7

I assume that $\frac{1}{3}$ is equal to $0.3333...$.

Let's define a sequence as follows: $0.3$, $0.33$, $0.333$, $0.3333$,...

Question: is $\frac{1}{3}$ included in this sequence?

Every item in the sequence clearly has finite number of decimals, and $\frac{1}{3}$ has infinite decimals so it is clearly not included.

On the other hand the limit of this sequence is $\frac{1}{3}$ so it seems ok to say that it "includes" $\frac{1}{3}$.

What is the correct answer?

psmith
  • 245
  • 24
    The correct answer is that it is not included. Every entry in the sequence is smaller than the limit. – Peter Apr 13 '24 at 09:39
  • 13
    is $\frac{1}{3}$ included in this series? You need to defined what "included" here means. As a limit, yes. – Dietrich Burde Apr 13 '24 at 09:39
  • 9
    What you mentioned is not a series , but a sequence. A series is a summation of entries of a sequence. And a better formulation would be "belongs to" instead of "included". – Peter Apr 13 '24 at 09:45
  • 1
    I'm not really sure what do i mean "included". This comment started my line of thinking, to give you more context: https://math.stackexchange.com/questions/237300/why-is-the-set-of-all-real-numbers-uncountable#comment525387_237300 – psmith Apr 13 '24 at 09:45
  • Thanks @Peter, updated to sequence. – psmith Apr 13 '24 at 09:48
  • As an engineer, I would say what is your margin of error in order to be close enough to be included. It is like the old example of placing all the girls on one side of the gym, and all the boys on the other side of the gym. Every 5 minutes the line of girls and the line of boys reduces the space between them by 1/2. The physicist will say never being a purest for numbers, and the engineer will say within the hour is close enough. – Forward Ed Apr 14 '24 at 04:46
  • 3
    @ForwardEd Physicists are well aware that, in the real world, "very close" is generally indistinguishable from "equal". – David Z Apr 14 '24 at 09:29
  • This is an equivalent question to whether the limit's index belongs to the set of indexes, ie, is $\infty$ a natural number. Please see MSE Question 1088877 and MSE Q 36289. – Jam Apr 14 '24 at 12:18
  • Well I guess you engineers and physicists would probably be just happy, then, to do away with all of calculus, all those divs grads curls, all that stuff whose mathematical formulation depends on limits :-) – Lee Mosher Apr 14 '24 at 12:56
  • “0.3, 0.33, 0.333, 0.3333, ...” does not define a sequence. – user3840170 Apr 14 '24 at 14:13
  • 1
    @DietrichBurde: There is no way that the value $\frac13$ is included in the sequence $(0.3,0.33,\ldots)$, and I am astonished to see you vacillating on the question. There is no need to define what "included" means; it is unambiguous in this context. – TonyK Apr 14 '24 at 23:33
  • 5
    @user3840170: don't be ridiculous. We all know what the OP means. – TonyK Apr 14 '24 at 23:36
  • @Peter I'm glad this got upvotes. It is possible to give a simple answer without confusing a beginner by saying "well, kind of no and kind of yes, you see if we consider the complexified bundle...." – Thierry Apr 15 '24 at 13:08
  • In simple point-set topology, a related concept is the set $A={0.3, 0.33, 0.333, \ldots}$. Then the number $\frac13$ is not a member of $A$. In symbols $\frac13 \notin A$. But the fact that the limit of your sequence is $\frac13$ which you correctly mention, shows that $A$ is not a closed set. There is a concept called the closure of $A$, sometimes denoted $\overline{A}$, and this is a larger set. And in this case, we have $\frac13 \in \overline{A}$, so "one third" is a member of the closure of $A$. – Jeppe Stig Nielsen Apr 15 '24 at 13:27
  • @Thierry I agree TonyK here. There is only one reasoanble meaning of "including" , namely that the desired number belongs to the sequence , in other words , there is a member of the sequence being the desired number. I would use other words to describe other concepts. – Peter Apr 15 '24 at 13:33
  • @TonyK I am not, and no you don’t. – user3840170 Apr 15 '24 at 14:41
  • There's something humorous about a question on ${0.3, 0.33, 0.333,\dots}$ being closed. – A rural reader Apr 16 '24 at 14:21

4 Answers4

36

One thing that hasn't been said in the previous answers, and really needs to be emphasized, is that mathematics is not about guesswork, so we really cannot think "so it seems ok to ...". In mathematics, everything has to be precisely defined, and then whatever (precise) statement you make would be either true or false, and this is not up for debate.

Every item in the sequence clearly has finite number of decimals, and $1/3$ has infinite decimals so it is clearly not included.

Although not totally precise, you are essentially correct here. Precisely, each item in the sequence is equal to $k/10^m$ for some natural numbers $k,m$, but $1/3$ is not. We can prove this: Given any $k,m∈ℕ$ such that $1/3 = k/10^m$, we have $10^m = 3·k$, but $10^m = (9+1)^m$ $≡ 1^m = 1$ (mod $3$), whereas $3·k ≡ 0$ (mod $3$), so this is impossible.

On the other hand the limit of this sequence is $1/3$ so it seems ok to say that it "includes" $1/3$.

You can very well define "include" to mean something in order to make that sequence include its limit. But you must be very clear about the fact that it is not the conventional meaning of "include". So you were correct to use scare-quotes around "include", and it is important to realize how crucial precision is in mathematics.

Here is a standard mathematical notion that you may be interested in. Given any set $S ⊆ ℝ$, we say that $c$ is an adherent point of $S$ iff $c$ is a limit of some sequence from $S$. Given any sequence $f∈ℕ→ℝ$ and any $c∈ℝ$, it turns out that the following are equivalent:

  • $c$ is an adherent point of $\{ \ f(k) : k∈ℕ \ \}$.
  • There is some sequence $i∈ℕ→ℕ$ such that $f∘i$ converges to $c$.
  • Either there is some increasing sequence $i∈ℕ→ℕ$ such that $f∘i$ converges to $c$ or there is some $k∈ℕ$ such that $f(k) = c$.

This notion may be the kind of relationship you are interested in, but it is not a good idea to call it "inclusion".

user21820
  • 60,745
  • 2
    Although it makes sense to say that in this context, but mathematics is about guesswork, and a large part of mathematics starts out from guesswork. Proofs and definitions are a necessary evil to make sure we communicate with each other and don't go crazy. They are not inherently a part of maths. – Trebor Apr 15 '24 at 04:01
  • 2
    @Trebor: Ok ok.. let's compromise. Rigorous mathematics is not about guesswork, though one is free to guess as much as they like about what statements to try proving and what proof ideas to try. – user21820 Apr 15 '24 at 10:13
23

That is a series which converges to $\frac{1}{3}$ which, informally, means it will eventually get arbitrarily close. However, no entry in the series will be exactly $\frac{1}{3}$.

Similarly, the series $3$, $3.1$, $3.14$, $3,141$, $3.1415$ $3.14159$, ... converges to $\pi$ but no term is equal to $\pi$.

As Peter says in the comments, it would be better to say "sequence".

badjohn
  • 8,854
10

A single term of your series can be expressed as

$$A_n=\frac{1}{3}\bigg(1-\frac{1}{10^n}\bigg)$$

Using basic limits, this means that if $n→∞,A_n→\frac{1}{3}$. But according to your definition, $n$ is a countable number. Hence logically,$A_n<\frac{1}{3}$.

If you consider including the value even when n is $\infty$, then you may say it is included in the series, which isn't possible though. Like it depends upon your definition of "inclusion". Hope I was able to answer your question

Gwen
  • 3,910
  • 6
  • 19
  • So if the sequence has countably infinite elements then it would be correct to say that that sequence lists 1/3 because 1/3 also has countably infinite decimals? – psmith Apr 13 '24 at 09:56
  • What do you mean by "countably infinite" @psmith – Gwen Apr 13 '24 at 09:57
  • Perhaps i'm mixing up multiple things but i mean as in this sentence: "The set of all integers is a countably infinite set". – psmith Apr 13 '24 at 10:02
  • 4
    A set is countably infinite if its elements can be put in one-to-one correspondence with the set of natural numbers. Each term in your series you provided can be corresponded to each n being a natural number. But you can't say that "∞" is included in the set of all natural numbers, can you? Similarly you actually can't say that "1/3" is a member of your series. Check my edited answer. – Gwen Apr 13 '24 at 10:08
  • The $n$th element in my sequence corresponds to the $n$th decimal in $0.3333...$ So my sequence includes $1/3$ in the same sense as $0.3333...$ has infinite decimals, no? – psmith Apr 14 '24 at 04:21
  • 3
    @psmith No, definitely not. Your sequence has infinitely many elements of the form $0.333\ldots3$, with finitely many $3$s, but it contains no element which is $0.333\ldots$, with infinitely many $3$s. – Sophie Swett Apr 14 '24 at 12:54
  • @TannerSwett if the sequence has infinitely many elements and each element corresponds to a decimal in 1/3 then we can conclude that there are elements in the sequence with infinitely many decimals, no? What gives? – psmith Apr 16 '24 at 01:50
  • 1
    @psmith Why would we be able to conclude that? The existence of infinitely many finite things does not entail the existence of a single infinite thing. – Sophie Swett Apr 16 '24 at 10:38
-1

As others have pointed out, this is a matter of definition. We might translate the statement that 1/3 is included in the sequence to something a little more precise, such as:

There exists an index such that the element of the sequence at that index is equal to 1/3.

In this case, if we're indexing by natural numbers, then the answer is NO (because it can be shown that every ℕ-indexed element of the sequence is distinct from 1/3).

However, a different choice would be to permit the index ω of a convergent sequence, in which case, the answer is YES, because it can be shown that the sequence converges to 1/3, and, by our definition, the element of the sequence at ω is that limit.

Now, this latter view is, I think, unconventional, but it is a choice that could be made.

  • 1
    Um. And what are the elements around that one? If you allow ω, clearly you must allow ω+1 and ω-1. – Sneftel Apr 15 '24 at 10:53
  • @Sneftel i guess he is talking about the ordinal number ω (omega) that is greater than every natural number. – psmith Apr 15 '24 at 15:13
  • @psmith Exactly. The "infinity" in ordinals isn't the same as the infinity in the limit of a sequence. If you never get there, you never get there. – Sneftel Apr 15 '24 at 15:55
  • @Sneftel Clearly? Again, it's a matter of definition. In this case, I have chosen to allow precisely a natural number, or a symbol ω as an index; nothing more. I do not require this index set to be a Rig, only to be totally ordered. Like I said, I don't claim this is a common definition, but it is a choice that could be made. – MCLooyverse Apr 16 '24 at 14:06
  • Sure. For that matter, you could simply make the choice that in this new system, the seventh element of the sequence is exactly 1/3, arithmetic be damned. My point is that somebody who would actually do math using this would not be able to do useful math if they had made either of those choices, because of the loss of consistency in the axioms underlying the math. – Sneftel Apr 16 '24 at 14:19
  • What is lost by extending the sequence with an unreachable index? It still contains the original sequence exactly, but we've baked in that it converges, and what its limit is. – MCLooyverse Apr 16 '24 at 14:24
  • What’s lost is, now your “infinite sequence” is no longer infinite, because it’s no longer defined for all ordinal numbers. ω is not normally unreachable (just unreachable from natural numbers) but you’ve rendered it so. So you can’t do arithmetic with ordinals anymore, because they no longer follow normal arithmetic rules. – Sneftel Apr 16 '24 at 14:39