11

I want to know: what is the last digit of $\pi$?

Some people say there are no such thing, but they fail to mention why.

amWhy
  • 209,954
Joel
  • 177
  • 3
    Sometimes even rational numbers don't have a "last digit", think of 1/3=0.33333... – Vhailor Jul 21 '11 at 22:12
  • 58
    Nice Poem. ${}$ – jspecter Jul 21 '11 at 22:13
  • 3
    If $\pi$ has a last digit, then $0.999\ldots\neq 1$. – Asaf Karagila Jul 21 '11 at 22:13
  • @Theo: indeed - I have started a meta question at http://meta.math.stackexchange.com/questions/2628/the-criteria-to-close-a-question Please add your thoughts. – davidlowryduda Jul 22 '11 at 00:39
  • @Theo Curiously this is currently the 16'th hottest question (27 hotness) across all SE sites. – Bill Dubuque Jul 22 '11 at 04:20
  • 9
    This question should specify "base 10". The "no last digit" phenomenon depends on how $\pi$ is represented. To take a contrived setting, base-$\pi$ numbers, then $\pi$ is written as $1$. I'm not trying to be pedantic here: representation is a fundamental part of this question. – Fixee Jul 22 '11 at 06:09
  • 1
    @Vhailor, Surely the last digit of 1/3 is 3? – Jonathan. May 31 '12 at 17:06
  • 16
    @jspecter: A slightly enhanced version:

    $$ \text{I desperately want to know}\ \text{The last digit of $\pi$}\ \text{Some people say there's no such thing}\ \text{They fail to mention why} $$

    – joriki Dec 22 '12 at 03:49
  • I upvote for the fantastic answer by @jspecter. I took a laughter break for that.. – WishingFish Jun 28 '13 at 17:33
  • and OP, perhaps this is exactly why you shouldn't ask mathematics questions in rhyme. @jspecter – WishingFish Jun 28 '13 at 17:48
  • I like the original better @joriki, but good try! – WishingFish Jun 28 '13 at 17:52
  • @Fixee Can you give an example of real numbers represented in a non-integer base, let alone an irrational base, let alone base $\pi$? – Simon S Jul 13 '15 at 23:03
  • @SimonS I think I did above: $\pi = 10_\pi$. Irrational bases do lead to weird results, but I'm not the first one to suggest them. I first encountered the idea of irrational (and transcendental) bases in Knuth's 2nd book, but the idea predates even that: http://mathworld.wolfram.com/Base.html – Fixee Jul 14 '15 at 02:24
  • @Fixee Yes of course an irrational or transcendental base is a theoretical possibility. But it strikes me as useless, not to mention computationally expensive. So what I meant is: does anyone ever use it? – Simon S Jul 14 '15 at 09:01
  • 1
    @SimonS The best way to answer that question would be a web search. My research area is cryptography, so it's too far afield for me to have anything but recreational interest, but a lot of very interesting math isn't that well-used. It doesn't mean it's not cool or fun to think about. For example, here's a question: we know that $\pi$ is irrational in base 10; is 10 irrational in base $\pi$? – Fixee Jul 14 '15 at 18:53
  • Related: http://math.stackexchange.com/questions/446664/what-is-the-most-efficient-numerical-base-system – Simon S Jul 16 '15 at 02:15

5 Answers5

17

There is no "last" digit of $\pi$. If there was a last digit, then there could only be finitely many digits in front so that $\pi$ would be a rational number. However $\pi$ was shown to be irrational in the 18th century by Lambert.

(This Meta.StackExchange post is a joke based on the impossibility of finding such a last digit)

Eric Naslund
  • 72,099
  • 3
    Thanks for the link to the Meta.StackExchange post: hilarious. I've got my laughs in for the day. We all need a good dose of humor now and then! – amWhy Jul 21 '11 at 22:16
9

Since you may have never seen the topics in my colleagues' answers, I'll try to explain them in some detail.

Suppose for the sake of argument that when $\pi$ is written as a decimal expansion ($3.1415 \dots$) it does have a final digit. This would clearly imply that there is a finite number of terms in the expansion. All real numbers with finite decimal expansions can be written in the form $\frac{a}{b}$ where $a$ and $b$ are integers (whole numbers).

By this reasoning we conclude that $\pi = \frac{a}{b}$ for some positive integers $a$ and $b$, i.e., that $\pi$ is rational. This is the starting point for this short proof given by I. Niven in 1946, which is especially easy to follow if you've had a little trigonometry and even less differential calculus. The proof concludes with an absurdity like the existence of an integer between $0$ and $1$, which implies that $a$ and $b$ do not exist and $\pi$ is irrational (and has an infinite decimal expansion). It should be noted that the irrationality of $\pi$ was first established by Lambert in 1761 by studying the continued fraction expansion of the tangent function and using the identity $\tan \frac{\pi}{4} = 1$. More generally, he proved that if $x$ is rational, then $\tan x$ is irrational.

In short, there is no final digit in the decimal expansion of $\pi$ because it is irrational.

user02138
  • 17,064
8

Proving that $\pi$ is irrational is more difficult than proving that $e$ or $\sqrt{2}$ or $\log_2 3$ is irrational. See http://en.wikipedia.org/wiki/Proof_that_pi_is_irrational .

Proving that an irrational number has no last digit is easier than that: http://en.wikipedia.org/wiki/Irrational_number

8

Even rational numbers usually have no "last digit": what is the last digit of $$ 0.1313131\dots = 0.\overline{13} = \frac{13}{99} ? $$

So what sort of numbers have a last digit?

One, numbers with a terminating decimal expansion: numbers like $\displaystyle 2.23627 = \frac{223627}{100000}$. As you can see, all such numbers can be written as a fraction with denominator being a power of $10$.

Two, depending on your definition of "last digit", numbers like $0.4677777\dots = \frac1{100}46.77777$ = $\displaystyle \frac1{100}\left(46 + \frac79\right) = \frac1{100}\frac{421}{9}$. These numbers can be written as $\displaystyle \frac1{10^k} \frac{n}9$ for some integers $k$ and $n$.

So a number $x$ has a "last digit" if and only if $(9\cdot 10^k)x$ is an integer for some $k$. Only very special numbers are of this form, and it should be no surprise that $\pi$ is not. (Admittedly, I don't actually see how to prove this without invoking $\pi$'s irrationality, but it's a much weaker property.)

ShreevatsaR
  • 41,374
4

HINT $\rm\ \pi = 3.1415\ \Rightarrow\ 10^4\: \pi = 31415\ \Rightarrow\ \pi = 31415/10^4\ $ is rational, contra Lambert's proof.

Bill Dubuque
  • 272,048