As most people know, an integer can only be finite length when expressed in the form of a series of digits in some base. However, real numbers in general can be infinite length, so long as there is a "first digit".

That is, if we express an "infinite integer" in the form $$ \sum_{n=0}^\infty a_n 10^n $$ where $a_n\in\mathbb{Z}_{10}$, then addition and multiplication should both be well-defined on this sequence. Indeed, I suspect that such a system may actually form a field, as additive and multiplicative inverses should exist (consider that $\bar{9}+1=0$ where $\bar{9}$ is the number for which $a_n=9$ for all $n$, and $\overline{285714}3\times7=1$, where $\overline{285714}3$ has $a_0=3$ and then $a_{6n+1}=4$, $a_{6n+2}=1$, and so on).

Has such a set with addition and multiplication been investigated? Does it have a name? Does it actually form a field, or does it fail one of the required properties of a field? Is the system functionally the same irrespective of chosen base, or do different bases change some properties of the system?


You've rediscovered the $10$-adic integers. Such "infinite decimal expansions of integers" form a ring using the usual rules for adding and multiplying decimals. However, they do not form a field. A quick and easy way to see this is that $10$ has no inverse (since if you multiply $10$ by anything, then the last digit of the product will be $0$).

Even worse, however, this ring has zero divisors: there are nonzero elements $x$ and $y$ such that $xy=0$. These take a bit of work to construct, but here's the idea. We'll construct the digits of $x$ and $y$ one at a time, starting from the end. Start by saying the last digit of $x$ is $2$ and the last digit of $y$ is $5$, so the last digit of $xy$ is $0$. Then choose the preceding digit of $x$ so that $x$ is divisible by $4$, and the the preceding digit of $y$ so that $y$ is divisible by $25$, so their product will be divisible by $100$ and end in two zeroes (for instance, $x$ might end in $12$ and $y$ might end in $25$). Continue choosing digits of $x$ and $y$ one at a time so that $x$ is divisible by every power of $2$ and $y$ is divisible by every power of $5$. Every digit of the product $xy$ will end up being $0$.

You can, of course, do the same thing with a different base instead of $10$, giving you the $b$-adic integers for any integer $b>1$. Unlike the case of finite base expansions, however, you get a genuinely different number system for different values of $b$! (To be precise, it turns out that the number system you get depends only on the set of prime factors of $b$, so for instance the $10$-adic integers are isomorphic to the $50$-adic integers.)

It turns out that whenever $b$ is composite, the $b$-adic integers will have zero divisors, as in the case $b=10$. When $b$ is a prime (say $b=p$), however, the $p$-adic integers are an integral domain (they have no zero divisors). The $p$-adic integers aren't quite a field, though, because $p$ has no inverse. If you adjoin an inverse to $p$, you do get a field, called the $p$-adic numbers and written $\mathbb{Q}_p$. Elements of $\mathbb{Q}_p$ are "base $p$ expansions that can be infinite on the left": that is, formal sums $$\sum_{n=k}^\infty a_np^n$$ where each $a_n$ is an integer between $0$ and $p-1$ and $k\in\mathbb{Z}$ (so $k$ can be negative, allowing finitely many negative powers of $p$ in the sum).

There is a much larger story here: $p$-adic numbers play a huge role in modern number theory, and have a very rich structure. Since it is easily possible to write an entire book on the topic, I'll end my answer here though. You can read a bit more about them on the Wikipedia page I linked at the beginning.

  • $\begingroup$ Thanks. This basically addresses all of what I was hoping to check. I hadn't considered the case of the final digit being a zero (I did suspect that the base would matter, though). I had seen the "p-adic numbers" referred to in other questions/answers, but had never really looked into them (as my expertise is in calculus and mathematical physics). $\endgroup$ – Glen O Mar 5 '17 at 5:04

Yes. What you've defined is called the 10-adic numbers. They don't form a field for two reasons, one of which is that $10$ is not invertible; the other reason is left as an exercise.

Most books will only talk about the $p$-adic numbers for $p$ prime, and this is because the general case turns out to reduce to the prime case, essentially because of the Chinese remainder theorem.

  • $\begingroup$ Thanks. Your answer addressed the most important part of my question, and did so very quickly. I've upvoted your answer. However, Eric's answer provides much more detail and addresses a number of the other questions I asked as well, so I'm afraid I have to accept his answer over yours. $\endgroup$ – Glen O Mar 5 '17 at 5:00

Another fun way that the rings differ depending on the base used is that in base 5, $-1$ (i.e. $\bar{4}$) has a square root while $2$ doesn't, but in base 7 it's the other way around (and $-1$ is of course $\bar{6}$ in that base).

I should note that I haven't proved those, but my messing around with the idea pointed strongly that way. If you like programming it's a fun exercise to write up a class that implements these numbers. I did it in Python, as its yield/generator system made it easy to model these numbers as a potentially infinite list that produced the digits from units place up.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.