p-adic Numbers: When 100000000 Is Close to Zero

A different meaning of "close"

Ordinary distance is about magnitude: |7 − 3| = 4, and two numbers are close if that gap is small. The p-adic numbers, introduced by Kurt Hensel in 1897, throw that idea out and replace it with divisibility. Fix a prime p. Two integers are p-adically close if their difference is divisible by a high power of p — the more times p divides the gap, the closer the numbers are considered, no matter how large that gap is in the ordinary sense.

Concretely: with p = 2, the numbers 0 and 100000000 (which is 2⁸ × 390625) are extremely close 2-adically, because their difference is divisible by 2⁸. Meanwhile 0 and 1 are as far apart as two numbers can be, since 1 isn't divisible by 2 at all. This flips ordinary intuition on its head — the "biggest" integers in the usual sense can be the ones sitting right next to zero.

The valuation and the metric formula

Formally, the p-adic valuation v_p(n) of a nonzero integer n is the exponent of p in its prime factorization — how many times p divides n. The p-adic absolute value and the p-adic distance are then defined directly from it:

v_p(n)   = the exponent of p in the prime factorization of n
|n|_p    = p^(-v_p(n))          (and |0|_p = 0 by convention)
|x-y|_p  = p^(-v_p(x-y))        <-- the p-adic DISTANCE between x and y

example, p = 2:
  v_2(8)  = 3   ->  |8|_2  = 2^-3 = 0.125
  v_2(1)  = 0   ->  |1|_2  = 2^0  = 1
  8 is "smaller" (closer to 0) 2-adically than 1, despite 8 > 1 normally

This is a genuine metric — it satisfies positivity, symmetry, and a triangle inequality — so the p-adic numbers form a legitimate metric space, just a deeply unfamiliar one. Completing the rationals under this new metric, the same way completing them under the ordinary absolute value produces the real numbers, produces an entirely different number system: ℚ_p, the field of p-adic numbers.

Ultrametric: the strong triangle inequality

The p-adic metric actually obeys something stronger than the usual triangle inequality — the ultrametric or "strong triangle" inequality, d(x,z) ≤ max(d(x,y), d(y,z)) instead of merely d(x,y) + d(y,z). This has strange but precise consequences: every triangle is isosceles, with the two longest sides always equal; every point inside an open ball is a valid center of that ball; and two balls are either disjoint or one contains the other entirely — they can never partially overlap. These are exactly the properties of a hierarchical branching structure, which is why p-adic numbers are naturally drawn as an infinite tree rather than a line: at each level you branch into p children according to the next digit of a base-p expansion, and distance in the metric corresponds directly to depth of the shared ancestor.

Why mathematicians actually use this

p-adic analysis isn't a curiosity kept in a museum case; it's load-bearing machinery in modern number theory. Hensel's lemma lets you lift a solution of a polynomial equation modulo p to an exact solution in ℚ_p, essentially a p-adic version of Newton's method. The Hasse local-global principle asks whether a Diophantine equation solvable in every ℚ_p (and in the reals) is solvable in the rationals — true for quadratic forms, famously false in general. And Andrew Wiles's 1994 proof of Fermat's Last Theorem is built on p-adic Galois representations of elliptic curves, connecting this 19th-century curiosity directly to one of the most famous theorems in mathematics.

Frequently asked questions

Try it live

Everything above runs in your browser — open p-adic Number Ultrametric Space and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.