> What number should you multiply by? Whatever you pick, there will be a digit in that base, but all the others are in a different base.
I feel like it would be easier to capitalize on the intuition from the circle diagram by abandoning that multiplication idea, and just taking a limit. In particular, consider the sequence .1, .12, .123 etc. in this mixed-base system, and the differences of those values from 1; and show how to calculate the difference for a given term of the sequence; and show that this goes to 0 in the limit.
The multiplication trick is cool and all and it lets you skip trying to explain limits to people. But if you draw that circle you're already 100-epsilon% of the way to explaining limits…
I think the key to this is that one should not look at 0.99999999.... as a regular decimal or a number in itself. Unlike finite decimals like 0.1235, infinitely continuing ones don't have an immediate physical interpretation. 0.9999...is just shorthand for an infinite process that in turn comes about due to the idiosyncracies of allowing infinity in the first place.
A lot of people just get confused because numbers like 0.23, 0.8, are intuitive and thus something that looks syntactically similar like 0.9999..... also feels like it should be intuitive.
But it's not. It's just something that crosses the border from the concrete into the much more abstract. Just an infinite process. Just 1.
0.9… should not be seen as any kind of process. It’s simply another way of spelling 1, just like 4/6 is another way of spelling 2/3, as is -2/-3. There’s nothing special, mysterious, or non-numeric about it. It’s simply another, redundant encoding of the same number.
Any place-value numeral is a shorthand for writing a particular geometric series. Only considering numbers between 0 and 1:
0.abc… = a*10^-1 + b*10^-2 + c*10^-3 + …
If there exists one last nonzero digit in the place-value representation, there is one last nonzero term in the series, and the sum is trivial to compute.
But if the place-value string is infinite? Well, we know that infinite series can have finite sums, and in particular all of these place-value series converge and have finite sums [exercise: show why].
Specifically, the sum of a converging infinite series, and thus the value of the corresponding place-value expression, is defined as the limit of its finite prefix sums. The general formula for the sum of a converging geometric series such as 0.9… is
S = a / (1 - r)
where a is the first term and r the ratio of two consecutive
terms. In this case S = 0.9…, a = 0.9 and r = 0.1. So
0.9… = 0.9 / (1 - 0.1)
= 1
Intuitively, the prefix sums of 0.9… are
0.9
0.99
0.999
0.9999
...
and it’s "clear" that the limit of this sequence at infinity is 1. And because 0.9… is by definition that limit,
This is a great explanation... I guess it hinges on "that infinite series can have finite sums", which seems to be the source of confusion here is for people (like me) coming from the programming and non-mathematical realm, who think about convergence as a process rather than a truth of some sort.
Sure, with that approach but every time I see this kind of explanation I always think that it sort of ignores what most laypeople really mean which I have always found to be 1 - ε which obviously is a less tidy answer and probably teaches more about the limits our understanding of numbers and the limits of mathematics – with real numbers anyway.
So for 0.999... = 1, you have to convince someone that 1 / 10^∞ = 0
Which is true in standard analysis - there are no non-zero infinitesimals - but there are other number systems besides the reals which do have infinitesimals.
While I accept that 0.999... = 1 in standard analysis, it is not at all intuitive to me that 0.999... is a number and not a limit.
---
Suppose we write a function to produce a number with `n` 9s after the decimal point:
f'(n, sum, dec) =
if n == 0 || dec == 0 then sum
else f'(n - 1, sum + dec, dec / 10)
f(n) = f'(n, 0, 0.9)
Then its clear that
f(1) = 0.9
f(2) = 0.99
f(3) = 0.999
...
For any `n`, the result will never reach `1`. Each successively larger `n` only ever closes 9/10ths of the gap. There is no way to close that extra 1/10th of the gap.
By definition of `...` meaning "repeat the previous 3 digits an infinite number of times", we can accept that f(∞) = 0.999... - two different ways to specify the same thing - a non-terminating sequence of 9s after the decimal point.
It's only intuitive to me then that:
lim f(n) = 1
n->∞
Since no single `n` can actually ever produce 1. Even f(∞) < 1, as we didn't include any clause that specifies if n = ∞, add in the additional, 1/10th the difference between the sum and 1.
If you plot f(n), you will see quite clearly that it's an asymptote that approaches 1 as n->∞
If repeated decimals are limits, it makes perfect sense. The claim that `0.999...` is not a limit, but just a number, is why its unobvious that 0.999... = 1
Perhaps the act of cutting a quantity into thirds is physically intuitive, but its representation as 0.333... as driven by the infinite series 3/10 + 3/100 + 3/1000 + ... = 1/3 is, to me anyway, less obviously physical.
That doesn't work with positive things. You don't think of 1 as being the process of adding one of something.
This is why trying to relate math to physical concrete things is folly. Math is inherently abstract. Clinging on to physical representations will ultimately end up limiting your ability to understand it.
Like, try and intuit imaginary numbers to physical things.
> I think the key to this is that one should not look at 0.99999999.... as a regular decimal or a number in itself. Unlike finite decimals like 0.1235, infinitely continuing ones don't have an immediate physical interpretation.
But it's not clear to me why. It's not an estimation like 3.14159... where the sequence would go on forever and I can't write the exact value. I know exactly what 0.1666... is.
The number 0.8 is really 0.8000... so in a sense nothing ever terminates. The only difference between 0.8000... and 0.1666... is that one continues with zeroes and one with sixes.
Think of it like a recursive process which terminates for 0, but produces an infinite recursion for anything else.
repeat_digit(x) = x / 10 + repeat_digit(x / 10)
Since 0 / 10 = 0, 0.0 / 10 = 0, and 0.000 / 10 = 0, and 0 + 0.0 + 0.000 = 0, etc - we know that this produces 0 and could short circuit it to terminate - maybe clearer instead to write:
repeat_digit(x) =
if x == 0 then 0
else x / 10 + repeat_digit(x / 10)
repeat_digit(0) = 0 ;; terminates
repeat_digit(9) = 0.999... ;; never terminates
Decimals are just a hack to make long arithmetic work on fractional amounts. Notation and bases are just a convenient representation, values are the same in all bases.
This is a good opportunity to remind a fun property about decimal expansions (& expansions in other bases). One might intuitively think that real numbers have a unique decimal expansion but the fact that 1.000... = 0.999... disproves it. Still, it's not too difficult to show that every real number admits a unique decimal expansion _unless_ it has one that ends with a sequence of zeros or a sequence of nines. In that case it has exactly two decimal expansions. This shows that 1.000... = 0.999... is the only such quirk, all the other cases are basically the same, 232.128000... = 232.127999... for instance.
Why? The property you describe isn't special to phi in any way. Phi is defined by the equation 1/φ = φ - 1. (It's the positive root.) And therefore, in base φ, 0.1 is necessarily equal to 10 - 1, which means that 10 = 1 + 0.1, or 10 = 1.1.
But obviously this generalizes. You could take the positive root of 1/ψ = ψ - 3, which is 3.3027756377... , more than 3.
In that case, by definition, 10 = 3.1 in base ψ. But why is that interesting? And why is it more interesting if we change the 3 in the definition of our base to be another 1?
I've always found this alternate way more intuitive:
⅓ = 0.333...
⅓ × 3 = 1
0.333... × 3 = 0.999...
I've been able to use this method to convince fifth graders who had limited algebra experience. (Instead of ×3, I used ⅓ + ⅓ + ⅓. The fifth graders knew multiplication but using addition made it slightly clearer, I thought.)
Neither of these are much of a proof, though. They're appeals to intuition that, in this instance, turn out to be correct. There are other appeals to intuition with infinite sequences / series that turn out to be dead wrong, so I feel that this is teaching the wrong thing.
I would say this is a good example where it's geometrically "proven" that pi = 4. It's convincing enough where if someone didn't already learn the value of pi, I might expect them to believe this proof.
The idea is that if you circumscribe a square around a unit circle and take its perimeter, you obtain 4. If you "invert" the corners of the square until they hit the circle, the perimeter remains unchanged. If you iterate infinitely, the sequence of polygons eventually converges to the circle.
In this case, the sequence truly does converge to a circle. However, the subtle detail is that, in the limit, this manner of measuring perimeter is the taxicab metric (and NOT the Euclidean metric), where pi actually does equal 4. It's not so obvious geometrically, but it shows that there's a distinction between a sequence of curves and the corresponding sequence of lengths.
EDIT: Though, having said all this, I feel the proof given about 0.999... equaling 1 is rigorous, but I suppose there's some subjectivity involved.
I haven't read or understood the author's proof, but the "standard" proof probably goes like this (any corrections? anyways though, I have to admit this approach sucks all of the joy out of the blog post)
We take the standard definition of the reals as equivalence classes of Cauchy sequences.
Let a_n be the mixed-base number up to n digits, e.g. a_4 = 0.1234. Clearly it's Cauchy
Let b_n = 1 - a_n, e.g. b_4 - 0.00001
Obviously, lim n->inf (a_n + b_n) = 1
But since the sequences a and b are Cauchy, we can exchange limit and sum
Here's how I would prove the same fact, which I think is better and more elegant. If we were to use base 12, that would mean that 1/3 would be .4 instead of .333... which completely eliminates the confusion. .999... is just three thirds, in base 12 again .4 +.4 +.4 = 1. The whole .333 repeating thing and by extension .999... is just a result of the dumb decision to use base 10 (dumb because if you have a system for counting objects, it makes more sense to use a grouping that is divisible as many ways as possible while being small enough to grok)
All these proofs that focus on a particular example with 0.999… or 0.333… are neat but they eventually obscure the real explanation: 0.999… is just notation which by definition refers to the number you are approaching as you add digits. Everyone will agree that 0.999… is approaching 1 so there's no argument anymore.
TL;DR: as often in math, lay people struggle because they don't start with clear definitions.
The trouble is that definition isn't clear. We're told that `...` means "repeat the last 3 digits" - this isn't the same as saying "it refers to the number it approaches" - in fact, mathematicians are the source of the confusion because they refuse to admit that repeating a decimal ∞ times is a way of expressing a limit.
For any finite N, we can say:
repeat_N_times(n : Nat, sum : Real, decimal : Real) =
if n == 0 or decimal == 0 then 0
else repeat_N_times(n - 1, sum + decimal, decimal / 10)
f(n, decimal) = repeat_N_times(n, 0, decimal)
Obviously, if we give ∞ for n when the decimal is nonzero, the function never terminates, but, the sum also never reaches 1 - since each successive call only closes the gap by 9/10th the difference between the previous sum and 1.
So, we can specify that
lim f(n, 0.9) = 1
n->∞
In other words,
0.999... = lim f(n, 0.9) = 1
n->∞
But f(∞, 0.9) and 0.999... are two equivalent ways of expressing an "infinitely repeated 9" after the decimal point.
So does `...` refer to an infinitely repeated decimal, or does it not? Are f(∞, 0.9) and 0.999... equivalent?
If not, then 0.999... is clearly not just a "repeated decimal", because that's what `f(∞, _)` is - by inductive definition.
So, we should either stop calling `...` a "repeating decimal" or accept that repeating decimals represent limits.
If we accept that repeating decimals are limits (they are), then nobody need challenge that 0.999... = 1 - it's obviously true.
Tell that to a mathematician and they say: NO! Repeating decimals are not limits!
> in fact, mathematicians are the source of the confusion because they refuse to admit that repeating a decimal ∞ times is a way of expressing a limit.
Nonsense. 0.999... = 1 and other repeating decimals are commonly mentioned or shown when teaching limits and geometric series. I am a mathematician, and have taught exactly this many times, as have my colleagues.
Technically the repeating part of the repeating decimal is notation for a geometric series with a negative power of 10 as the common ratio, and the number being represented is the sum of the series, which is the limit of the sequence of partial sums (a limit!). Describing repeating decimals as series rather than limits does seem clearer though (the limit is of a sequence you're not explicitly mentioning or seeing).
It's also true that repeating decimals are rational numbers, a fact often taught well before limits. This is often a simpler way to think about them, and is what the algebraic proof at that of the article uses, showing 0.999... = 9/9 = 1.
At uni in Statistics 101 me and my friend quickly found out that when the professor turned away from the blackboard and asked students what P(something) was, most common answer was 0 or 1.
"Mixed base"?! I mean, it's a neat looking thing, 0.12345... but it's a tautology that if every digit represents the next base, and every digit increases, you're approaching 1.
Hah! What a wonderful observation. I feel like this is a test; the previous statement seemed obvious to me, but now I'm past anything I can visualize. My gut feeling is that your train arrives at the same moment at an infinitely distant station as a function, even though algorithmically, it would always be trailing.
I should clarify that I have no idea whether 0.999... === 1. I'm kinda guessing it depends on how you arrived at that 0.999... but that's probably my own floating point survivor's bias creeping in.
Well it would have to because it’s bounded (must be less than 1¹) but continuously increasing.
⸻
1. That’s a weak upper bound. I hesitate to give a strong bound because from this view of the discussion, I can’t even see what I wrote and I’m too lazy to deal with it. This sort of thing is covered first in Calc II (or Calculus BC if you were fortunate enough to have it at your high school—I was a snotty kid and took the BC test and passed it even though I only had an AB class, so this is one of many weaknesses I have mathematically) and then again, assuming you take it in either real analysis or mathematical analysis.²
2. Fun fact, I did my undergraduate at the Claremont Colleges where what would be 100 classes are numbered 1–49, 200 classes 50–99, 300 classes 100–149 and 400 classes 150–199 with graduate level classes being 200+. My first undergraduate class, was titled “Fundamentals of Mathematics” and numbered 103 and my grad school advisor thought it was a remedial class and not the in-depth introduction to advanced mathematical proof techniques that it was.
It is not a "fact" that .999... is 1. You are assuming (!) that applying "+" or "-" to infinite terms is 'defined' but clearly it is undefined behavior.
This is trivial to see: We add d=0.000....1 to .999... Here we add 0.000001.
0.999... + d = 1.000000999....
1.0 + d = 1.0000010000000
Now the "fact" that .999... is "1" seems contradicted given that 1+d - (.999... +d) =/= 0
Moral of the story: don't forget all formalisms have a bounded domain and there is no such thing as a universal operation.
p.s. the attempt to mystify the obvious misuse of the mathemtical functions and number representation by the progressively shifting bases in the second example is relying in the same phenomena when misapplying these arithmatic operations beyond their applicable domain.
To be perfectly clear: the 'undefined' operation that the OP relies on is subtracting .999... from 9.999... He asserts this results in 9 plus a 0/zero residual value. That 'residual value' being zero is suggested merely by the graphic representation of .999.... that was "shifted". In actual reality, the subtraction makes as much sense as subtracting infinity from infinity and claiming result is zero.
> This is trivial to see: We add d=0.000....1 to .999... Here we add 0.000001.
0.000...1 =/= 0.000001
> 0.999... + d = 1.000000999...
> 1.0 + d = 1.0000010000000
> 1+d - (.999... +d) =/= 0 [claiming that 1 + d =/= 0.999... + d]
This is the same claim as 0.999... =/= 1, so your argument is tautological. 1+d =/= 0.999...+d if and only if 1 =/= 0.999...; this isn't a proof, you just moved the number a few decimal places to the right and then claimed it was proven.
It's follows as an elementary extension of the problem in the original post.
Given that we've proved there that 0.99... = 1, we can use this fact to prove that any number with a repeating set of 9s after the decimal place n is equal to the number with those 9s removed and the n-1th place incremented by one.
Take this example:
1.09... ?= 1.1
10 * (1.09...) ?= 10 * 1.1
10.9... ?= 11
10 + .9... ?= 11
10 + 1 ?= 11 --> trivially true.
It just requires the exact same manipulation trick used in the original proof.
The original post is playing with your head. Let's do grown up math.
We define L_0, limit to zero in the Real domain. This is as close as you can get to 0. We define L_1, limit to(wards) 1 as that which leaves a difference of L0 to 1. I.e. L_1 := 1 - L_0.
This L_1 is what is 'pictorially' (even in the mind's eye) imagined as representable by an infinite sequence of 9s in 0.999.... But it has a precise "value": L_1. It is informative here to recognize that we can not do the same for L_0, since that is 0.000.... and that 'inability to visualize' right there should be meta information for you that 'this is a formalism that is being pushed to its limits'. But we also have a precise definition for value of L_0 as noted. And we have defined L_1 in terms of (a 'graspable' 1) and an hypothetical L_0.
Now (N x L_1) -- which requires the 'shifty' shifting the digits bit in the mathematic schtick here to "visualize" -- also has a 'precise' value:
N x L_1 = N x (1 - L_0) = N - NxL_0.
So we now do the magic trick of substracting x (which is L_1) from Nx, that is N x L_1. Amazing: Nx - x where x is L_1 gives us 9 - 9 x L_0, that is 9x! But in the 'pictures' the '9 x L_0' terms is no longer seen! Hmm. Is that possibly because we can't "visually shift" L_0 ? Is that possibly because we can't, using the conceptual lens of digital numbers, 'see' 9 times L_0?
Another way of looking at this is to ask if you are OK with considering L_0 to be 1/ℵ0 . and .999... another way of saying 1 - 1/ℵ0. Then multiplying the number by 10 gives rise to things like '10 - 10/ℵ0'.
Does 1/ℵ0 = n/ℵ0 = 0 in your mathematical universe? If yes, then there is nothing to see here folks.
Otherwise, we have a case of someone trying to pass off 9-9/ℵ0 (which 'looks like' 9.0000000000.....?) as 9.
The nerve.
p.s. you are correct in the sense of a simple extention of the original approach. I hope however it is clear now that the entire trick is resting on tautologies and 'misrepresentation' of values. This is partly I think due to notation. We really should make a distinction between say 1/3 (0.333...) and numbers that deal with infinities and infinitesimals. For example, if we were in the habit of writing these infinite terms as (L_0): '.000...?' and (L_1): '0.999...?') then even pictorially, the original multiplication and substractions would have had terms ending with '...?'. That would clearly distinguish '9.000...' from '9.000...?'.
I'm disappointed that everyone here is only talking about the 0.9999... equals 1 bit, and not talking about the really fun part, which is the expansion where the n^{th} digit is in base (n+1).
I'm starting to feel like no one is reading past the initial section of the post.
Did you read beyond the first paragraph? This is about 0.1234567... where the n-th digit is written in base n. I found the visual proof more interesting since it is quite intuitive. And also unlike 0.(9), for which you have to accept some unconventional arithmetic facts as part of the reasoning.
This is purely a fact of notation though. 0.999... is defined to be a "destination" and not an ongoing process such as "lim".
So you cannot take 0.999... to the infinity-th power to get it to zero, you still get one. The issue is that you need to couple the exponent to the length of digits because the rate at which the exponent grows decides whether you go to zero or one.
No, that’s not the definition of the … in a repeating decimal. The definition is the limit of the sequence .9, .99, .999, … as the number of digits reaches infinity.
I feel like it would be easier to capitalize on the intuition from the circle diagram by abandoning that multiplication idea, and just taking a limit. In particular, consider the sequence .1, .12, .123 etc. in this mixed-base system, and the differences of those values from 1; and show how to calculate the difference for a given term of the sequence; and show that this goes to 0 in the limit.
The multiplication trick is cool and all and it lets you skip trying to explain limits to people. But if you draw that circle you're already 100-epsilon% of the way to explaining limits…
A lot of people just get confused because numbers like 0.23, 0.8, are intuitive and thus something that looks syntactically similar like 0.9999..... also feels like it should be intuitive.
But it's not. It's just something that crosses the border from the concrete into the much more abstract. Just an infinite process. Just 1.
Any place-value numeral is a shorthand for writing a particular geometric series. Only considering numbers between 0 and 1:
If there exists one last nonzero digit in the place-value representation, there is one last nonzero term in the series, and the sum is trivial to compute.But if the place-value string is infinite? Well, we know that infinite series can have finite sums, and in particular all of these place-value series converge and have finite sums [exercise: show why].
Specifically, the sum of a converging infinite series, and thus the value of the corresponding place-value expression, is defined as the limit of its finite prefix sums. The general formula for the sum of a converging geometric series such as 0.9… is
where a is the first term and r the ratio of two consecutive terms. In this case S = 0.9…, a = 0.9 and r = 0.1. So Intuitively, the prefix sums of 0.9… are and it’s "clear" that the limit of this sequence at infinity is 1. And because 0.9… is by definition that limit,https://en.wikipedia.org/wiki/Cauchy_sequence
Particularly https://en.wikipedia.org/wiki/Construction_of_the_real_numbe...
---
Intuitively:
Then for an infinite number of decimal places: So for 0.999... = 1, you have to convince someone that 1 / 10^∞ = 0Which is true in standard analysis - there are no non-zero infinitesimals - but there are other number systems besides the reals which do have infinitesimals.
While I accept that 0.999... = 1 in standard analysis, it is not at all intuitive to me that 0.999... is a number and not a limit.
---
Suppose we write a function to produce a number with `n` 9s after the decimal point:
Then its clear that For any `n`, the result will never reach `1`. Each successively larger `n` only ever closes 9/10ths of the gap. There is no way to close that extra 1/10th of the gap.By definition of `...` meaning "repeat the previous 3 digits an infinite number of times", we can accept that f(∞) = 0.999... - two different ways to specify the same thing - a non-terminating sequence of 9s after the decimal point.
It's only intuitive to me then that:
Since no single `n` can actually ever produce 1. Even f(∞) < 1, as we didn't include any clause that specifies if n = ∞, add in the additional, 1/10th the difference between the sum and 1.If you plot f(n), you will see quite clearly that it's an asymptote that approaches 1 as n->∞
If repeated decimals are limits, it makes perfect sense. The claim that `0.999...` is not a limit, but just a number, is why its unobvious that 0.999... = 1
Even infinite irrational numbers have physical representation, like pi.
Is -1 physically intuitive? Have you ever tried to have -1 of something?
This is why trying to relate math to physical concrete things is folly. Math is inherently abstract. Clinging on to physical representations will ultimately end up limiting your ability to understand it.
Like, try and intuit imaginary numbers to physical things.
Yes, I do. why not? Maps nicely to Peano arithmetic.
MIDWIT: Nooo, infinite decimals are real numbers just like any other! They all have a perfectly reasonable physical interpretation!
ENLIGHTENED: Real numbers are unphysical
0.23 == 0.23000…
As another commenter stated, use a different base than can exactly represent one third and the whole thing becomes a non-issue.
But it's not clear to me why. It's not an estimation like 3.14159... where the sequence would go on forever and I can't write the exact value. I know exactly what 0.1666... is.
The number 0.8 is really 0.8000... so in a sense nothing ever terminates. The only difference between 0.8000... and 0.1666... is that one continues with zeroes and one with sixes.
- "tens" place: 1.61803
- ones place: 1
- decimals place: 0.61803
So in base phi, you could write Phi as EITHER 10.0, or as 1.1 - they both add up to 1.61803!
Most interesting non integer base, I think.
Why? The property you describe isn't special to phi in any way. Phi is defined by the equation 1/φ = φ - 1. (It's the positive root.) And therefore, in base φ, 0.1 is necessarily equal to 10 - 1, which means that 10 = 1 + 0.1, or 10 = 1.1.
But obviously this generalizes. You could take the positive root of 1/ψ = ψ - 3, which is 3.3027756377... , more than 3.
In that case, by definition, 10 = 3.1 in base ψ. But why is that interesting? And why is it more interesting if we change the 3 in the definition of our base to be another 1?
I've always found this alternate way more intuitive:
⅓ = 0.333...
⅓ × 3 = 1
0.333... × 3 = 0.999...
I've been able to use this method to convince fifth graders who had limited algebra experience. (Instead of ×3, I used ⅓ + ⅓ + ⅓. The fifth graders knew multiplication but using addition made it slightly clearer, I thought.)
Except if you start rearranging infinite series that don't converge absolutely, then things get nasty.
Could you give an example? This sounds interesting, both in context and on its own.
I would say this is a good example where it's geometrically "proven" that pi = 4. It's convincing enough where if someone didn't already learn the value of pi, I might expect them to believe this proof.
The idea is that if you circumscribe a square around a unit circle and take its perimeter, you obtain 4. If you "invert" the corners of the square until they hit the circle, the perimeter remains unchanged. If you iterate infinitely, the sequence of polygons eventually converges to the circle.
In this case, the sequence truly does converge to a circle. However, the subtle detail is that, in the limit, this manner of measuring perimeter is the taxicab metric (and NOT the Euclidean metric), where pi actually does equal 4. It's not so obvious geometrically, but it shows that there's a distinction between a sequence of curves and the corresponding sequence of lengths.
EDIT: Though, having said all this, I feel the proof given about 0.999... equaling 1 is rigorous, but I suppose there's some subjectivity involved.
Someone who doesn't believe the former won't believe the latter.
There are also non-standard analyses where 0.999... ≠ 1 - when you include non-zero infinitesimals like with hyperreal/surreal numbers.
Discussed April 28, 2020, 626 comments: https://news.ycombinator.com/item?id=23004086
We take the standard definition of the reals as equivalence classes of Cauchy sequences.
Let a_n be the mixed-base number up to n digits, e.g. a_4 = 0.1234. Clearly it's Cauchy
Let b_n = 1 - a_n, e.g. b_4 - 0.00001
Obviously, lim n->inf (a_n + b_n) = 1
But since the sequences a and b are Cauchy, we can exchange limit and sum
1 = lim n->inf (a_n) + lim n -> inf (b_n)
But the second limit = 0, therefore the first = 1
So your comment doesn't really address the main point, and why it's surprising.
TL;DR: as often in math, lay people struggle because they don't start with clear definitions.
For any finite N, we can say:
So eg, Obviously, if we give ∞ for n when the decimal is nonzero, the function never terminates, but, the sum also never reaches 1 - since each successive call only closes the gap by 9/10th the difference between the previous sum and 1.So, we can specify that
In other words, But f(∞, 0.9) and 0.999... are two equivalent ways of expressing an "infinitely repeated 9" after the decimal point.So does `...` refer to an infinitely repeated decimal, or does it not? Are f(∞, 0.9) and 0.999... equivalent?
If not, then 0.999... is clearly not just a "repeated decimal", because that's what `f(∞, _)` is - by inductive definition.
So, we should either stop calling `...` a "repeating decimal" or accept that repeating decimals represent limits.
If we accept that repeating decimals are limits (they are), then nobody need challenge that 0.999... = 1 - it's obviously true.
Tell that to a mathematician and they say: NO! Repeating decimals are not limits!
Nonsense. 0.999... = 1 and other repeating decimals are commonly mentioned or shown when teaching limits and geometric series. I am a mathematician, and have taught exactly this many times, as have my colleagues.
Technically the repeating part of the repeating decimal is notation for a geometric series with a negative power of 10 as the common ratio, and the number being represented is the sum of the series, which is the limit of the sequence of partial sums (a limit!). Describing repeating decimals as series rather than limits does seem clearer though (the limit is of a sequence you're not explicitly mentioning or seeing).
It's also true that repeating decimals are rational numbers, a fact often taught well before limits. This is often a simpler way to think about them, and is what the algebraic proof at that of the article uses, showing 0.999... = 9/9 = 1.
I should clarify that I have no idea whether 0.999... === 1. I'm kinda guessing it depends on how you arrived at that 0.999... but that's probably my own floating point survivor's bias creeping in.
⸻
1. That’s a weak upper bound. I hesitate to give a strong bound because from this view of the discussion, I can’t even see what I wrote and I’m too lazy to deal with it. This sort of thing is covered first in Calc II (or Calculus BC if you were fortunate enough to have it at your high school—I was a snotty kid and took the BC test and passed it even though I only had an AB class, so this is one of many weaknesses I have mathematically) and then again, assuming you take it in either real analysis or mathematical analysis.²
2. Fun fact, I did my undergraduate at the Claremont Colleges where what would be 100 classes are numbered 1–49, 200 classes 50–99, 300 classes 100–149 and 400 classes 150–199 with graduate level classes being 200+. My first undergraduate class, was titled “Fundamentals of Mathematics” and numbered 103 and my grad school advisor thought it was a remedial class and not the in-depth introduction to advanced mathematical proof techniques that it was.
It is not a "fact" that .999... is 1. You are assuming (!) that applying "+" or "-" to infinite terms is 'defined' but clearly it is undefined behavior.
This is trivial to see: We add d=0.000....1 to .999... Here we add 0.000001.
0.999... + d = 1.000000999.... 1.0 + d = 1.0000010000000
Now the "fact" that .999... is "1" seems contradicted given that 1+d - (.999... +d) =/= 0
Moral of the story: don't forget all formalisms have a bounded domain and there is no such thing as a universal operation.
p.s. the attempt to mystify the obvious misuse of the mathemtical functions and number representation by the progressively shifting bases in the second example is relying in the same phenomena when misapplying these arithmatic operations beyond their applicable domain.
To be perfectly clear: the 'undefined' operation that the OP relies on is subtracting .999... from 9.999... He asserts this results in 9 plus a 0/zero residual value. That 'residual value' being zero is suggested merely by the graphic representation of .999.... that was "shifted". In actual reality, the subtraction makes as much sense as subtracting infinity from infinity and claiming result is zero.
Lets suppose 1/10^∞ is a number (obviously not a real number, but a number nonetheless). Call it ϵ
Now, lets look at this common "proof", using this There's a flaw here on the second line: The assumption that 10 * 0.999... = 9.999... In the original proof, they're silently adding 9ϵ to the result when multiplying by 10. Which of course is incorrect. 10*(1 - ϵ) = 10 - 10ϵ, by usual mathematical convention. Where did the additional 9ϵ magically appear from?The original "proof" assumes that ϵ = 0 and therefore 10ϵ = ϵ
But if ϵ is not 0, then the proof falls apart, because 10ϵ != ϵ
So 0.999... = 1 iff ϵ = 0
Does 1/10^∞ = 0?
For real numbers, it's just a given. You must simply accept it and move on.
0.000...1 =/= 0.000001
> 0.999... + d = 1.000000999...
> 1.0 + d = 1.0000010000000
> 1+d - (.999... +d) =/= 0 [claiming that 1 + d =/= 0.999... + d]
This is the same claim as 0.999... =/= 1, so your argument is tautological. 1+d =/= 0.999...+d if and only if 1 =/= 0.999...; this isn't a proof, you just moved the number a few decimal places to the right and then claimed it was proven.
These two values are equal: 1.000000999... --> 1.000001
prove it.
Given that we've proved there that 0.99... = 1, we can use this fact to prove that any number with a repeating set of 9s after the decimal place n is equal to the number with those 9s removed and the n-1th place incremented by one.
Take this example:
1.09... ?= 1.1
10 * (1.09...) ?= 10 * 1.1
10.9... ?= 11
10 + .9... ?= 11
10 + 1 ?= 11 --> trivially true.
It just requires the exact same manipulation trick used in the original proof.
We define L_0, limit to zero in the Real domain. This is as close as you can get to 0. We define L_1, limit to(wards) 1 as that which leaves a difference of L0 to 1. I.e. L_1 := 1 - L_0.
This L_1 is what is 'pictorially' (even in the mind's eye) imagined as representable by an infinite sequence of 9s in 0.999.... But it has a precise "value": L_1. It is informative here to recognize that we can not do the same for L_0, since that is 0.000.... and that 'inability to visualize' right there should be meta information for you that 'this is a formalism that is being pushed to its limits'. But we also have a precise definition for value of L_0 as noted. And we have defined L_1 in terms of (a 'graspable' 1) and an hypothetical L_0.
Now (N x L_1) -- which requires the 'shifty' shifting the digits bit in the mathematic schtick here to "visualize" -- also has a 'precise' value:
N x L_1 = N x (1 - L_0) = N - NxL_0.
So we now do the magic trick of substracting x (which is L_1) from Nx, that is N x L_1. Amazing: Nx - x where x is L_1 gives us 9 - 9 x L_0, that is 9x! But in the 'pictures' the '9 x L_0' terms is no longer seen! Hmm. Is that possibly because we can't "visually shift" L_0 ? Is that possibly because we can't, using the conceptual lens of digital numbers, 'see' 9 times L_0?
Another way of looking at this is to ask if you are OK with considering L_0 to be 1/ℵ0 . and .999... another way of saying 1 - 1/ℵ0. Then multiplying the number by 10 gives rise to things like '10 - 10/ℵ0'.
Does 1/ℵ0 = n/ℵ0 = 0 in your mathematical universe? If yes, then there is nothing to see here folks.
Otherwise, we have a case of someone trying to pass off 9-9/ℵ0 (which 'looks like' 9.0000000000.....?) as 9.
The nerve.
p.s. you are correct in the sense of a simple extention of the original approach. I hope however it is clear now that the entire trick is resting on tautologies and 'misrepresentation' of values. This is partly I think due to notation. We really should make a distinction between say 1/3 (0.333...) and numbers that deal with infinities and infinitesimals. For example, if we were in the habit of writing these infinite terms as (L_0): '.000...?' and (L_1): '0.999...?') then even pictorially, the original multiplication and substractions would have had terms ending with '...?'. That would clearly distinguish '9.000...' from '9.000...?'.
why do you think adding zero changes a number?
Well I didn't ever think about it that way and now I'm blown away. I'm lucky today!
I'm starting to feel like no one is reading past the initial section of the post.
1 - 0.(9) = 0
So you cannot take 0.999... to the infinity-th power to get it to zero, you still get one. The issue is that you need to couple the exponent to the length of digits because the rate at which the exponent grows decides whether you go to zero or one.
0.999... = lim n to inf sum i=1 to n 10^-i
how can you say its not a limit? the value of the limit of partial sums = 1
defined to be a destination is meaningless
Huh? I didn't get that conversion part. How is he doing it?
Like, how can you have a "3" in base 2?
Likewise AB in base 10 is A*10(=10*10)+B(=11)=111.
?