p-adic Numbers — Topology, Algebra, and Hensel's Lemma

⚡ Key Takeaways

For each prime $p$ , the field of $p$ -adic numbers $\mathbb{Q}_p$ is the completion of $\mathbb{Q}$ under the $p$ -adic absolute value $|x|_p = p^{-v_p(x)}$ . It is a complete, totally disconnected, locally compact non-archimedean field whose ring of integers $\mathbb{Z}_p$ is a compact DVR. Hensel's Lemma lets one lift approximate roots of polynomials over $\mathbb{Z}_p$ to exact roots, making $\mathbb{Z}_p$ behave like a "local power series ring." Used pervasively in number theory (local-global principles), arithmetic geometry, and representation theory. Originated with Kurt Hensel (1897).

💡 The Core Insight

Replace "closeness on the number line" with "closeness in divisibility." Two integers are $p$ -adically close when their difference is divisible by a high power of $p$ . Completing $\mathbb{Q}$ under this metric turns infinite base- $p$ expansions to the left into convergent series, and the non-archimedean inequality $|x+y|_p \le \max(|x|_p,|y|_p)$ makes analysis astonishingly rigid: series converge iff their terms tend to zero, and Newton's method either fails immediately or converges in one symbolic step (Hensel).

Definition

p

-adic valuation and absolute value

Fix a prime $p$ . For a nonzero integer $n$ , the $p$ -adic valuation $v_p(n)$ is the exponent of $p$ in its prime factorization. Extend to $\mathbb{Q}^\times$ by $v_p(a/b) = v_p(a) - v_p(b)$ , and set $v_p(0) = +\infty$ . The $p$ -adic absolute value is

1

\quad |x|_p = p^{-v_p(x)}, \qquad |0|_p = 0.

Derivation hint: Higher divisibility by $p$ ( $v_p$ large) yields smaller $|x|_p$ ; the exponential base $p^{-1}$ is chosen so that the product formula holds. The function $v_p$ satisfies, for all $x,y \in \mathbb{Q}$ ,

2

\quad v_p(xy) = v_p(x) + v_p(y), \qquad v_p(x+y) \ge \min\bigl(v_p(x), v_p(y)\bigr),

with equality in the second when $v_p(x)\ne v_p(y)$ . Derivation hint: Multiplicativity follows from unique factorization; the ultrametric estimate from factoring out the minimal power of $p$ before adding.

DefinitionThe field

\mathbb{Q}_p

and ring

\mathbb{Z}_p

$\mathbb{Q}_p$ is the completion of $(\mathbb{Q}, |\cdot|_p)$ : equivalence classes of Cauchy sequences modulo null sequences, with the metric and field operations extended by continuity. The ring of $p$ -adic integers is

3

\quad \mathbb{Z}_p = \{x \in \mathbb{Q}_p : |x|_p \le 1\} = \{x : v_p(x) \ge 0\}.

Derivation hint: $\mathbb{Z}_p$ is the closed unit ball; closure under addition uses the ultrametric, closure under multiplication uses additivity of $v_p$ . Equivalently, $\mathbb{Z}_p$ is the inverse limit

4

\quad \mathbb{Z}_p = \varprojlim_{n} \mathbb{Z}/p^n\mathbb{Z}, \qquad \mathbb{Z}_p \ni x \;\longleftrightarrow\; (x_n)_n,\ x_{n+1}\equiv x_n \pmod{p^n}.

Derivation hint: A Cauchy sequence's residues mod $p^n$ stabilize, defining compatible elements of $\mathbb{Z}/p^n$ ; conversely a compatible tower assembles into a Cauchy sequence. Every $x \in \mathbb{Q}_p^\times$ has a unique $p$ -adic expansion

5

\quad x = \sum_{n=v_p(x)}^{\infty} a_n p^{n}, \qquad a_n \in \{0,1,\dots,p-1\},\ a_{v_p(x)}\neq 0.

Derivation hint: Greedily extract digits $a_n$ by reducing mod $p$ after dividing by $p^n$ ; convergence holds because $|a_n p^n|_p \le p^{-n} \to 0$ .

TheoremStructure of

\mathbb{Z}_p

and

\mathbb{Q}_p

1. $\mathbb{Z}_p$ is a complete discrete valuation ring (DVR) with unique maximal ideal $\mathfrak{m} = p\mathbb{Z}_p$ and residue field $\mathbb{Z}_p/p\mathbb{Z}_p \cong \mathbb{F}_p$ . 2. The units are $\mathbb{Z}_p^\times = \{x : |x|_p = 1\}$ , and every nonzero $x \in \mathbb{Q}_p$ factors uniquely as $x = p^{v_p(x)} u$ with $u \in \mathbb{Z}_p^\times$ . 3. $\mathbb{Q}_p = \mathbb{Z}_p[1/p] = \operatorname{Frac}(\mathbb{Z}_p)$ . 4. Topologically: $\mathbb{Z}_p$ is compact, $\mathbb{Q}_p$ is locally compact, both are totally disconnected, perfect, and homeomorphic (as topological spaces) to a Cantor set.

ProofCompactness of

\mathbb{Z}_p

via the inverse limit

1. By [4], $\mathbb{Z}_p = \varprojlim \mathbb{Z}/p^n\mathbb{Z}$ sits as a closed subset of the product $\prod_n \mathbb{Z}/p^n\mathbb{Z}$ , each factor finite (hence compact, discrete). 2. By Tychonoff's theorem the product is compact. 3. The inverse limit is the subset cut out by the compatibility conditions $x_{n+1} \equiv x_n \pmod{p^n}$ , which are closed (intersections of preimages of points under continuous maps to discrete spaces). 4. A closed subspace of a compact space is compact; hence $\mathbb{Z}_p$ is compact. 5. $\mathbb{Q}_p = \bigcup_{k\ge 0} p^{-k}\mathbb{Z}_p$ is a countable increasing union of compact open sets, so it is locally compact (each point lies in the compact open neighborhood $p^{-k}\mathbb{Z}_p$ ). 6. Total disconnectedness: the sets $a + p^n\mathbb{Z}_p$ are simultaneously open and closed (clopen) and form a basis; any two distinct points are separated by such a clopen partition, so connected components are singletons. $\blacksquare$

∎

TheoremHensel's Lemma — strong form

Let $f \in \mathbb{Z}_p[X]$ and suppose $a_0 \in \mathbb{Z}_p$ satisfies

6

\quad |f(a_0)|_p < |f'(a_0)|_p^{2}.

Derivation hint: This says the value $f(a_0)$ is much smaller than the squared derivative — the precise condition under which Newton iteration is contracting in the ultrametric. Then there exists a unique $a \in \mathbb{Z}_p$ with

7

\quad f(a) = 0 \qquad\text{and}\qquad |a - a_0|_p \le \frac{|f(a_0)|_p}{|f'(a_0)|_p} < |f'(a_0)|_p .

Derivation hint: $a$ is the limit of the Newton sequence; the bound is the first Newton step's size, and uniqueness follows from quadratic contraction.

Simple form. If $\bar f \in \mathbb{F}_p[X]$ (reduction mod $p$ ) has a simple root $\bar a_0$ (so $\bar f(\bar a_0)=0$ , $\bar f'(\bar a_0)\neq 0$ ), then $\bar a_0$ lifts to a unique root of $f$ in $\mathbb{Z}_p$ reducing to $\bar a_0$ .

ProofHensel via Newton iteration

Define the Newton sequence

8

\quad a_{k+1} = a_k - \frac{f(a_k)}{f'(a_k)}, \qquad k \ge 0.

Derivation hint: Classical Newton's method; in $\mathbb{Q}_p$ we track $|\cdot|_p$ instead of monotone real convergence. 1. Set-up. Let $c = |f(a_0)|_p$ and $d = |f'(a_0)|_p$ . Hypothesis [6] is $c < d^2$ ; set $\theta = c/d^2 < 1$ . 2. Taylor expansion. For any $a,h \in \mathbb{Z}_p$ , since $f \in \mathbb{Z}_p[X]$ there exist $g_j \in \mathbb{Z}_p$ (binomial-type coefficients) with

f(a+h) = f(a) + f'(a)h + h^2\!\sum_{j\ge 0} g_j h^j.

All coefficients lie in $\mathbb{Z}_p$ , so $|f(a+h) - f(a) - f'(a)h|_p \le |h|_p^2$ . 3. One step. Take $h = -f(a_0)/f'(a_0)$ , so $|h|_p = c/d$ . Then

|f(a_1)|_p = |f(a_0) + f'(a_0)h + (\text{error})|_p = |\text{error}|_p \le |h|_p^2 = c^2/d^2.

Also the derivative is stable: $|f'(a_1) - f'(a_0)|_p \le |a_1-a_0|_p \le c/d < d$ , so by the ultrametric $|f'(a_1)|_p = d$ (since $c/d < d$ ). 4. Induction. With $\theta_k = |f(a_k)|_p / d^2$ one shows $\theta_{k+1} \le \theta_k^2$ , hence $\theta_k \le \theta^{2^k} \to 0$ . Thus $|f(a_k)|_p \to 0$ and $|a_{k+1}-a_k|_p = |f(a_k)|_p/d \to 0$ . 5. Convergence. Since the increments tend to $0$ and $\mathbb{Q}_p$ is complete (and the ultrametric makes $\sum$ converge iff terms $\to 0$ ), $a_k \to a \in \mathbb{Z}_p$ with $f(a) = 0$ by continuity, and $|a-a_0|_p \le c/d$ . 6. Uniqueness. If $f(a) = f(b) = 0$ with both in the ball $|x-a_0|_p < d$ , then $0 = f(a)-f(b) = (a-b)\bigl(f'(a_0) + \delta\bigr)$ with $|\delta|_p < d = |f'(a_0)|_p$ , forcing $|f'(a_0)+\delta|_p = d \ne 0$ , so $a = b$ . $\blacksquare$

∎

mathematician · Kurt Hensel | 1861–1941

A student of Kronecker and Weierstrass, Hensel introduced the $p$ -adic numbers in 1897, seeking an arithmetic analogue of the Laurent/power-series expansions that complex analysts used so fruitfully near a point. His insight: just as a meromorphic function is studied via its local expansion at each point of a Riemann surface, an integer should be studied via its "local expansion at each prime $p$ ." His 1908 book Theorie der algebraischen Zahlen developed the theory and the lemma now bearing his name. The viewpoint was initially regarded as exotic, but Hasse (his student) elevated it to a central tool of number theory through the local-global principle.

🧠 Intuition

Think of a prime $p$ as a "scale" and the $p$ -adic expansion as an odometer running leftward. In the reals, " $0.999\ldots$ " packs digits to the right and they shrink; $p$ -adically, digits pack to the left and the high powers $p^n$ are the small quantities. So $1 + p + p^2 + \cdots = \frac{1}{1-p}$ converges, and famously $\sum_{n\ge 0} (p-1)p^n = -1$ in $\mathbb{Q}_p$ — the "all-9s" number is $-1$ . The ultrametric inequality $|x+y|_p \le \max(|x|_p,|y|_p)$ produces a geometry where intuition from $\mathbb{R}$ fails productively: every triangle is isosceles, every point inside a ball is its center, and two balls are either nested or disjoint (never partially overlapping). This rigidity is why analysis is easy: a series converges precisely when its terms shrink, and Newton's method, once it gets one foothold (the Hensel condition), snaps to a root with quadratically accelerating $p$ -adic precision — each step doubles the number of correct base- $p$ digits.

Figure

The nested-ball ("tree") structure of

\mathbb{Z}_p

for

p=3

: cosets

a+p^n\mathbb{Z}_p

form a Cantor-like tree; descending one level fixes one more leftward digit.

```svg

```

ExampleA square root of

-1

in

\mathbb{Q}_5

via Hensel

Given. $f(X) = X^2 + 1$ over $\mathbb{Z}_5$ . Does $\sqrt{-1}$ exist in $\mathbb{Q}_5$ ?

Step 1 — Reduce mod $5$ . $\bar f(X) = X^2 + 1 \in \mathbb{F}_5[X]$ . Seek simple roots: $2^2 = 4 \equiv -1$ , so $\bar a_0 = 2$ is a root, as is $3$ . Check it is simple: $\bar f'(X) = 2X$ , $\bar f'(2) = 4 \neq 0$ in $\mathbb{F}_5$ . ✓

Step 2 — Hensel condition. Take $a_0 = 2$ . Then $f(2) = 5$ , so $|f(2)|_5 = 5^{-1}$ . And $f'(2) = 4$ , a unit, $|f'(2)|_5 = 1$ . Check [6]: $|f(2)|_5 = 5^{-1} < 1 = |f'(2)|_5^2$ . ✓

Step 3 — Newton iterate.

a_1 = a_0 - \frac{f(a_0)}{f'(a_0)} = 2 - \frac{5}{4}.

In $\mathbb{Z}_5$ , $4^{-1} \equiv 4 \pmod 5$ (since $4\cdot4=16\equiv1$ ), so $5/4 = 5\cdot 4^{-1}$ has $|5/4|_5 = 5^{-1}$ . Numerically mod $5^3$ : $4^{-1}\equiv 94 \pmod{125}$ , so $5/4 \equiv 470 \equiv 95 \pmod{125}$ , giving $a_1 \equiv 2 - 95 \equiv -93 \equiv 32 \pmod{125}$ . Check: $32^2 + 1 = 1025 = 8\cdot125+25$ , so $f(a_1)\equiv 25 \pmod{125}$ , i.e. $|f(a_1)|_5 \le 5^{-2}$ . Precision doubled.

Step 4 — Expansion. Iterating yields $a = 2 + 1\cdot5 + 2\cdot 5^2 + \cdots$ .

\boxed{\;\sqrt{-1} = 2 + 1\cdot 5 + 2\cdot 5^2 + \cdots \in \mathbb{Q}_5,\quad |a^2+1|_5 = 0\;}

(No square root of $-1$ exists in $\mathbb{Q}_2,\mathbb{Q}_3$ , or $\mathbb{R}$ ; this is genuinely a $5$ -adic phenomenon, available iff $p\equiv 1 \pmod 4$ .)

ExampleSolvability of

X^2 = 2

in

\mathbb{Q}_7

and failure in

\mathbb{Q}_5

Given. Is $2$ a square in $\mathbb{Q}_7$ ? In $\mathbb{Q}_5$ ?

$\mathbb{Q}_7$ case. $f(X)=X^2-2$ . Mod $7$ : $3^2 = 9 \equiv 2$ , so $\bar a_0 = 3$ is a root, $\bar f'(3) = 6 \neq 0$ — simple. Hensel: $f(3) = 7$ , $|f(3)|_7 = 7^{-1} < 1 = |6|_7^2$ . ✓ Hence $\sqrt2 \in \mathbb{Z}_7$ :

\sqrt 2 \equiv 3 \pmod 7, \quad \text{lift: } a_1 = 3 - \tfrac{7}{6},\ 6^{-1}\equiv 8 \pmod{49},\ a_1 \equiv 3 - 56 \equiv -53\equiv 45\pmod{49}.

Check $45^2 = 2025 = 41\cdot49+16$ , $f(a_1)\equiv 14 \pmod{49}$ ? Recompute: $2025 - 2 = 2023 = 41\cdot49 + 14$ , $|f(a_1)|_7\le 7^{-1}$ ... refine with full $6^{-1}\bmod 49 = 41$ : $a_1 \equiv 3 - 7\cdot41 = 3-287\equiv -284\equiv 10 \pmod{49}$ ; $10^2-2=98=2\cdot49$ , so $|f(a_1)|_7\le 7^{-2}$ . ✓ Precision doubled.

\boxed{\;\sqrt 2 \in \mathbb{Q}_7,\quad \sqrt2 = 3 + 1\cdot 7 + \cdots\;}

$\mathbb{Q}_5$ case. Mod $5$ the squares are $\{0,1,4\}$ . Since $2 \notin \{0,1,4\}$ , $\bar f(X) = X^2 - 2$ has no root in $\mathbb{F}_5$ .

\boxed{\;X^2 = 2 \text{ has no solution in } \mathbb{Q}_5\;}

Lesson. For a unit $u$ , $u$ is a square in $\mathbb{Z}_p$ ( $p$ odd) iff its reduction is a square in $\mathbb{F}_p^\times$ — Hensel reduces the question to the residue field.

⚠️ Common mistake

" $p$ -adic numbers are just formal power series in $p$ ." The digits obey carrying rules (base- $p$ addition with carries to the left); $\mathbb{Z}_p$ is not the polynomial-style $\mathbb{F}_p[[t]]$ . Indeed $\mathbb{Z}_p$ has characteristic $0$ while $\mathbb{F}_p[[t]]$ has characteristic $p$ , though both are complete DVRs with residue field $\mathbb{F}_p$ .
" $|f(a_0)|_p < 1$ is enough for Hensel." No — you need the squared-derivative condition [6]. When $f'(a_0)$ is itself divisible by $p$ (a non-simple root mod $p$ ), naive lifting fails; one must use the stronger inequality or Hensel does not apply.
" $\mathbb{Q}_p$ is ordered like $\mathbb{R}$ ." $\mathbb{Q}_p$ admits no field ordering compatible with its arithmetic; " $|x|_p$ small" has nothing to do with sign or magnitude.
" $\mathbb{Z}_p$ is countable." It is uncountable (continuum cardinality), like $\mathbb{R}$ — there are uncountably many leftward digit strings.

CounterexampleHensel fails without the squared-derivative bound

Consider $f(X) = X^2 - p \in \mathbb{Z}_p[X]$ with $p$ odd, and try $a_0 = 0$ . Then $f(0) = -p$ , $|f(0)|_p = p^{-1} < 1$ , which looks small. But $f'(0) = 0$ , so $|f'(0)|_p = 0$ and [6] fails utterly. Indeed $X^2 = p$ has no root in $\mathbb{Q}_p$ : if $x^2 = p$ then $2v_p(x) = v_p(p) = 1$ , impossible for an integer valuation. The valuation obstruction is exactly what the squared-derivative condition guards against. Likewise $f(X)=X^2-2$ over $\mathbb{Z}_2$ : $\bar f = X^2 = (X)^2$ has a double root at $0$ , $\bar f'=0$ , Hensel is silent — and one checks $2$ is not a square in $\mathbb{Q}_2$ (it requires the refined dyadic test $u\equiv 1 \bmod 8$ ).

connection

$\mathbb{Q}_p$ sits among the completions of $\mathbb{Q}$ : by Ostrowski's Theorem every nontrivial absolute value on $\mathbb{Q}$ is equivalent to the usual archimedean one (giving $\mathbb{R}$ ) or to some $|\cdot|_p$ (giving $\mathbb{Q}_p$ ). The product formula $\prod_v |x|_v = 1$ over all places $v$ unifies them and underlies Adeles and Ideles, the locally compact ring $\mathbb{A}_\mathbb{Q} = \mathbb{R}\times \prod'_p \mathbb{Q}_p$ on which modern automorphic theory lives. Hensel's Lemma is the arithmetic shadow of the Newton–Kantorovich convergence criterion and is functorially the statement that $\mathbb{Z}_p$ is a henselian local ring; the abstraction leads to Étale Cohomology (strictly henselian local rings as stalks) and to Smoothness and Formal Lifting in deformation theory. Algebraically, $\mathbb{Z}_p$ is the prototypical complete DVR, linking to Discrete Valuation Rings, Krasner's Lemma (stability of roots), and the classification of Local Fields (finite extensions of $\mathbb{Q}_p$ ). Galois-theoretically, the inertia and Frobenius structure of $\mathrm{Gal}(\overline{\mathbb{Q}_p}/\mathbb{Q}_p)$ feeds Local Class Field Theory and the Local Langlands Correspondence.

🔬 Where this appears

⏳ History

- 1897 — Kurt Hensel introduces

p

-adic numbers in a note to the DMV, by analogy with power-series expansions of analytic functions.

- 1908 — Hensel's Theorie der algebraischen Zahlen systematizes the theory and presents the lifting lemma.

- 1913 — Hensel develops the formal calculus; the field gains his student Helmut Hasse's attention.

- 1920s — Hasse formulates and proves the local-global principle for quadratic forms (Hasse–Minkowski), demonstrating the power of completions.

- 1932 — Krull's general valuation theory and the abstract notion of complete valued fields place

\mathbb{Q}_p

in a broad algebraic frame.

- 1936–1945 — Chevalley and others build local class field theory purely in terms of

\mathbb{Q}_p

and its extensions; the idele/adele formalism emerges (Chevalley 1936, Weil, Tate's 1950 thesis).

- 1962 — Tate's thesis (published) recasts

L

-functions adelically, harnessing all

\mathbb{Q}_p

simultaneously.

- 1970s–80s — Lubin–Tate theory,

p

-adic

L

-functions (Kubota–Leopoldt, Iwasawa), and Coates–Wiles bring

p

-adic analysis to the heart of arithmetic geometry.

- 1995 — Wiles's proof of Fermat's Last Theorem uses deformation theory of Galois representations over

p

-adic rings, where Hensel-type lifting (formal smoothness) is structural.

open-problem

$\mathbb{Q}_p$ is not algebraically closed; its algebraic closure $\overline{\mathbb{Q}_p}$ is not complete, and completing it yields $\mathbb{C}_p$ , which is algebraically closed but not locally compact — the loss of local compactness obstructs naive harmonic analysis and motivated rigid/Berkovich analytic geometry. Deep open and active areas: the precise structure of the absolute Galois group $\mathrm{Gal}(\overline{\mathbb{Q}_p}/\mathbb{Q}_p)$ beyond tame quotients; the $p$ -adic Langlands correspondence for $\mathrm{GL}_n$ , $n\ge 3$ (only $\mathrm{GL}_2(\mathbb{Q}_p)$ is fully understood, via Colmez); effective bounds in Hensel-type lifting for systems of polynomials (multivariate / Newton-polygon refinements and Tougeron's theorem boundaries); and the integrality and special-value behaviour of $p$ -adic $L$ -functions (Iwasawa Main Conjectures, largely proven in cases but with frontier generalizations). The anabelian program (Mochizuki) probes how much arithmetic is encoded in the étale fundamental group of $p$ -adic spaces — among the most contested frontiers.

p-adic Numbers — Topology, Algebra, and Hensel's Lemma

Formal Statement

Building Intuition

Worked Examples

Common Misconceptions

Connections

Real-World Applications

Historical Development

Subtleties & Limits