4102, 2051, 6154, 3077, 9232, 4616, 2308, 1154, 577, 1732, 866, 433, 1300, 650, 325, 976, 488,
244, 122, 61, 184, 92, 46, 23, 70, 35, 106, 53, 160, 80, 40, 20, 10, 5, 16, 8, 4, 2, 1, 4, 2, 1, ...

Example 1 illustrate that despite the variance in sequence lengths both sequences confirm the conjecture by ultimately converging
to the loop 4, 2, 1, ....

Building on the framework of the Collatz sequence as defined in Equation (1), the weak Collatz conjecture introduces a modification
to ensure the sequence terminates when Ti < T0 . The sequence now called the weak Collatz sequence is defined as:

 3Ti + 1 Ti is odd and Ti > T0
Ti+1 = Ti /2 Ti is even and Ti > T0 (2)
stop Ti < T0


The weak Collatz conjecture posits that for any starting integer T0 > 1, there exists an Ti term at which Ti < T0 . This condition
introduces a finite boundary to the sequence, different from the infinite cycle proposed by the strong conjecture.

1.2 Example 2
Starting with T0 = 51 we find the sequence:
51, 154, 77, 232, 116, 58, 29
While starting with T0 = 31 we find the sequence:

31, 94, 47, 142, 71, 214, 107, 322, 161, 484, 242, 121, 364, 182, 91, 274, 137, 412, 206, 103, 310, 155, 466, 233,

700, 350, 175, 526, 263, 790, 395, 1186, 593, 1780, 890, 445, 1336, 668, 334, 167, 502, 251, 754, 377, 1132, 566, 283,
850, 425, 1276, 638, 319, 958, 479, 1438, 719, 2158, 1079, 3238, 1619, 4858, 2429, 7288, 3644, 1822, 911, 2734, 1367,
4102, 2051, 6154, 3077, 9232, 4616, 2308, 1154, 577, 1732, 866, 433, 1300, 650, 325, 976, 488, 244, 122, 61, 184, 92, 46, 23

In each case in Example 2, the sequences validate the weak conjecture by eventually reaching a term less than the initial value,
specifically, 29 < 51 and 23 < 31.

Implication Proof by Argument: Connection Between the Weak and Strong Collatz Conjectures
To establish the connection between the weak and strong Collatz conjectures, lets assume the weak Collatz conjecture holds true,
which posits for any starting integer T0 > 1, the sequence defined by Equation (2) will eventually reach a term Ti that is less
than T0 . Now lets consider a sequence starting at any integer T0 > 1, which will, according to the weak conjecture, diminish until
a smaller integer Ti < T0 is reached. If Ti remains above 1, the weak conjecture can be reapplied n times, resulting in further
reduction until Ti+n < Ti is achieved. This process iterates until the sequence cannot decrease further, culminating at the integer
1.

This recurring reduction forms an overarching downward trajectory in the sequence, characterized by fluctuations that nonetheless
conform to a general trend of decrease. Although individual segments of the sequence might exhibit temporary increases, the essen-
tial nature of the sequence under the Weak Collatz Conjecture ensures that it will eventually reduce to the smallest positive integer

2
1. Once 1 is reached, the sequence predictably follows the infinite cyclic behavior of 4, 2, 1, as established by the Collatz Conjecture.

Thus the connection between the weak and strong Collatz Conjectures is such that if the week Collatz Conjecture
hold true, by implication so does the strong Collatz Conjecture.

2 Matrix-Based Representation of the Weak Collatz Conjecture Sequence

To introduce a matrix-based representation for the weak Collatz conjecture sequence, we first define a mapping for the initial term
of the sequence, T0 , as a function of k ∈ N+ :

T0 : k → ak + b where a ∈ N+ , b ∈ Z, and T0 > 1.

For an odd T0 , the next term in the sequence according to Equation (2) T1 , can be obtained by:

T1 = 3T0 + 1 = 3(ak + b) + 1 = (3a)k + (3b + 1).

Thus, the mappings can be expressed as:

T0 : k → ak + b,
T1 : k → (3a)k + (3b + 1).
The two terms can be represented using matrices:
    
T0 a b k
:k→ ,
T1 3a 3b + 1 1
 
a b
And if we let M1 = which we can rewritten as:
3a 3b + 1
    
1 0 a b 0 0
M1 = + thus
0 3 a b 0 1
   
1 0 0 0
M1 = M0 + · · · case 1
0 3 0 1
 
a b
Where M0 = and case 1 represents the matrix transformation from T0 to T1 for an odd T0 .
a b

On the other hand if T0 is even, then by the definition of Equation (2);

T1 = T0 /2 = (ak + b)/2 = (a/2)k + b/2

Thus, the mappings can be expressed as:

T0 : k → ak + b,
T1 : k → (a/2)k + b/2,
 
a b
Yet again the two terms can be represented using matrices M1 = where one can also represent the matrix M1 as:
a/2 b/2
  
1 0 a b
M1 = thus
0 1/2 a b
 
1 0
M1 = M0 · · · case 2
0 1/2
 
a b
Where M0 = and case 2 represents the matrix transformation from T0 to T1 for an even T0 .
a b
   
a b a b
Now if we allow ourselves to generalize Mi = such that M0 = then using both case 1 and 2 we can formulate
c d a b
a matrix-based representation of Equation (2):
    
1 0 0 0
M + if c is even, d is odd, a ≤ c and |b| ≤ |d|

i

0 3 0 1






  
Mi+1 = 1 0 (3)
Mi if c is even, d is even, a ≤ c and |b| ≤ |d|
0 1/2








stop if a > c and |b| > |d|


3
 2.1 Example 3
 
1024 −449
Starting with M0 = using Equation (3) one finds:
1024 −449
           
1024 −449 1024 −449 1024 −449 1024 −449 1024 −449 1024 −449
···
1024 −449 3072 −1346 1536 −673 4608 −2018 2304 −1009 729 −319

Equation (3) is the matrix representation of Equation (2) it translates the iterative application described by Equation (2) into
matrix operations that encapsulate these transformations including the the condition to stop the iterative process if Ti < T0 this
can be demonstrated in Example 4 below.

2.2 Example 4
 
1024 −449
Lets consider M16 = from Example 3, since the matrix is derived from Equations (3) then there exists in-
729 −319
finite sequences with initial integer T0 = 1024k − 449 and after subjecting T0 to Equation (2) 15 times, we get the term
T16 = 729k − 319 : T16 < T0 ∀k ∈ N+ , this can be demonstrated for a few k ∈ N+ value sequences:

• For k = 7 we get T0 = 6719 and after subjecting T0 to Equation (2) 15 times, we get the term T16 = 4784 in the finite
sequence
6719, 20158, 10079, 30238, 15119, 45358, 22679, 68038, 34019, 102058, 51029, ..., 4784

• For k = 52 we get T0 = 52799 and after subjecting T0 to Equation (2) 15 times, we get the term T16 = 37589 in the finite
sequence
52799, 158398, 79199, 237598, 118799, 356398, 178199, 534598, 267299, 801898, 400949, ..., 37589

• For k = 113 we get T0 = 115263 and after subjecting T0 to Equation (2) 15 times, we get the term T16 = 82058 in the
finite sequence

115263, 345790, 172895, 518686, 259343, 778030, 389015, 1167046, 583523, 1750570, 875285, ..., 82058

However Equation (3) is incomplete since the ith term Mi depends on the parity of the second rows elements c and d, this implies
there are two more conditions unaccounted for in Equation (3) since the possible parity combinations are:

c Parity d Parity State

Even Odd Accounted
Even Even Accounted
Odd Even Unaccounted
Odd Odd Unaccounted
 
a b
Thus given a matrix Mi = , where c is odd and d is either even or odd, one will find that Equation (3) can not be applied
c d
to Mi as a result that the weak Collatz Equation (2) is not directly applicable to Ti : k → ck + d for Ti exhibits distinct patterns
of parity behaviour when c is odd and d is even then:
• Ti is an even integer ∀k is even.
• Ti is an odd integer ∀k is odd.
While Ti exhibits a distinct patterns of parity behavior when c is odd and d is odd then:
• Ti is an even integer ∀k is odd.
• Ti is an odd integer ∀k is even.
To rigorously address this disparity, we introduce the concept of partitioning. We define two disjoint mappings Ti,0 and Ti,1 with
in the maping of Ti such that in the case when c is odd and d is even:
(
Ti,0 : k → c(2k) + d
Ti : k →
Ti,1 : k → c(2k − 1) + d

And partitioning such that in the case when c is odd and d is odd:
(
Ti,0 : k → c(2k − 1) + d
Ti : k →
Ti,1 : k → c(2k) + d

It is clear that in both cases:

Ti = (Ti,0 ⊎ Ti,1 ) = (Ti,0 ∪ Ti,1 ) − (Ti,0 ∩ Ti,1 ) (4)

4
Thus the partitioning of Ti is disjoint and is based on the parity behavior of the coefficient and constant terms of Ti ∀k ∈ N+ . Due
to the shared parameter k between T0 and Ti , we extend the partitioning directly to T0 , and define two disjoint discrete functions
T0,0 and T0,1 such that in the case when c is odd and d is even:

T0,0 = T0 [2k]
T0,1 = T0 [2k − 1]

And partitioning such that in the case when c is odd and d is odd:

T0,0 = T0 [2k − 1]
T0,1 = T0 [2k]

Once again in both cases:

T0 = (T0,0 ⊎ T0,1 ) = (T0,0 ∪ T0,1 ) − (T0,0 ∩ T0,1 ) (5)
Thus the partitioning of T0 is disjoint and is based on the parity behavior of the coefficient and constant terms of Ti ∀k ∈ N+ . The
partitioning identities as in Equations (4) and (5) can be expressed as a single matrix-based representation in the following way,
given;
T0 = (T0,0 ∪ T0,1 ) − (T0,0 ∩ T0,1 )
Ti = (Ti,0 ∪ Ti,1 ) − (Ti,0 ∩ Ti,1 )
Thus in the case when c is odd and d is even, we can write:

T0 = (T0 [2k] ∪ T0 [2k − 1]) − (T0 [2k] ∩ T0 [2k − 1])

Ti = (Ti [2k] ∪ Ti [2k − 1]) − (Ti [2k] ∩ Ti [2k−])

And since T0 : k → ak + b while Ti : k → ck + d we can substitute T0 and Ti to find;

ak + b = ((a[2k] + b) ∪ (a[2k − 1] + b)) − ((a[2k] + b) ∩ (a[2k − 1] + b))

ck + d = ((c[2k] + d) ∪ (c[2k − 1] + d)) − ((c[2k] + d) ∩ (c[2k − 1] + b))

One can further simplify to get;

ak + b = (([2a]k + b) ∪ ([2a]k + [b − a])) − (([2a]k + b) ∩ ([2a]k + [b − a]))

ck + d = (([2c]k + d) ∪ ([2c]k + [d − c])) − (([2c]k + d) ∩ ([2c]k + [d − c]))

And by using two-by-two matrix made of the coefficients and constant terms of the two Equations, one finds;
         
a b 2a b 2a (b − a) 2a b 2a (b − a)
= ∪ − ∩
c d 2c d 2c (d − c) 2c d 2c (d − c)

One can also represent the matrix-based representation as;

             
a b a b 2 0 a b 2 −1 a b 2 0 a b 2 −1
= ∪ − ∩
c d c d 0 1 c d 0 1 c d 0 1 c d 0 1

Which can be simplified to get;

         
2 0 2 −1 2 0 2 −1
Mi = M i ∪ Mi − Mi ∩ Mi (6)
0 1 0 1 0 1 0 1

And since the union and intersection operations are commutative that is A ∪ B = B ∪ A and A ∩ B = B ∩ A, then it can be easily
shown that Equation (6) in the case when c is odd and d is odd still holds true as a single matrix-based representation for the
partitioning identities Equations (4) and (5). Feature more since Equation (6) expresses how each matrix Mi can be decomposed
into a union of transformed matrices while ensuring the transformations are applied disjointly. Then Equation (6) allows for the
extension of Equation (3) resulting in the complete matrix-based representation of the weak Collatz conjecture sequence as:
    
1 0 0 0
M(i, j) + if c is even, d is odd, a ≤ c and |b| ≤ |d|


0 3 0 1








  
 1 0
M(i, j) if c is even, d is even, a ≤ c and |b| ≤ |d|


0 1/2






  
M(i+1, j) = 2 0 (7)
M
 (i, j) if c is odd, d is even, a ≤ c and |b| ≤ |d|


 0 1



  
2 −1


M if c is odd, d is odd, a ≤ c and |b| ≤ |d|

(i, j)

0 1








stop if a > c and |b| > |d|


5
   
2 −1
M if c is odd, d is even, a ≤ c and |b| ≤ |d|

 (i, j)

0 1


M(1, j+1) =   (8)
2 0


 M(i, if c is odd, d is odd, a ≤ c and |b| ≤ |d|

j)

0 1
 
a b
Where M(i, j) = and Equation (7) and (8) totherger describe that following array of matrices:
c d
M(0, 0) M(1, 0) M(2, 0) ··· M(i, 0)
M(0, 1) M(1, 1) M(2, 1) ··· M(i, 1)
M(0, 2) M(1, 2) M(2, 2) ··· M(i, 2) (9)
.. .. .. .. ..
. . . . .
M(0, j) M(1, j) M(2, j) ··· M(i, j)

Where the elements of the matrix M(i, j) in the first row are the coefficient and constant terms of a discrete function describing the
initial term T0 of the weak Collatz conjecture sequence defined in Equation (2) thus:

T0 : k → ak + b (10)

While the elements of the matrix M(i, j) in the second row are the coefficient and constant terms of a discrete function describing
the term Ti after subjecting T0 some nth iterations of the weak Collatz sequence defined in Equation (2) we get:

Ti : k → ck + d (11)

2.3 Example 5
 
15 2
Starting with M(0, 0) = we find the array:
15 2
     
15 2 30 2 30 2
15 2 30 2 15 1
         
30 −13 30 −13 30 −13 60 −43 60 −43
30 −13 90 −38 45 −19 90 −64 45 −32
             
60 −13 60 −13 60 −13 120 −13 120 −13 240 −13 240 −13
90 −19 270 −56 135 −28 270 −28 135 −14 270 −14 135 −7
             
120 −73 120 −73 120 −73 240 −73 240 −73 480 −73 480 −73
270 −163 810 −488 405 −244 810 −244 405 −122 810 −122 405 −61
         
240 −133 240 −133 240 −133 480 −373 480 −373
270 −149 810 −446 405 −223 810 −628 405 −314
     
240 −193 240 −193 240 −193
M(3, 5) M(4, 5) M(5, 5) ··· M(i, 5)
810 −649 2430 −1946 1215 −973
.. .. .. .. .. .. .. ..
. . . . . . . .
M(0, j) M(1, j) M(2, j) M(3, j) M(4, j) M(5, j) ··· M(i, j)

Figure 1 below is a graphical depiction that helps in visualizing the mathematical transformations that the array in Example 5
compactly lists. While the figure gives a step-by-step visual flow of the transformations, the array provides a straightforward and
immediate reference to each state of the matrix through the sequence.

6
 Figure 1: Graphical Representation of Example 5

Equation (7) and (8) form the complete matrix representation of Equation (2) it translates the iterative application described by
Equation (2) into matrix operations that encapsulate these transformations including the the condition to stop the iterative process
if Ti < T0 this can be demonstrated in Example 6 below.

2.4 Example 6
 
1920 −1093
Lets consider M(6, =
7) from example 5, since the matrix is derived from Equations (7) and (8) then there
1215 −691
exist infinitely finite sequences with initial integer T0 = 1920k − 1093 and after subjecting T0 to Equation (2) 11 times, we get
the term T12 = 1215k − 691 : T12 < T0 ∀k ∈ N+ , this can be demonstrated for a few k value sequences:

• For k = 1 we get T0 = 827 and after subjecting T0 to Equation (2) 11 times, we get the term T12 = 524 in the sequence

827, 12482, 1241, 3724, 1862, 931, 2794, 1397, 4192, 2096, 1048, 524

• For k = 5 we get T0 = 8507 and after subjecting T0 to Equation (2) 11 times, we get the term T12 = 5384 in the sequence

8507, 25522, 12761, 38284, 19142, 9571, 28714, 14357, 43072, 21536, 10768, 5384

• For k = 20 we get T0 = 37307 and after subjecting T0 to Equation (2) 11 times, we get the term T12 = 23609 in the
sequence
37307, 111922, 55961, 167884, 83942, 41971, 125914, 62957, 188872, 94436, 47218, 23609

3 Implications of Matrix-Based Representation of the Weak Collatz Conjecture

Sequence
The implications of the matrix-based representation of the weak Collatz conjecture sequence are profound, as they provide a struc-
tured method to analyze and understand the behavior of the sequence. Here is the key implications:

Finite Sequences: The matrix-based representation yields an array as in Equation(9) based on the iterative process of the
Equations (7) and (8). If it can be shown that every row of this array contains finite columns that meet the stopping condition, it
suggests that the weak Collatz conjecture holds true. This is because the sequence will eventually reach a term Ti that is less than
the initial term T0 , thus terminating the sequence. By demonstrating that every row in the array has finite columns, we can define
a set M that contains every last entry matrix of each row:
∞
]
M= {Mi,j | a ≤ c ∧ |b| ≤ |d|} (12)
j=0

This disjoint set union M represents the collection of all matrices Mi,j that satisfy the conditions a ≤ c ∧ |b| ≤ |d|. Since equation
(6) and the iterative processes in equations (7) and (8) ensure finite columns in each row of the array, it follows that every element
of T0 generated by the initial matrix M0,0 will be part of the set M.

T0 ∈ M(0, 0) = T0 ∈ M
 
1 0
This implies that if M(0, 0) = , then the weak Collatz conjecture holds true ∀T0 ∈ N+ .
1 0

7
 Appendix
1 import numpy as np
2 import sys
3
4 # Initialize constants for matrix values
5 a , b , c , d = 1 ,0 ,1 ,0
6
7 # Array to store matrices
8 array_rows = []
9 array_rows . append ( np . array ([[ a , b ] , [c , d ]]) ) # Starting matrix
10
11 def iterate_fun_1 (M , i , j ) :
12 a , b = M [0 , 0] , M [0 , 1]
13 c , d = M [1 , 0] , M [1 , 1]
14 print ( str ( M ) , " ... M_ ({} , {}) " . format (i , j ) )
15
16 # Transf ormatio n rules based on matrix elements ( similar to weak Collatz conjecture conditions )
17 if c % 2 == 0 and d % 2 == 1 and a <= c and abs ( b ) <= abs ( d ) :
18 M_new = np . array ([[1 , 0] , [0 , 3]]) @ M + np . array ([[0 , 0] , [0 , 1]])
19 elif c % 2 == 0 and d % 2 == 0 and a <= c and abs ( b ) <= abs ( d ) :
20 M_new = np . array ([[1 , 0] , [0 , 1 / 2]]) @ M
21 elif c % 2 == 1 and d % 2 == 0 and a <= c and abs ( b ) <= abs ( d ) :
22 M_new = M @ np . array ([[2 , 0] , [0 , 1]])
23 array_rows . append ( iterate_fun_2 ( M ) )
24 elif c % 2 == 1 and d % 2 == 1 and a <= c and abs ( b ) <= abs ( d ) :
25 M_new = M @ np . array ([[2 , -1] , [0 , 1]])
26 array_rows . append ( iterate_fun_2 ( M ) )
27 else :
28 return
29
30 i += 1
31 iterate_fun_1 ( M_new , i , j )
32
33 def iterate_fun_2 ( M ) :
34 a , b = M [0 , 0] , M [0 , 1]
35 c , d = M [1 , 0] , M [1 , 1]
36
37 # Alternate tr an s fo rm at i on s based on matrix parity
38 if c % 2 == 1 and d % 2 == 0 and a <= c and abs ( b ) <= abs ( d ) :
39 M_new = M @ np . array ([[2 , -1] , [0 , 1]])
40 else :
41 M_new = M @ np . array ([[2 , 0] , [0 , 1]])
42 return M_new
43
44 # Main loop to process each matrix transfo rmation
45 j = 0
46 while array_rows :
47 iterate_fun_1 ( array_rows . pop (0) , 0 , j )
48 print ()
49 j += 1
Listing 1: Python Code Matrix Representation of the Weak Collatz Conjec- ture