-3

Is there any way to make sense of expressions with infinitely many digits to the left of the decimal point?

There's a famous proof that $0.999\dots=1$ which starts with $x=0.999\dots$ and derives $10x=x+9$, therefore $x=1$. We can do the same thing with an infinite string of $9$'s to the right of the decimal point: Assume $x=9999\dots99$

Then $10x=9999\dots990$

Then $10x+9=9999\dots999=x$

Finally we get $10x+9=x$

So $x=-1$
$\qquad$ My question is: Is it possible to make sense of the expressions ''$99999\dots9$'' and ''$99999\dots90$'' and the algebraic manipulations above? If not, why does this work for $x=0.999\dots$ and not for $x=999\dots$?

Andrew Li
  • 431
  • 2
    Pay attention. You can't work with $9999...$ the same way you work with $0.9999...$ – Or Shahar Jul 15 '22 at 14:03
  • 1
    $99999$$\cdots$ looks as if it is bigger than any positive integer – Henry Jul 15 '22 at 14:03
  • 2
    No, we can't... – Mauro ALLEGRANZA Jul 15 '22 at 14:06
  • 2
    The sequence of infinitely many 9's does not mean anything at all. Whenever you deal with infinitely long expressions you are really dealing with limits. That's how you define the meaning of $0.999\ldots$. – Ethan Bolker Jul 15 '22 at 14:09
  • 3
    Note that $0.999...$ is a sum of the form $\Sigma_{k=1}^n 9 \cdot 10^{-k}$, which converges while $999....$ would be the sum $\Sigma_{k=1}^n 9 \cdot 10^{k}$ which diverges to positive infinity. – CyclotomicField Jul 15 '22 at 14:09
  • Your $10x+9=99999\cdots=x$ looks unjustified. If this were meaningful, you would have $10x+9=99999\cdots9$ where you would know the final digit while with $99999\cdots$ you would not – Henry Jul 15 '22 at 14:11
  • Are you suggesting that $10x+9=x$ is not true for $x=\infty$? – Henry Jul 15 '22 at 14:13
  • Let $x=10^n-1$, then $10x=10^{n+1}-10$, $10x+9=10^{n+1}-1 \neq x$ – Ivan Kaznacheyeu Jul 15 '22 at 15:40
  • However, if $x=\lim_{n\rightarrow\infty} 10^{n}-1$, then you do get to conclude $10x+9=x$. In the standard metric on $\mathbb{R}$ that limit is infinite. With respect to the 10-adic metric, the limit is $-1$. – Dark Malthorp Jul 17 '22 at 21:54
  • 1
    I think the question is now clear, asking why this works with $0.999\dots$ but not $999\dots$ – Dark Malthorp Jul 18 '22 at 13:57

2 Answers2

1

The problem with these equations is the "$\dots$". In infinite expressions such as the infamous $0.9999\dots$, the ellipsis represents a limit. In the decimal case, what it means is the limit of the sequence $$ 0.9, 0.99, 0.999, \text{ and so on} $$ which you can prove converges in various ways, for example it is trivially increasing, and is bounded above by $1$. The manipulations that follow are valid assuming the limit exists. Let $x=0.9999\dots$ (i.e., the limit of the above sequence). Then:\begin{eqnarray} x &=& 0.9999\dots\\ 10x&=&9.9999\dots\\ 10x&=&9+x\\ x &=& 1 \end{eqnarray} Now, we turn to your expression $999\dots$. The ellipsis here would indicate that this is meant to be the limit of the sequence $$ 9, 99, 999, \text{ and so on} $$ Of course, the limit doesn't exist under the standard metric on $\mathbb{R}$, but if you pretend you don't know that, you can do the kind of manipulations you wrote out and say that if the limit $999\dots$ exists, it is $-1$.

It looks very similar to an expression for a $p$-adic number, which are typically represented by an infinite string of base-$p$ digits for some prime $p$ (see this question and its answers for why it should be a prime). In $p$-adic spaces, you can do those sorts of manipulations you wrote down and make meaningful sense of them. Wikipedia presents the nice example of the $5$-adic $$ \dots1313132 = \frac13 $$ Indeed, you can represent any rational number as an infinite string of base $p$ digits. So, for example, in $5$-adics again, you can write $$ \dots 44444 = -1 $$ which follows from the same kind of manipulations you did: Let $x = \dots 4444$. Then (keep in mind I'm in base $5$ here):\begin{eqnarray} x &=& \dots 4444\\ 10x &=& \dots 4440\\ 10x+4&=&\dots 4444 = x\\ 4x +4&=&0\\ x &=&-1 \end{eqnarray}

0

No, you cannot express numbers that way. Technically, you could let $x=99999\dots$, but it makes absolutely no sense to write $10x = 99999\dots0$ since there are an infinite number of $9$'s before the $0$. Even if you could, solving $10x+9=x$ involves subtracting $x$ from both sides. Since $x$ is infinite, $10x+9$ is also infinite and you cannot subtract one infinity from another like that.

bobeyt6
  • 1,272
  • 1
    "but it makes absolutely no sense to write $10x = 99999...0$ since there are an infinite number of 9's before the 0" -- a priori there is no reason why you couldn't have numbers (or some other mathematical object) given by a sequence of digits indexed by $\omega + 1$. – Mees de Vries Jul 15 '22 at 14:12
  • @MeesdeVries but then I would have thought $99999\cdots9 \not= 99999\cdots$ – Henry Jul 15 '22 at 14:15
  • Well, maybe. Since these things are not yet defined in any way, who knows what it means to multiply one by 10? – Mees de Vries Jul 15 '22 at 14:16
  • 1
    I guess what I'm trying to say is that in writing $9999\dots0$ we are assuming $99999\dots$ has an ending point which it does not since there an infinite number of $9$'s – bobeyt6 Jul 15 '22 at 14:16
  • 1
    @bobeyt6, it does not imply that "999..." has an ending point, any more than writing ordinals as $0, 1, 2, 3, \ldots, \omega, \omega+1, \omega+2, \ldots, \omega \cdot 2, \ldots$ implies that there are finitely many natural numbers. – Mees de Vries Jul 15 '22 at 14:17
  • 1
    @MeesdeVries writing $99999\dots0$ means that there is a $0$ at the end of the infinite number of $9$'s – bobeyt6 Jul 15 '22 at 14:18
  • Yes, and that's fine. You can have an infinite linear order followed by another larger element. Consider writing down a subset of the rational numbers ${0, 1/2, 2/3, 3/4, 4/5, 5/6, 6/7, \ldots, 1}$. Does that notation suggest that there is a largest natural number? – Mees de Vries Jul 15 '22 at 14:23
  • 'you cannot subtract one infinity from another like that.' But in $x=0.999...$ when you multiply x by 10, you get $10x=9.999...$ Then $10x-x$=$9$. Here you just subtract one infinity from another
    @bobeyt6
    – Andrew Li Jul 15 '22 at 14:37
  • 1
    This is different since $0.999\dots$ is not infinite even if it has an infinite number of $9$'s – bobeyt6 Jul 15 '22 at 14:39
  • what about $x=0.8888...$ $10x=8.8888$ Then $10x-x$=$8$ – Andrew Li Jul 15 '22 at 14:42
  • 1
    In which $x=\frac89$ which matches its decimal expansion. My point is is that you cannot subtract one infinity from another but you can subtract one convergent sum from another, such as $0.999\cdots$ or $0.888\cdots$, – bobeyt6 Jul 15 '22 at 14:44
  • (+1) , although $x=9\cdots 9$ with infinite many nines should already be avoided. – Peter Jul 16 '22 at 07:58