Properties of Cantor's Ternary Set
By Tom Brown
- 56 reads
Properties of Cantor's Ternary Set
First I describe two different constructions for Cantor's Set, then investigate some of it's very special properties.
It is not hard to prove they are equivalent.
I am very limited here with typesetting in places it is a bit crude so bear with me and use your imagination. Please refer to the End Note.
A n = { a n }
=
First, as intersecting closed intervals. Consider the interval [0,1] i.e all real numbers between 0 and 1 and including the endpoints 0 and 1.
The first step is removing the middle open third interval ( 1/3 , 2/3 ) resulting in two closed intervals [0, 1/3] and [2/3, 1]
Then each of these in turn, leaving 2 x 2 = 4
________________________ S 0 = [0,1]
________ ________ S 1
__ __ __ __ S 2
… ..
… .. . . .. … S n
The procedure is repeated indefinitely (to infinitely) leaving a resulting set of real numbers not containing any intervals.
= = =
Each real number b in the interval [0,1] has a ternary expansion, meaning it has an expansion in digits 0, 1 and 2 as the sum of powers of 1/3 with bn the n'th digit
Such as 0. b1 b2 b3 b4 … with bn = 1, 2, or 3 and the number b is
b = b1/ 3 + b2 /9 + b3/ 27 + … (with 9 = 3^2 27 = 3^3 etc.)
The Cantor Set is then all the ternary expansions in [0,1] without a digit 1
= = =
A set of real numbers is called “countably infinite” if it is infinite but can not be written as a sequence. Cantor's diagonal argument proves that the Cantor Set is uncountable. I give a detailed proof.
Suppose that the Cantor Set can be written as a a sequence b i, we will demonstrate a number that is not in the set
b1 = b11 b12 b13 b14 b15 …
b2 = b21 b22 b23 b24 b25 …
b3 = b31 b32 b33 b34 b35 …
…
bi = bi1 bi2 bi3 …
…
b5 = b51 b52 b53 b54 b55 …
Each digit (entry) bij is now either a zero or a one because it is a Cantor Set. When working down the diagonals (right–>left) of the “infinite matrix”
b11 b12 – b21 b13 – b22 – b31 b14 – b23 – b32 – b41
b42 – b33 – b42 – … …
down the main diagonal, construct the number a as a = 0. a1 a2 a3 a4 …
where aj = 1 if bjj=0 and aj=0 if bjj=1
But now a cannot be in the sequence bi since it differs from each number in the sequence (in the place i = j ). and thus a not in the sequence.
This is a contradiction it is not, since Cantor's Set as assumed and cannot be written as a sequence and is uncountably infinite.
= = =
There are further very interesting (desirable) properties and examples.
The property of compactness is very important in tropological spaces and for the real line is characterised by a subset of real numbers being closed and bounded.
Clearly for the Cantor Set each element is between zero and one, thus the set is bounded. Also a closed interval (containing endpoints) is topologically “closed” so that Cantor's Set is compact as being the intersection of closed sets.
These are very basic ideas.
We have already brought in the concept as a countable set being a sequence. Further examples are the integers, indeed the union of countably many sets is also, it is easy to prove.
Surprisingly the rational numbers (fractions) can also be written (exhausted) in a sequence and it is easy to demonstrate. A countable union of countable sets is also countable.
= = =
The concept of a Null Set (as a set of of measure zero) enables one to study the Lebesgue integral for real numbers without formal Measure Theory.
Given any positive epsilon ( eps > 0 ), if a set B of real numbers can be covered by a sequence of intervals {I n } with total length less than epsilon the set B is a null Set.
It means each element of B is contained in an interval I n ( { I n } covers B), as in the union of the intervals A n, and the total sum of lengths of the A n is less than any given eps.
=
Suppose B is countable (eg. integers) thus is a sequence { b n }. Let b n be contained in an interval I n chosen such that length ( I n ) is less than eps/ 2^n.
The sum of lengths of the I n is less than eps.( ½ + ¼ + 1/8 + … ) as a geometric series,
and is less than epsilon. So a countable set (eg. rationals) has measure zero
=
Note that for each n the union { I n } = S n covers the Cantor Set. Note that the total length of S n is length (S n ) = (2/3)^n from the construction of S n.
Given eps > 0 choose N such that eps n < eps N < eps for all n > N
For any eps > 0 Cantor's Set may be covered with intervals the sum of which is less then epsilon and it is thus a Null Set from the definition.
So we have an example of an (over) countable set that is a Null Set.
=
As another property the Cantor Set also is a fractal. One can see that it has scaled subsets, proper subsets which are identical as smaller and smaller sets and subsets ad infinitum (to infinitesimal).
=
End Note
Sets will be in capitals, elements in small type and digits indexed and doubled index. We work with variables as integers, the set of real numbers, subsets, intervals and sets of intervals.
As slightly more familiar notation n is an integer ; a n = a_n is a real number, (sub) sets of real numbers is A.
A = { a n } = { a_n } a sequence, I n = I_n an interval ; b mn = b_mn each a digit, a 0, 1 or 2
-
My Links Strongly related also read my AbcTales stories “Naive Set Theory”, “Basic Set Theory” in my Numbers and Infinities collection.
Literature as standard references, a comprehensive treatment of real analysis “Mathematical Analysis” by Tom Apostol one of the best formal textbooks I have come across, and then more of a curiosity, “Lebesgue Integration & Measure”, Alan J. Weir.
Comments and replies, as remarks or observations most welcome.
- Log in to post comments
Comments
Beyond addition and
Beyond addition and subtraction, I am embarrassed to say Math eludes me. I don't understand the concept of Algebra or Trigonometry even though I solved problems to complete these courses in school, I didn't retain any of it. But it's fascinating, how everything is mathematical in the universe, and with any science there are mathematical properties and understanding them can answer questions. I respect the brilliance it takes to understand equations, to measure the world by it- I wish I understood this numeric language as you do, but I've 9accepted my limited capability. ![]()
- Log in to post comments


