International Journal of Scientific and Research Publications, Volume 14, Issue 10, October 2024 349
ISSN 2250-3153
Construction and Properties of Ternary Cantor set
1
Paramacutty Paramadevan
1
Department of Mathematics, Eastern University Sri Lanka.
[email protected] DOI: 10.29322/IJSRP.14.10.2024.p154XX
Paper Received Date: 24th September 2024
Paper Acceptance Date: 25th October 2024
Paper Publication Date: 30th October 2024
Abstract: A remarkable mathematical construct, the Ternary Cantor set is distinguished by its special qualities, such as being an
uncountable set of measure zero, a perfect set, and a nowhere dense set. Despite its relatively straightforward construction, this set
has extensive applications in computer science, physics, and mathematics. Among other areas in which it is important are chaotic
dynamical systems, fractal theory, and set theory. Moreover, the Cantor set has been extended to higher dimensions and used in a
variety of mathematical environments, including fractal, metric, and topological spaces. With an explanation of its recursive
construction and a focus on its essential qualities, this study attempts to present the Cantor set in a way that is easily understood by
novices. Furthermore, a number of significant theorems about the Cantor set are explored to provide a deeper understanding of its
structure and significance in mathematics.
Keywords: uncountable set, measure, nowhere dense, fractal theory, chaotic dy- namical systems.
Introduction
Cantor set is a set possess uncountable many points in the closed interval [0, 1] and was discovered in 1874 by Henry John Stephen
Smith. Later on, after the first research article published on Cantor ternary set by Georg Cantor, a German mathematician, in 1883,
Cantor set become famous and started to play important roles in many area of mathematics such as set theory, measure theory to
real analysis, fractal theory, topology and chaotic dynamical systems and in many area of computer science. The interplay between
ternary arithmetic and the iterative construction of the Cantor set has implications for signal processing, information theory and the
broader study of non-linear dynamics. Cantor’s original construction involved removing the middle third of a line segment
repeatedly, producing a set with remarkable properties. The classical construction of the Cantor set involves a process known as
iterated removal of intervals. This approach, while conceptually simple, has inspired numerous variations and extensions.
Researchers have explored modifications to the removal process considering different fractions and patterns leading to rich variety
of Cantor like sets.
Even though the Cantor set can be define in many ways in various literature[1], the most accepted and comprehensive form [2,3] is
as follows:
∞ 3𝑛−1
3𝑘 + 1 3𝑘 + 2
𝐶 = [0,1] / ⋃ ⋃ ( , ),
3𝑛+1 3𝑛+1
𝑛=0 𝑛=1
3𝑘+1 3𝑘+2 3𝑘+0 3𝑘+3
where, every middle third is removed as the open interval ( , ) from the closed interval [ , ]=
3𝑛+1 3𝑛+1 3𝑛+1 3𝑛+1
𝑘+0 𝑘+1
[ 3𝑛 , 𝑛 ] 𝑠urrounding it. While in some literatures Canter set defined quite different form such as:
3
∞ 3𝑛−1 −1
3𝑘 + 0 3𝑘 + 1 3𝑘 + 2 3𝑘 + 3
𝐶=⋂ ⋃ ([ 𝑛 , ]∪[ 𝑛 , ]),
3 3𝑛 3 3𝑛
𝑛=0 𝑘=0
3𝑘+1 3𝑘+2
where, the middle third ( , ) of the foregoing closed interval
3𝑛 3𝑛
𝑘+0 𝑘+1 3𝑘+0 3𝑘+3 3𝑘+0 3𝑘+1 3𝑘+2 3𝑘+3
[ , ]=[ , ] is removed by intersecting with [ , ]∪[ , ].
3𝑛−1 3𝑛−1 3𝑛 3𝑛 3𝑛 3𝑛 3𝑛 3𝑛
This publication is licensed under Creative Commons Attribution CC BY.
10.29322/IJSRP.14.10.2024.p15447 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 14, Issue 10, October 2024 350
ISSN 2250-3153
Let us explore the construction process of the Cantor set in the next section.
Construction of ternary cantor set
In this section, let us explore the iterative process to form the Cantor set by recursion method. At first trisect the closed interval [0,1]
at two points 1/3, 2/3 and remove the open interval (1/3,2/3), which is called the middle third, and obtained the set 𝐶1 = [0,1/3] ∪
[2/3,1]. Again by trisecting each of the closed intervals [0, 1/3] and [2/3,1] and removing the open middle thirds (1/9,2/9) and
(7/9,8/9), respectively, we obtain
𝐶2 = [0,1/9]∪[2/9,3/9]∪[6/9,7/9]∪[8/9,9/9]. Repeatedly, by removing the middle third of each of the above four intervals, we obtain
𝐶3 = [0, 1/27] ∪ [2/27, 3/27] ∪ [6/27, 7/27] ∪ [8/27, 9/27] ∪ [18/27, 19/27] ∪ [20/27, 21/27] ∪ [24/27, 25/27] ∪ [26/27, 27/27].
By continuing in the same manner recursively, we can obtain sets 𝐶1 , 𝐶2 , 𝐶3 , … .
Then the Ternary Cantor set, denoted by 𝐶,is the intersection of all 𝐶1 , 𝐶2 , 𝐶3 …
∞
𝑖𝑒. 𝐶 = ⋂ 𝐶𝐾
𝑘=1
Alternatively, as discussed in [4], if 𝑆 is a subset of the set of real numbers, then we define
1 1 1 1
𝑠 = { 𝑠: 𝑠 ∈ 𝑆} 𝑎𝑛𝑑 𝑆 + = {𝑠 + : 𝑠 ∈ 𝑆}.
3 3 3 3
𝐿𝑒𝑡 𝐶0 = [0,1],
1 1 2 1 2
𝑡ℎ𝑒𝑛 𝐶1 = 𝐶0 ∪ ( 𝐶0 + ) = [0, ] ∪ [ 1 ],
3 3 3 3 3
1 1 2 1 2 3 6 7 8 9
𝐶2 = 𝐶1 ∪ ( 𝐶1 + ) = [0, ] ∪ [ , ] ∪ [ , ] ∪ [ , ] by continuing, we can obtain
3 3 3 9 9 9 9 9 9 9
1 1 2
𝐶𝑘+1 = 𝐶𝑘 ∪ ( 𝐶𝑘 + ) .
3 3 3
1 1 2
Since 𝐶 = ⋂∞
𝑘=1 𝐶𝐾 , 𝑤𝑒 ℎ𝑎𝑣𝑒 𝐶 = 𝐶 ∪ ( 𝐶 + ). 3 3 3
In these ways the Cantor set can be generated recursively.
Simply, the graphical representation of the elements of the Cantor set can be represented as follows:
0 1
1 2
0 1
3 3
1 2 1 6 7 8
0 1
9 9 3 9 9 9
3 6 9 18 21 24
0 1
27 27 27 27 27 27
9 18
0 1
27 27
Figure (i)
In the above Figure (i), each line segment indicates real numbers in the respective closed intervals. The first line shows that, initially
we consider all real numbers in the closed interval [0, 1]. Second line shows two disjoint closed intervals [0, 1/3] and [2/3, 1] after
the removal of the middle third (1/3,2/3) of [0,1]. Meantime, third line shows the four closed intervals remaining once the middle
thirds (1/9, 2/9) and (7/9, 8/9) of (0, 1/3) and (2/3, 1) respectively are removed and the iterative process continuous infinitely may
times.
In the next section let us discuss some properties of Ternary Cantor set.
Some properties of ternary cantor set
In this section let us explore some basic properties of Ternary Cantor set.
This publication is licensed under Creative Commons Attribution CC BY.
10.29322/IJSRP.14.10.2024.p15447 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 14, Issue 10, October 2024 351
ISSN 2250-3153
Property 1: 𝐶 is non-empty
We can observe that during each step finding𝐶𝑘 , the end points of 𝐶𝑘−1 remain in 𝐶𝑘 . Therefore, C contains all the end points, hence
C is non-empty.
Property 2: 𝐶 is closed
Since each 𝐶𝑘 is closed, as it is a finite union of closed intervals and 𝐶 =∩ 𝐶𝐾 is a countable intersection of closed sets and is
therefore closed.
Property 3: 𝐶 is perfect
We have seen that the Cantor set is closed and it is proved in Theorem 6 it has no isolated points. These are the conditions for a set
to be perfect[5].
Theorems related with ternary cantor set
Before get in to the theorem let us start with the following lemma:
Lemma .1
Every real number in [0, 1] can be expressed of the form
∞
𝑎𝑘
∑ , 𝑤ℎ𝑒𝑟𝑒 𝑎𝑘 ∈ {0,1,2}.
2𝑘
𝑘=1
Proof: Define a function 𝑓 ∶ [0,1,2]𝑁 → [0,1] such that
𝑎𝑘 𝑎𝑘
𝑓(𝑎1 , 𝑎2 , 𝑎3 , … . . ) = 𝑓 (∑∞
𝑘=1 ) = 𝑐 ∈ [0,1], where the sequence (𝑎1 , 𝑎2 , 𝑎3 , … . . ) represents the series ∑∞
𝑘=1 .
2𝑘 3𝑘
𝑎𝑘
Clearly 𝑓 is a well-defined function and the sequence of partial sum 𝑠𝑛 = ∑𝑛𝑘=1
3𝑘
is increasing in n and bounded above by
∞ 1
1
2 ∑ 𝑘 = 2. 3 = 1.
3 1
𝑘=1 1−
3
Therefore, the infinite series is convergent in [0, 1].
First, let us prove 𝑓 is onto.
𝑎1
Let 𝑐 ∈ [0,1) and 𝑎1 be the largest integer such that ≤ 𝑐. Since 0 ≤ 𝑐 < 1
3
𝑎1 1
we have 0 ≤ 𝑎1 ≤ 2 𝑎𝑛𝑑 0 ≤ 𝑐 − < .
3 3
𝑎2 𝑎1 1 𝑎1 𝑎2 1
Let 𝑎2 be the largest integer such that ≤𝑐− < . We must have 0 ≤ 𝑎2 ≤ 2 𝑎𝑛𝑑 0 ≤ 𝑐 − − < .
3 3 3 3 32 32
By continuing in this way we obtain a sequence of 𝑎𝑖 ∈ {0,1,2} such that
𝑛
𝑎𝑘 1
0≤𝑐−∑ < .
3𝑘 3𝑛
𝑘=1
This implies,
𝑛 ∞
𝑎𝑘 𝑎𝑘
𝑐 = 𝑙𝑖𝑚 ∑ = ∑ 𝑘.
3𝑘 3
𝑘=1 𝑘=1
Further, 𝑐 = 1, we take all 𝑎𝑗 = 2 then
This publication is licensed under Creative Commons Attribution CC BY.
10.29322/IJSRP.14.10.2024.p15447 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 14, Issue 10, October 2024 352
ISSN 2250-3153
𝑛 𝑛−1 ∞
2 2 1 1 2
∑ 𝑘 = ∑ 𝑘 = 1 − 𝑛 . 𝐻𝑒𝑛𝑐𝑒 ∑ 𝑘 = 1
3 3 3 3 3
𝑘=1 𝑘=0 𝑘=1
Thus we proved ∀𝑐 ∈ [0,1] , there exists a sequence (𝑎1 , 𝑎2 , 𝑎3 , … . . ) such 𝑓(𝑎1, 𝑎2 , 𝑎3 , … . ) =
∞
𝑎𝑘
∑ = 𝑐, 𝑡ℎ𝑎𝑡 𝑖𝑠 𝑓 𝑖𝑠 𝑜𝑛𝑡𝑜.
3𝑘
𝑘=1
However, this mapping is not one to one. Let we explore in which situation it is not one to one.
Suppose we have two different sequences (𝑎1 , 𝑎2 , 𝑎3 , … . ) and (𝑏1 , 𝑏2 , 𝑏3 , … ) such that
∞ ∞
𝑎𝑘 𝑏𝑘
∑ 𝑘
= ∑ 𝑘.
3 3
𝑘=1 𝑘=1
Since these are different sequences there exists some k such that 𝑎𝑘 ≠ 𝑏𝑘 .
Suppose 𝑗 is the smallest integer such that 𝑎𝑗 ≠ 𝑏𝑗 .
Without loss of generality as- same 𝑎𝑗 > 𝑏𝑗 then we have
𝑎𝑗 −𝑏𝑗 𝑎𝑘 𝑏
+ ∑∞
𝑘=𝑗+1 = ∑∞ 𝑘
𝑘=𝑗+1 𝑘 .
3𝑗 3𝑘 3
𝑎𝑗 −𝑏𝑗 1 𝑏𝑘 2 1
Since ≥ 𝑎𝑛𝑑 ∑∞
𝑘=𝑗+1 ≤ ∑∞
𝑘=𝑗+1 = .
3𝑗 3𝑗 3𝑘 3𝑘 3𝑛
The only one possibility to have this inequalities and the previous equation are that
𝑎𝑗 − 𝑏𝑗 = 1 and 𝑎𝑘 = 0, 𝑏𝑘 = 2 for 𝑘 ≥ 𝑗 + 1.
That is the sequence 𝑎𝑘 has terminal 0’s and 𝑏𝑘 has terminal 2’s. Except this situation the ternary expansion of 𝑐 ∈ {0,1} is unique.
Theorem 1. Each number of the Cantor set can be expressed as a series, called
𝑎𝑘
Ternary expansion, of the form ∑∞
𝑘=1 , 𝑤ℎ𝑒𝑟𝑒 𝑎𝑘 ∈ {0,2}.
3𝑘
𝑝𝑟𝑜𝑜𝑓. Note that in the proof of the above lemma, 𝑎𝑗 = 𝑏𝑗 + 1 then we have either 𝑎𝑗 = 1 or 𝑏𝑗 = 1. From this we can conclude
that different sequence in {0, 2}N give different values of c in the cantor set C ⊆ [0, 1].
Theorem 2: Every real number in [0, 1] can be expressed in the scale of 2, called
𝑏𝑘
Binary expansion, of the form ∑∞
𝑘=1 , 𝑤ℎ𝑒𝑟𝑒 𝑏𝑘 ∈ {0,1}.
2𝑘
𝑎𝑘
𝑃𝑟𝑜𝑜𝑓. Define a map 𝑓: 𝐶 → [0,1] such that 𝑓(𝑥) = ∑∞
𝑘=1
2
, 𝑤ℎ𝑒𝑟𝑒 𝑥 ∈ 𝐶
2𝑘
𝑎𝑘
has unique expression such a way that 𝑥 = ∑∞
𝑘=1 , 𝑎𝑘 ∈ {0,2}.
3𝑘
𝑎 𝑎2 𝑎3
We can see that if x has ternary expansion (𝑎1 , 𝑎2 , 𝑎3 , … ), 𝑡ℎ𝑒𝑛 𝑓(𝑥) has binary expansion ( 1 , , , … ). That is the binary
2 2 2
expansion become an element of [0,1]. Thus the function is well-defined and clearly this function is on to since every sequence in
{0,1}N can be obtained in this manner.
1 1 1 2 2
But this mapping is not one to one since = (0,2,2,2, … ), then 𝑓 ( ) = (0,1,1,1, … ) = , while = (2,0,0,0, … ), then 𝑓 ( ) =
3 3 2 3 3
1
(1,0,0,0, … ) = .
2
Theorem 3: The Canter set is non-denumerable.
𝑃𝑟𝑜𝑜𝑓. By Theorem 1, every number in the Cantor set has ternary expansion of the form
This publication is licensed under Creative Commons Attribution CC BY.
10.29322/IJSRP.14.10.2024.p15447 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 14, Issue 10, October 2024 353
ISSN 2250-3153
∞
𝑎𝑘
∑ , 𝑤ℎ𝑒𝑟𝑒 𝑎𝑘 ∈ {0,2} .
3𝑘
𝑘=1
While Theorem 2, every number in [0, 1] has binary expansion of the form
𝑏𝑘
∑∞
𝑘=1 , 𝑤ℎ𝑒𝑟𝑒 𝑏𝑘 ∈ {0,1} with the assumption that not all 𝑏𝑘 after certain terms are 0.
2𝑘
Now we construct a one to one correspondence between these two expansions such that 𝑏𝑘 = 0 when ever 𝑎𝑘 = 0 and 𝑏𝑘 = 1 when
ever 𝑎𝑘 = 2. Then, since the set of all real numbers in [0, 1] is non-denumerable so also the Cantor set.
Theorem 4: The Cantor set is measurable.
𝑃𝑟𝑜𝑜𝑓. We can obtain a sequence, when see what is removed from each iteration. At first, the middle third of [0, 1], interval of
1 1
length is removed. During the second iteration, two middle third intervals each of length are removed. In the next iteration, four
3 9
1 1
middle third intervals each has length is removed. In general at the 𝑘 𝑡ℎ step 2𝑘−1 intervals with length is removed. Therefore,
27 3𝑘
2𝑛−1
measure of C, 1 − ∑∞
𝑛=1 = 0.
3𝑛
Theorem 5: The Cantor set is nowhere dense.
𝑃𝑟𝑜𝑜𝑓. It is enough to proof 𝐶 does not contain any open intervals.
Let 𝑂 be an arbitrary open interval in [0, 1] with length 𝑙. By Archimedean property,
1 1
choose 𝑘 such that < 𝑙. We know that each 𝐶𝑘 has 2𝑘 disjoint closed intervals each has length . Let 𝐼 be one of such interval.
3𝑘 3𝑘
Therefore, length of 𝐼 is less than the length of the open interval 𝑂. Hence 𝑂 ⊈ 𝐼 ⊆ 𝐶𝑘 , thus 𝑂 ⊈ 𝐶.
Theorem 6: The Cantor set has no isolated points.
𝑃𝑟𝑜𝑜𝑓. It is enough to proof every point 𝑐 of 𝐶 is a limit point of 𝐶. Which is
equivalent to show for any arbitrary 𝜖, we can find 𝑡 ∈ 𝐶 such that 0 < |𝑐 − 𝑡| < 𝜖
1
By Archimedean property, choose 𝑘 such that < 𝑙. Since 𝑐 𝜖 𝐶, 𝑐 ∈ 𝐶𝐾. for some k.
3𝑘
1
That is, there is a closed interval 𝐼 ∈ 𝐶 with length
3𝑘
When we consider 𝐼 ∩ 𝐶𝑘+1 , it contains two component intervals in which one, say
1
𝐼0 must contain c with an end point t. Thus 0 < |𝑐 − 𝑡| ≤ < 𝜖.
3𝑘
Conclusion
The Cantor set has many interesting properties and consequences in the field of pure mathematics, computer science and
engineering. One of the interesting property addressed in this research work is that it is a uncountable set with measure zero. In this
study the non-denumerable property has been proved using the fact that every number in the Cantor set has ternary expansion in
which only the integers 0, 2 involve and every real number in [0,1] has binary expansion. While the classical construction method
is well established, alternative approaches also to constructing Cantor set have been proposed in various literatures. These include
probabilistic methods, symbolic dynamics and number theory connections. Exploring this alternative avenues not only deepens our
understanding of Cantor set but also opens up new perspective for future research. In conclusion, the construction of the Cantor set
has been a central theme in mathematical research for over a century. The rich tapestry of approaches, generalizations, and
applications underscores the enduring significance of this fundamental mathematical concept. As researchers continue to explore
new directions and address open questions, the literature on the construction of Cantor set is likely to evolve and expand contributing
to the broader landscape of mathematical knowledge.
This publication is licensed under Creative Commons Attribution CC BY.
10.29322/IJSRP.14.10.2024.p15447 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 14, Issue 10, October 2024 354
ISSN 2250-3153
REFERENCES
1. Dylan R. Nelson. The Cantor set- A Brief Introduction. https://www.scribd.com/document/764456
2. Murray R. Spiegal (1979). Theory and Problems of Real Analysis. Schaums outline series.
3. Royden, H.L. Real Analysis (3rd edition)(2004). Prentice-Hall of India Private Limited, New Delhi.
4. https://sites.math.washington.edu/hart/m424/Cantorset.pdf
5. Rebecca Stettin. Generalization and Properties of the ternary Cantor set and Exploration in Similar sets.
This publication is licensed under Creative Commons Attribution CC BY.
10.29322/IJSRP.14.10.2024.p15447 www.ijsrp.org
