If you've read about
p-adic integers (and maybe even if you haven't...) you'll know about the
explicit representation of p-adic
integers, as well as how to perform most types of
arithmetic on them.
But p-adic integers are also the completion of the integers (or the natural numbers) in the p-adic norm. So we can perform analysis on p-adics, just like we can perform analysis on real numbers. Of course, our "usual" metric of the absolute value |.| isn't defined on p-adics, so we can't use it. Instead, we use the p-adic norm |.|p.
Most of the results in this writeup may be easily extended also apply to p-adic rationals; when they don't, I'll try to note it.
Take some p-adic d=...d
2d
1d
0. If d were a
natural number (i.e. all but finitely many d
i's were
0) then
vp(d) would be
min {
i : d
i != 0 }. So the same is true for
any p-adic.
Recall that the distance between 2 p-adics x,y is |x-y|p = p-vp(x-y). What this means is that p-adics are "backwards": x is close to y if they share many "low" digits.
Every p-adic integer is a limit of natural numbers
Since the p-adic integers are the
completion of the natural numbers, this is obvious. Indeed, we see that the
sequence of (
base-
p) natural numbers d
0, d
1d
0, d
2d
1d
0, ...
converges to d (every element shares another
digit with d).
In particular, we see that if x is a natural number, we may write down -x as a limit of natural numbers; there is no concept of a "positive" p-adic integer!
Ultrametrics are preserved in the completion; alternatively, this can be shown directly: the number of low digits shared by x and z is at least the smaller of the number of low digits shared by x and y and the number of low digits shared by y and z, for any y.
(This is not true of Qp)
Students of analysis should already have recognised the topology given by the p-adic norm: it's the (infinite) Tychonoff product {0,...,p-1}×{0,...,p-1)×(0,...,p-1)×...
By Tychonoff's theorem, it is compact.
Or we can do it directly. Say we have a sequence of p-adics
a1=...a1,2a1,1a1,0
a2=...a2,2a2,1a2,0
a3=...a3,2a3,1a3,0
...
Then infinitely many of the a's share the same 1's
digit; pick a
n1 to be the first such. Of these, infinitely many share the same
p's digit; pick a
n2 to be the first of these after a
n1. Continuing in this fashion, we build a
sequence {a
nk}
k such that a
nj share the same left k digits for all j>k.
This sequence clearly converges, so we see Zp is a compact metric space.
As a perfect set that is dense in itself, it must be.
But we can see it directly. Any p-adic integer is, after all, an infinite string of the digits {0,...,p-1} that just happens to go left when we write it down. Basic set theory shows that the cardinality of the set of such strings (at least when they go to the right) is aleph.
(So it's what's often loosely called a Cantor set, and if p=2 then it has the same topology.
This follows directly from being the infinite product {1,...,p} × {1,...,p} × ...