Cylinder matrices

Paulus Gerdes


pgerdes@virconn.com



 
 
 

Abstract: In this paper I will introduce the concepts of horizontal and vertical, positive and negative cylinder matrices of period p, and present results concerning the multiplication of cylinder matrices. Various products of cylinder matrices are cycle matrices or helix matrices.


 

Introduction

Figure 1 presents an example of a horizontal cylinder matrix of dimensions 6×4. Its six matrix elements are 0, 1, 2, 3, 4, and 5.

0
1
3
4
5
5
2
2
4
3
1
0
1
0
4
3
2
2
5
5
3
4
0
1

Example of a horizontal cylinder matrix

Figure 1

The particular structure of the matrix may be understood if we join horizontally several copies of it. The numbers 4, 5, and 3 are recurring along a horizontal zigzag line (Figure 2).

Figure 2

Likewise the numbers 0, 1, and 2 are recurring along another horizontal zigzag line (Figure 3).

Figure 3

Imagine the matrix wrapped around a cylinder. Each horizontal zigzag line becomes a closed loop. Figure 4 presents the flat image of these loops. The matrix is composed of two loops. Each loop has period 3. The matrix will be called a horizontal cylinder matrix of period 3.

 

First loop

(a)

Second loop

(b)

Both loops

(c)

Figure 4

The matrix in Figure 5 is slightly different from the matrix in Figure 1: The first column of the matrix in Figure 1 became the last column in the matrix in Figure 5.

1
3
4
0
5
2
2
5
3
1
0
4
0
4
3
1
2
5
5
2
4
0
1
3

Figure 5

Figure 6 presents the loops of the new matrix. Each loop continues to have period 3, but their position is displaced by one unitsquare.

First loop

(a)

Second loop

(b)

Both loops

(c)

Figure 6

To distinguish the two possible positions, we will call the matrix in Figure 1 a negative horizontal 6×4 cylinder matrix of period 3, and the matrix in Figure 5 a positive horizontal 6×4 cylinder matrix of period 3. The choice of the terms "negative" and "positive" will be justified by the analysis of the properties of the multiplication of cylinder matrices.

Figure 7 presents an example of a positive vertical cylinder matrix of period 3. A vertical cylinder matrix is the transpose of a horizontal cylinder matrix.
 

a
b

Example of a positive vertical cylinder matrix of period 3

Figure 7





Horizontal zigzag loops

Let A be a matrix of dimensions (2m)×(2n). The first positive, horizontal loop (L1) may be characterised as follows:

(1) a11ÎL1,

(2) if aijÎL1, i+j even, i = 1, …, 2m-1, j = 1, …,2n-1, then a(i+1)(j+1)ÎL1,

(3) if aijÎL1, i+j odd, i = 2,…,2m, j = 1,…,2n-1, then a(i-1)(j+1)ÎL1,

(4) if a(2m)(j)ÎL1, j even, j = 2,…,2n-2, then a(2m)(j+1)ÎL1,

(5) if a1jÎL1, j even, j = 2,…, 2n-2, then a(1)(j+1)ÎL1,

(6) if a(i)(2n)ÎL1, i even, i = 2,.., 2m-2, then a(i+1)(1)ÎL1,

(7) if a(i)(2n)ÎL1, i odd, i = 3,…, 2m-1, then a(i-1)(1)ÎL1,

(8) if a(2m)(2n)ÎL1, then a(2m)(1) ÎL1.

The other positive, horizontal loops and the negative, horizontal loops may be defined analogously.

The following theorem about the number of different negative, horizontal loops may be proven:

Theorem 1: The number of negative, horizontal loops of a matrix of dimensions (2m)×(2n) is equal to the greatest common divisor of 2m and n.

The same holds for the number of positive, horizontal loops.
 
 

Cylinder matrices of period p

A matrix of dimensions (2m)×(2n) will be called a positive, horizontal cylinder matrix of period p if all its positive horizontal loops have period p.

Negative horizontal, positive vertical and negative vertical cylinder matrices of period p may be defined analogously.

The number of matrix elements in each loop should be a multiple of p. As the total number of loops is gcd(2m,n) and the total number of matrix elements is 4mn, it follows that p has to be a divisor of 4mn/[gcd(2m,n)].
 
 

Multiplication of horizontal cylinder matrices

Figure 8 presents examples of the multiplication of horizontal cylinder matrices of dimensions 6×4 and dimensions 4×12. The product matrices of dimensions 6×12 are composed of six loops. Like the multiplicands, the product matrices have period 3.

3
0
4
5
 
4
0
2
4
3
2
5
3
1
5
0
1
 
52
12
20
52
24
20
56
24
16
56
12
16
6
2
2
6
 
5
3
1
5
0
1
4
0
2
4
3
2
 
68
12
28
68
36
28
76
36
20
76
12
20
0
3
5
4
 
5
3
1
5
0
1
4
0
2
4
3
2
 
56
24
16
56
12
16
52
12
20
52
24
20
5
4
0
3
 
4
0
2
4
3
2
5
3
1
5
0
1
 
52
12
20
52
24
20
56
24
16
56
12
16
2
6
6
2
                           
76
36
20
76
12
20
68
12
28
68
36
28
4
5
3
0
                           
56
24
16
56
12
16
52
12
20
52
24
20
positive
positive
positive

(a)


1
2
0
4
 
4
0
2
4
3
2
5
3
1
5
0
1
 
30
6
12
30
15
12
33
15
9
33
6
9
5
5
3
3
 
5
3
1
5
0
1
4
0
2
4
3
2
 
72
24
24
72
24
24
72
24
24
72
24
24
4
0
2
1
 
5
3
1
5
0
1
4
0
2
4
3
2
 
30
6
12
30
15
12
33
15
9
33
6
9
2
1
4
0
 
4
0
2
4
3
2
5
3
1
5
0
1
 
33
15
9
33
6
9
30
6
12
30
15
12
3
3
5
5
                           
72
24
24
72
24
24
72
24
24
72
24
24
0
4
1
2
                           
33
15
9
33
6
9
30
6
12
30
15
12
negative
positive
negative

(b)


3
0
4
5
 
3
2
-1
4
2
5
4
0
5
3
0
-1
 
40
16
12
44
16
36
44
8
36
40
8
12
6
2
2
6
 
4
0
5
3
0
-1
3
2
-1
4
2
5
 
52
24
8
60
24
56
60
8
56
52
8
8
0
3
5
4
 
4
0
5
3
0
-1
3
2
-1
4
2
5
 
44
8
36
40
8
12
40
16
12
44
16
36
5
4
0
3
 
3
2
-1
4
2
5
4
0
5
3
0
-1
 
40
16
12
44
16
36
44
8
36
40
8
12
2
6
6
2
                           
60
8
56
52
8
8
52
24
8
60
24
56
4
5
3
0
                           
44
8
36
40
8
12
40
16
12
44
16
36
positive
negative
negative

(c)


1
2
0
4
 
3
2
-1
4
2
5
4
0
5
3
0
-1
 
23
10
5
26
10
23
26
4
23
23
4
5
5
5
3
3
 
4
0
5
3
0
-1
3
2
-1
4
2
5
 
56
16
32
56
16
32
56
16
32
56
16
32
4
0
2
1
 
4
0
5
3
0
-1
3
2
-1
4
2
5
 
23
10
5
26
10
23
26
4
23
23
4
5
2
1
4
0
 
3
2
-1
4
2
5
4
0
5
3
0
-1
 
26
4
23
23
4
5
23
10
5
26
10
23
3
3
5
5
                           
56
16
32
56
16
32
56
16
32
56
16
32
0
4
1
2
                           
26
4
23
23
4
5
23
10
5
26
10
23
negative
negative
positive

(d)

Figure 8

In agreement with what the examples suggest, the following theorem may be proven:

Theorem 2: The multiplication table of positive and negative horizontal cylinder matrices of period p is analogue to the multiplication table of positive and negative numbers (Figure 9).
 
 

A
B
AB
Horizontal cylinder matrix
Horizontal cylinder matrix
Horizontal cylinder matrix
+
+
+
-
+
-
+
-
-
-
-
+

Multiplication table of horizontal cylinder matrices

Figure 9

Transposition of the horizontal cylinder matrices leads immediately to Theorem 3.

Theorem 3: The multiplication table of positive and negative vertical cylinder matrices of period p is analogue to the multiplication table of positive and negative numbers.

What may be said about the multiplication of horizontal and vertical cylinder matrices?
 
 
 

Multiplication of horizontal and vertical cylinder matrices

Figure 10 presents the negative, horizontal cylinder matrix P and the negative, vertical cylinder matrix Q. Both matrices have period 3.

-1
2
0
1
-1
2
0
1
 
3
5
2
4
-1
-2
4
4
-3
-3
4
4
-3
-3
 
4
5
-2
3
-1
2
1
0
2
-1
1
0
2
-1
 
-2
-1
4
2
5
3
2
-1
1
0
2
-1
1
0
 
2
-1
3
-2
5
4
-3
-3
4
4
-3
-3
4
4
 
3
5
2
4
-1
-2
0
1
-1
2
0
1
-1
2
 
4
5
-2
3
-1
2
                 
-2
-1
4
2
5
3
                 
2
-1
3
-2
5
4
P
 
Q

Figure 10

As their dimensions are 6×8 and 8×6 respectively we can calculate both PQ and QP. Figure 11 presents PQ and QP.
 
 

14
8
-6
0
8
20
 
30
23
-9
-22
30
23
-9
-22
56
92
-42
56
-76
-42
 
23
30
-22
-9
23
30
-22
-9
-6
8
14
20
8
0
 
-9
-22
30
23
-9
-22
30
23
0
8
20
14
8
-6
 
-22
-9
23
30
-22
-9
23
30
-42
-76
56
-42
92
56
 
30
23
-9
-22
30
23
-9
-22
20
8
0
-6
8
14
 
23
30
-22
-9
23
30
-22
-9
             
-9
-22
30
23
-9
-22
30
23
             
-22
-9
23
30
-22
-9
23
30
PQ
 
QP

Figure 11

The matrix PQ is a positive cycle matrix of period 3 (cf. Gerdes, 2002a) (see Figure 12a, b), whereas the matrix QP is a positive simple helix matrix (cf. Gerdes, 2002b) (see Figure 12c, d).


 

Structure of a positive cycle matrix of dimensions 6×6

(a)

PQ as a positive cycle matrix 
of period 3

(b)

Structure of a positive simple helix matrix of dimensions 8×8

(c)

QP as a positive simple helix matrix

(d)

Figure 12

Figures 13, 14, and 15 present further examples of the multiplication of horizontal cylinder matrices and vertical cylinder matrices of period 3. Each time horizontal times vertical leads to cycle matrices and vertical times horizontal leads to simple helix matrices. The product matrices are positive if both cylinder matrices have the same sign; the product matrices are negative if the two cylinder matrices have opposite signs.

3
-1
2
-2
3
-1
2
-2
 
3
5
2
4
-1
-2
0
4
4
0
0
4
4
0
 
4
5
-2
3
-1
2
-1
3
-2
2
-1
3
-2
2
 
-2
-1
4
2
5
3
-2
2
-1
3
-2
2
-1
3
 
2
-1
3
-2
5
4
4
0
0
4
4
0
0
4
 
3
5
2
4
-1
-2
2
-2
3
-1
2
-2
3
-1
 
4
5
-2
3
-1
2
                 
-2
-1
4
2
5
3
                 
2
-1
3
-2
5
4
P (hor. +)
 
Q (vert. -)

 
-6
20
20
34
-4
-20
 
-9
35
12
8
-9
35
12
8
16
32
16
40
32
40
 
8
12
35
-9
8
12
35
-9
34
20
-20
-6
-4
20
 
12
8
-9
35
12
8
-9
35
20
-4
-6
-20
20
34
 
35
-9
8
12
35
-9
8
12
40
32
40
16
32
16
 
-9
35
12
8
-9
35
12
8
-20
-4
34
20
20
-6
 
8
12
35
-9
8
12
35
-9
             
12
8
-9
35
12
8
-9
35
             
35
-9
8
12
35
-9
8
12
                   PQ
 
QP

 

Structure of a negative cycle matrix of dimensions 6×6

(a)

PQ as a negative cycle matrix
of period 3

(b)

Structure of a negative simple helix matrix of dimensions 8×8

(c)

QP as a negative simple helix matrix

(d)

Figure 13

-1
2
0
1
-1
2
0
1
 
0
4
2
1
5
3
4
4
-3
-3
4
4
-3
-3
 
2
5
0
3
4
1
1
0
2
-1
1
0
2
-1
 
3
5
1
2
4
0
2
-1
1
0
2
-1
1
0
 
1
4
3
0
5
2
-3
-3
4
4
-3
-3
4
4
 
0
4
2
1
5
3
0
1
-1
2
0
1
-1
2
 
2
5
0
3
4
1
                 
3
5
1
2
4
0
                 
1
4
3
0
5
2
P (hor. -)
 
Q (vert. +)

 
10
20
2
10
16
2
 
5
3
10
12
5
3
10
12
-8
18
-8
20
18
20
 
12
10
3
5
12
10
3
5
10
20
2
10
16
2
 
10
12
5
3
10
12
5
3
2
16
10
2
20
10
 
3
5
12
10
3
5
12
10
20
18
20
-8
18
-8
 
5
3
10
12
5
3
10
12
2
16
10
2
20
10
 
12
10
3
5
12
10
3
5
             
10
12
5
3
10
12
5
3
             
3
5
12
10
3
5
12
10
Negative cycle matrix

PQ

 
Negative simple helix matrix

QP

Figure 14


 

3
-1
2
-2
3
-1
2
-2
 
2
1
5
4
-2
-2
0
4
4
0
0
4
4
0
 
5
-2
2
-2
1
4
-1
3
-2
2
-1
3
-2
2
 
-2
-2
4
5
1
2
-2
2
-1
3
-2
2
-1
3
 
4
1
-2
2
-2
5
4
0
0
4
4
0
0
4
 
2
1
5
4
-2
-2
2
-2
3
-1
2
-2
3
-1
 
5
-2
2
-2
1
4
                 
-2
-2
4
5
1
2
                 
4
1
-2
2
-2
5
P (hor. +)
 
Q (vert. +)

 
-22
-2
50
40
-2
-32
 
-19
29
-12
12
-19
29
-12
12
24
-32
48
24
16
48
 
29
-19
12
-12
29
-19
12
-12
50
-2
-22
-32
-2
40
 
-12
12
-19
29
-12
12
-19
29
40
-2
-32
-22
-2
50
 
12
-12
29
-19
12
-12
29
-19
48
16
24
48
-32
24
 
-19
29
-12
12
-19
29
-12
12
-32
-2
40
50
-2
-22
 
29
-19
12
-12
29
-19
12
-12
             
-12
12
-19
29
-12
12
-19
29
             
12
-12
29
-19
12
-12
29
-19
Positive cycle matrix

PQ

 
Positive simple helix matrix

QP

Figure 15

Figure 16 presents the multiplication table for horizontal and vertical cylinder matrices as may be conjectured on the basis of the experimentation.

A
B
AB
BA
Horizontal cylinder matrix
Vertical cylinder matrix
Cycle matrix
Simple helix matrix
+
+
+
+
-
+
-
-
+
-
-
-
-
-
+
+

Multiplication table of horizontal and vertical cylinder matrices

Figure 16




The following conjectures may be formulated:

Conjecture 4: The products of horizontal and vertical cylinder matrices of dimensions (2m)×(2n) and (2n)×(2m), respectively, that have the same period p, are cycle matrices of period p. The cycle matrix is positive if both cylinder matrices have the same sign; the cycle matrix is negative if the cylinder matrices have opposite signs.

In Gerdes (2002a) only square cycle matrices are defined. The concept of cycle matrix may be extended to non-square matrices. For instance, the matrix in Figure 17 may be considered a positive 6×12 cycle matrix of period 3. It appears as the product of the horizontal and vertical cylinder matrices of period 3 in Figure 18.

26
19
6
24
37
8
8
37
24
6
19
26
19
58
37
19
40
37
37
40
19
37
58
19
6
37
26
8
19
24
24
19
8
26
37
6
24
19
8
26
37
6
6
37
26
8
19
24
37
40
19
37
58
19
19
58
37
19
40
37
8
37
24
6
19
26
26
19
6
24
37
8

Figure 17


0
1
3
4
 
0
5
4
1
2
3
3
2
1
4
5
0
5
5
2
2
 
1
5
3
0
2
4
4
2
0
3
5
1
4
3
1
0
 
3
2
1
4
5
0
0
5
4
1
2
3
1
0
4
3
 
4
2
0
3
5
1
1
5
3
0
2
4
2
2
5
5
                         
3
4
0
1
                         
Hor., -
 
Vert. -

Figure 18

Conjecture 5: The products of vertical and horizontal cylinder matrices of dimensions (2m)×(2n) and (2n)×(2s) respectively, that have the same period p, are single helix matrices. The simple helix matrix is positive if both cylinder matrices have the same sign and is negative if the two cylinder matrices have opposite signs.

0
1
0
1
 
1
5
7
9
3
2
4
2
4
 
3
9
7
5
1
3
3
3
3
 
1
5
7
9
3
4
2
4
2
 
3
9
7
5
1
1
0
1
0
           
A
 
B

Figure 19

6
18
14
10
2
 
70
60
70
60
28
92
84
76
20
 
60
70
60
70
24
84
84
84
24
 
70
60
70
60
20
76
84
92
28
 
60
70
60
70
2
10
14
18
6
         
AB
 
BA

Figure 20


For certain dimensions and periods, cylinder matrices may be simultaneously positive and negative. For instance, the horizontal and vertical matrices of dimensions 5×4 and 4×5 with period 5 presented in Figure 19 are simultaneously positive and negative. Their product matrices AB and BA (Figure 20) are also simultaneously positive and negative. In the case of the helix matrices, this means that they are helix matrices of period 1.
 
 

Multiplication of vertical cylinder matrices and cycle matrices

Figure 21 presents examples of the multiplication of vertical cylinder matrices and cycle matrices. 

Extrapolation on the basis of these examples leads to the following conjecture:

Conjecture 5: The product of a vertical cylinder matrix and a square cycle matrix having the same period p is a vertical cylinder matrix of period p. The sign of the product matrix is equal to the product of the signs of the multiplicands.

The reader is invited to (im)prove the conjectures and to discover more properties of cylinder matrices.

-2
5
2
3
4
1
 
3
0
2
-3
-2
4
 
27
-17
39
60
-9
24
3
5
1
-2
4
2
 
1
-1
1
6
-1
6
 
39
-9
27
24
-17
60
1
4
3
2
5
-2
 
-3
0
4
3
-2
2
 
24
-9
60
39
-17
27
2
4
-2
1
5
3
 
2
-2
3
4
0
-3
 
60
-17
24
27
-9
39
-2
5
2
3
4
1
 
6
-1
6
1
-1
1
 
27
-17
39
60
-9
24
3
5
1
-2
4
2
 
4
-2
-3
2
0
3
 
39
-9
27
24
-17
60
1
4
3
2
5
-2
               
24
-9
60
39
-17
27
2
4
-2
1
5
3
               
60
-17
24
27
-9
39
Negative vertical cylinder matrix of period 3

P

 
Negative cycle matrix of period 3

Q

 
Positive vertical cylinder matrix of period 3

PQ

(a)


-2
5
2
3
4
1
 
6
0
1
3
4
8
 
37
83
70
59
77
32
3
5
1
-2
4
2
 
2
7
5
2
9
5
 
59
83
32
37
77
70
1
4
3
2
5
-2
 
1
4
6
8
0
3
 
32
77
59
70
83
37
2
4
-2
1
5
3
 
3
0
8
6
4
1
 
70
77
37
32
83
59
-2
5
2
3
4
1
 
5
9
2
5
7
2
 
37
83
70
59
77
32
3
5
1
-2
4
2
 
8
4
3
1
0
6
 
59
83
32
37
77
70
1
4
3
2
5
-2
               
32
77
59
70
83
37
2
4
-2
1
5
3
               
70
77
37
32
83
59
Negative vertical cylinder matrix of period 3

P

 
Positive cycle matrix of period 3

Q

 
Negaitive vertical cylinder matrix of period 3

PQ

(b)


2
1
5
4
-2
-2
 
3
0
2
-3
-2
4
 
-20
-3
31
25
-13
4
5
-2
2
-2
1
4
 
1
-1
1
6
-1
6
 
25
-3
4
-20
-13
31
-2
-2
4
5
1
2
 
-3
0
4
3
-2
2
 
4
-13
25
31
-3
-20
4
1
-2
2
-2
5
 
2
-2
3
4
0
-3
 
31
-13
-20
4
-3
25
2
1
5
4
-2
-2
 
6
-1
6
1
-1
1
 
-20
-3
31
25
-13
4
5
-2
2
-2
1
4
 
4
-2
-3
2
0
3
 
25
-3
4
-20
-13
31
-2
-2
4
5
1
2
               
4
-13
25
31
-3
-20
4
1
-2
2
-2
5
               
31
-13
-20
4
-3
25
Positive vertical cylinder matrix of period 3

P

 
Negative cycle matrix of period 3

Q

 
Negative vertical cylinder matrix of period 3

PQ

(c)


2
1
5
4
-2
-2
 
4
-1
2
0
3
5
 
9
7
39
38
16
5
5
-2
2
-2
1
4
 
3
-2
1
3
-3
1
 
39
16
9
5
7
38
-2
-2
4
5
1
2
 
2
3
4
5
-1
0
 
5
16
38
39
7
9
4
1
-2
2
-2
5
 
0
-1
5
4
3
2
 
38
7
5
9
16
39
2
1
5
4
-2
-2
 
1
-3
3
1
-2
3
 
9
7
39
38
16
5
5
-2
2
-2
1
4
 
5
3
0
2
-1
4
 
39
16
9
5
7
38
-2
-2
4
5
1
2
               
5
16
38
39
7
9
4
1
-2
2
-2
5
               
38
7
5
9
16
39
Positive vertical cylinder matrix of period 3

P

 
Positive cycle matrix of period 3

 
Positive vertical cylinder matrix of period 3

PQ

(d)

Figure 21





References

Gerdes, Paulus (2002a), From Liki-designs to cycle matrices, Visual Mathematics, Vol. 7, March 2002 (http://members.tripod.com/vismath7/gerd/)

Gerdes, Paulus (2002b), Helix matrices (submitted for publication in Visual Mathematics). 
 
 

VisMathHOME