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

  • p-adic numbers ℚₚ complete ℚ using non-Archimedean metric: closeness measured by divisibility by powers of prime p
  • Two numbers are p-adically close if their difference is divisible by high power of p, inverting intuition about size
  • Hensel's Lemma lifts polynomial solutions mod pⁿ to solutions mod pⁿ⁺¹, analogous to Newton-Raphson method
  • p-adic topology violates Archimedean property but enables simpler computations than reals for number theory problems
  • Natural framework for Diophantine equations, local-global principles, and algebraic number theory applications
  • Kurt Hensel (1897) developed this by treating primes as "points at infinity" to do analysis at individual primes

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

Overview

The p-adic numbers Qp\mathbb{Q}_p provide an alternative completion of the rational numbers Q\mathbb{Q} that is fundamentally different from the familiar real numbers R\mathbb{R}. While the reals arise from the standard absolute value measuring "size," the p-adic numbers emerge from a different notion of distance: two numbers are "close" if their difference is divisible by a high power of a fixed prime pp. This seemingly exotic construction reveals profound structure in number theory, providing a natural setting for studying Diophantine equations, local-global principles, and algebraic number theory. The p-adic numbers were introduced by Kurt Hensel in 1897 as an algebraic tool for studying problems in number theory. Hensel's insight was to treat primes like points at infinity and develop analysis "at a prime," analogous to how power series work in complex analysis. This perspective transforms questions about integers and congruences into problems about continuous functions and convergent series. The p-adic topology is non-Archimedean—violating our geometric intuition—yet possesses rich algebraic structure that makes many computations simpler than their real counterparts. The culmination of this algebraic structure is Hensel's Lemma, a fundamental lifting principle that allows solutions to polynomial equations modulo pnp^n to be lifted to solutions modulo pn+1p^{n+1} under suitable conditions. This makes Qp\mathbb{Q}_p remarkably amenable to computation and provides a powerful tool for existence proofs. The lemma is the p-adic analogue of the Newton-Raphson method and serves as the foundation for understanding the interplay between topology and algebra in p-adic analysis.

💡 The Core Insight

The p-adic numbers complete Q\mathbb{Q} with respect to a non-Archimedean absolute value where divisibility by high powers of pp measures closeness. This topology makes Zp\mathbb{Z}_p compact and allows polynomial equations to be solved by successive approximation via Hensel's Lemma.

Definition

p-adic Numbers — Topology, Algebra, and Hensel's Lemma Fix a prime pp. The p-adic absolute value on Q\mathbb{Q} is defined by:

∣x∣p=p−vp(x) for x≠0,∣0∣p=0|x|_p = p^{-v_p(x)} \text{ for } x \neq 0, \quad |0|_p = 0

where vp(x)v_p(x) is the p-adic valuation: if x=pnabx = p^n \frac{a}{b} with gcd⁡(a,p)=gcd⁡(b,p)=1\gcd(a,p) = \gcd(b,p) = 1, then vp(x)=nv_p(x) = n. The p-adic metric is dp(x,y)=∣x−y∣pd_p(x,y) = |x-y|_p. The completion of Q\mathbb{Q} with respect to this metric is the field Qp\mathbb{Q}_p of p-adic numbers. The ring Zp={x∈Qp:∣x∣p≤1}\mathbb{Z}_p = \{x \in \mathbb{Q}_p : |x|_p \leq 1\} is the ring of p-adic integers.

Hensel's Lemma (Classical Form): Let f(X)∈Zp[X]f(X) \in \mathbb{Z}_p[X] be a polynomial. If there exists a∈Zpa \in \mathbb{Z}_p such that ∣f(a)∣p<∣f′(a)∣p2|f(a)|_p < |f'(a)|_p^2, then there exists a unique α∈Zp\alpha \in \mathbb{Z}_p with f(α)=0f(\alpha) = 0 and ∣α−a∣p<∣f′(a)∣p|\alpha - a|_p < |f'(a)|_p.

Core Theory

TheoremMain Results

Theorem [1] (Ostrowski): Every non-trivial absolute value on Q\mathbb{Q} is equivalent either to the standard absolute value ∣⋅∣∞|\cdot|_\infty or to a p-adic absolute value ∣⋅∣p|\cdot|_p for some prime pp.

Theorem [2] (Ultrametric Inequality): For all x,y∈Qpx, y \in \mathbb{Q}_p, |x + y|_p \leq \max\{|x|_p, |y|_p\} with equality if ∣x∣p≠∣y∣p|x|_p \neq |y|_p.

Theorem [3] (Compactness of Zp\mathbb{Z}_p): The ring Zp\mathbb{Z}_p is a compact topological ring. Every element of Zp\mathbb{Z}_p has a unique representation x = \sumi=0^\infty a_i p^i where 0≤ai<p0 \leq a_i < p, and this series converges in the p-adic topology.

Theorem [4] (Hensel's Lemma — Strong Form): Let f(X)∈Zp[X]f(X) \in \mathbb{Z}_p[X]. If a∈Zpa \in \mathbb{Z}_p satisfies vp(f(a))>2vp(f′(a))v_p(f(a)) > 2v_p(f'(a)), then there exists a unique α∈Zp\alpha \in \mathbb{Z}_p with f(α)=0f(\alpha) = 0, vp(α−a)>vp(f′(a))v_p(\alpha - a) > v_p(f'(a)), and f′(α)=f′(a)f'(\alpha) = f'(a).

Proof

Proof of Theorem [4] (Hensel's Lemma): We construct α\alpha as a limit of a sequence using Newton iteration. Let a0=aa_0 = a and define recursively:

an+1=an−f(an)f′(an)a_{n+1} = a_n - \frac{f(a_n)}{f'(a_n)}

Step 1 (Sequence is well-defined): We have f′(a0)=f′(a)≠0f'(a_0) = f'(a) \neq 0 in Zp\mathbb{Z}_p since vp(f′(a))v_p(f'(a)) is finite. By induction, we'll show f′(an)=f′(a)f'(a_n) = f'(a) for all nn, so division is always possible.

Step 2 (Improvement at each step): We claim that

vp(f(an+1))>2vp(f(an))−2vp(f′(a))v_p(f(a_{n+1})) > 2v_p(f(a_n)) - 2v_p(f'(a))

By Taylor expansion in Zp\mathbb{Z}_p:

f(an+1)=f(an−f(an)f′(an))=f(an)−f′(an)⋅f(an)f′(an)+f′′(ξ)2(f(an)f′(an))2f(a_{n+1}) = f(a_n - \frac{f(a_n)}{f'(a_n)}) = f(a_n) - f'(a_n) \cdot \frac{f(a_n)}{f'(a_n)} + \frac{f''(\xi)}{2}\left(\frac{f(a_n)}{f'(a_n)}\right)^2

for some ξ\xi between ana_n and an+1a_{n+1}. This gives:

f(an+1)=f′′(ξ)2f′(an)2⋅f(an)2f(a_{n+1}) = \frac{f''(\xi)}{2f'(a_n)^2} \cdot f(a_n)^2

Taking valuations:

vp(f(an+1))≥2vp(f(an))−2vp(f′(an))v_p(f(a_{n+1})) \geq 2v_p(f(a_n)) - 2v_p(f'(a_n))

Step 3 (Derivative is constant): For n=0n = 0, we have vp(a1−a0)=vp(f(a0)/f′(a0))=vp(f(a0))−vp(f′(a0))>vp(f′(a0))v_p(a_1 - a_0) = v_p(f(a_0)/f'(a_0)) = v_p(f(a_0)) - v_p(f'(a_0)) > v_p(f'(a_0)) by hypothesis. By Taylor:

f′(a1)=f′(a0)+f′′(η)(a1−a0)f'(a_1) = f'(a_0) + f''(\eta)(a_1 - a_0)

Thus vp(f′(a1)−f′(a0))≥vp(a1−a0)>vp(f′(a0))v_p(f'(a_1) - f'(a_0)) \geq v_p(a_1 - a_0) > v_p(f'(a_0)), which by the ultrametric inequality implies vp(f′(a1))=vp(f′(a0))v_p(f'(a_1)) = v_p(f'(a_0)). By induction, f′(an)=f′(a)f'(a_n) = f'(a) for all nn.

Step 4 (Cauchy sequence): We have

vp(an+1−an)=vp(f(an))−vp(f′(an))>vp(f(an−1))−vp(f′(a))>⋯v_p(a_{n+1} - a_n) = v_p(f(a_n)) - v_p(f'(a_n)) > v_p(f(a_{n-1})) - v_p(f'(a)) > \cdots

The sequence of valuations vp(an+1−an)v_p(a_{n+1} - a_n) is strictly increasing, so ∣an+1−an∣p→0|a_{n+1} - a_n|_p \to 0. Thus (an)(a_n) is Cauchy in Zp\mathbb{Z}_p, which is complete, so an→α∈Zpa_n \to \alpha \in \mathbb{Z}_p.

Step 5 (Limit is a root): Since ff is continuous in the p-adic topology and vp(f(an))→∞v_p(f(a_n)) \to \infty, we have f(α)=lim⁡f(an)=0f(\alpha) = \lim f(a_n) = 0.

Step 6 (Uniqueness): Suppose β∈Zp\beta \in \mathbb{Z}_p also satisfies the conditions. Then:

0=f(α)−f(β)=f′(α)(α−β)+O((α−β)2)0 = f(\alpha) - f(\beta) = f'(\alpha)(\alpha - \beta) + O((\alpha-\beta)^2)

If α≠β\alpha \neq \beta, then vp(f′(α)(α−β))<vp((α−β)2)v_p(f'(\alpha)(\alpha - \beta)) < v_p((\alpha-\beta)^2) since vp(α−a),vp(β−a)>vp(f′(a))v_p(\alpha - a), v_p(\beta - a) > v_p(f'(a)) implies both have derivative f′(a)f'(a). This contradicts the equation above. □

∎
⚠️ Common mistake

Common Misconception: The p-adic absolute value is not Archimedean. In particular, all integers have norm at most 1: ∣n∣p≤1|n|_p \leq 1 for n∈Zn \in \mathbb{Z}. This means "large" integers (in the usual sense) can be "small" p-adically if they're divisible by high powers of pp. The triangle inequality becomes stronger (ultrametric), which implies all triangles are isosceles: if ∣x∣p≠∣y∣p|x|_p \neq |y|_p, then ∣x+y∣p=max⁡{∣x∣p,∣y∣p}|x+y|_p = \max\{|x|_p, |y|_p\}. This violates geometric intuition from Euclidean geometry.

mathematician

Kurt Hensel (1897) — Introduced p-adic numbers as a tool for applying analytic methods to number theory, developing the theory in his book Theorie der algebraischen Zahlen and subsequent works on p-adic analysis.

visualization

3-adic balls illustrating ultrametric property B(0,1) = ℤ₃ B(0, 1/3) = 3ℤ₃ B(1, 1/3) 9ℤ₃ 0 1 9 Any two distinct balls are either disjoint or nested

Examples and Applications

Example1

Computing in Q7\mathbb{Q}_7: Find the 7-adic expansion of 13\frac{1}{3}. We seek x=∑i=0∞ai7ix = \sum_{i=0}^\infty a_i 7^i with 0≤ai<70 \leq a_i < 7 such that 3x=13x = 1. Start with 3a0≡1(mod7)3a_0 \equiv 1 \pmod{7}. Since 3⋅5=15=2⋅7+13 \cdot 5 = 15 = 2 \cdot 7 + 1, we have a0=5a_0 = 5. Now 3⋅5=15=1+2⋅73 \cdot 5 = 15 = 1 + 2 \cdot 7, so we carry 2. For the next digit: 3a1+2≡0(mod7)3a_1 + 2 \equiv 0 \pmod{7}, giving 3a1≡−2≡5(mod7)3a_1 \equiv -2 \equiv 5 \pmod{7}, so a1=4a_1 = 4 (since 3⋅4=12=5+73 \cdot 4 = 12 = 5 + 7). Continuing: 3⋅4+2=14=2⋅73 \cdot 4 + 2 = 14 = 2 \cdot 7, carry 2. 3a2+2≡0(mod7)3a_2 + 2 \equiv 0 \pmod{7} gives a2=4a_2 = 4 again. This pattern repeats: 13=5+4⋅7+4⋅72+4⋅73+⋯\frac{1}{3} = 5 + 4 \cdot 7 + 4 \cdot 7^2 + 4 \cdot 7^3 + \cdots in Z7\mathbb{Z}_7. Verification: 3(5+4⋅7+4⋅72+⋯ )=15+12⋅7+12⋅72+⋯=(1+2⋅7)+(1+2⋅7)⋅7+⋯=1+2⋅7+2⋅72+⋯3(5 + 4 \cdot 7 + 4 \cdot 7^2 + \cdots) = 15 + 12 \cdot 7 + 12 \cdot 7^2 + \cdots = (1 + 2 \cdot 7) + (1 + 2 \cdot 7) \cdot 7 + \cdots = 1 + 2 \cdot 7 + 2 \cdot 7^2 + \cdots. Actually, more carefully: the geometric series gives 13=5+4⋅7⋅11−7=5−286\frac{1}{3} = 5 + 4 \cdot 7 \cdot \frac{1}{1-7} = 5 - \frac{28}{6}... Let me recalculate. Actually: 13=13⋅11−0=∑k=0∞5⋅7k\frac{1}{3} = \frac{1}{3} \cdot \frac{1}{1-0} = \sum_{k=0}^\infty 5 \cdot 7^k where we solve 3x≡1(mod7k)3x \equiv 1 \pmod{7^k} iteratively. The answer is 13=5+2⋅7+2⋅72+⋯=5(1+2⋅75+⋯ )\frac{1}{3} = 5 + 2 \cdot 7 + 2 \cdot 7^2 + \cdots = 5(1 + \frac{2 \cdot 7}{5} + \cdots)... Direct: 13=524‾\frac{1}{3} = \overline{524} in base 7 (periodic).

Example2

Applying Hensel's Lemma: Solve x2≡2(mod7n)x^2 \equiv 2 \pmod{7^n} for all nn. First, check if 22 is a quadratic residue mod 77: we need x2≡2(mod7)x^2 \equiv 2 \pmod{7}. Testing: 12=1,22=4,32=2(mod7)1^2=1, 2^2=4, 3^2=2 \pmod 7. So x0=3x_0 = 3 works. Apply Hensel: f(x)=x2−2f(x) = x^2 - 2, f′(x)=2xf'(x) = 2x. At x0=3x_0 = 3: f(3)=7f(3) = 7, f′(3)=6f'(3) = 6. Check condition: ∣f(3)∣7=∣7∣7=7−1|f(3)|_7 = |7|_7 = 7^{-1} and ∣f′(3)∣72=∣6∣72=1|f'(3)|_7^2 = |6|_7^2 = 1. Since 7−1<17^{-1} < 1, Hensel applies. Newton iteration: x1=x0−f(x0)f′(x0)=3−76x_1 = x_0 - \frac{f(x_0)}{f'(x_0)} = 3 - \frac{7}{6}. In Z7\mathbb{Z}_7, we need 6−16^{-1}. Since 6⋅6=36=1+5⋅76 \cdot 6 = 36 = 1 + 5 \cdot 7, we have 6−1≡6(mod7)6^{-1} \equiv 6 \pmod{7}. So x1=3−7⋅6=3−42=−39≡3+7⋅4=31(mod49)x_1 = 3 - 7 \cdot 6 = 3 - 42 = -39 \equiv 3 + 7 \cdot 4 = 31 \pmod{49}. Check: 312=961=2+959=2+7⋅137=2+72⋅19+7⋅431^2 = 961 = 2 + 959 = 2 + 7 \cdot 137 = 2 + 7^2 \cdot 19 + 7 \cdot 4. So 312≡2+7⋅4(mod49)31^2 \equiv 2 + 7 \cdot 4 \pmod{49}... Actually, 31=3+4⋅731 = 3 + 4 \cdot 7, and (3+4⋅7)2=9+24⋅7+16⋅49=2+7+24⋅7(mod49)=2+25⋅7≡2+4⋅7(mod49)(3 + 4 \cdot 7)^2 = 9 + 24 \cdot 7 + 16 \cdot 49 = 2 + 7 + 24 \cdot 7 \pmod{49} = 2 + 25 \cdot 7 \equiv 2 + 4 \cdot 7 \pmod{49}. Hmm, let me use the formula: x1=3−76(mod49)x_1 = 3 - \frac{7}{6} \pmod{49}. We have 76=7⋅6−1=7⋅6=42≡−7(mod49)\frac{7}{6} = 7 \cdot 6^{-1} = 7 \cdot 6 = 42 \equiv -7 \pmod{49}, so x1=3+7=10(mod49)x_1 = 3 + 7 = 10 \pmod{49}. Check: 102=100=2+98=2+2⋅49≡2(mod49)10^2 = 100 = 2 + 98 = 2 + 2 \cdot 49 \equiv 2 \pmod{49}. ✓ Hensel guarantees this lifts to a unique square root of 22 in Z7\mathbb{Z}_7: 2=3+1⋅7+a2⋅72+⋯\sqrt{2} = 3 + 1 \cdot 7 + a_2 \cdot 7^2 + \cdots

Counterexample

Failure without derivative condition: Consider f(x)=x2f(x) = x^2 over Q2\mathbb{Q}_2 with initial approximation a=0a = 0. We have f(0)=0f(0) = 0 and f′(0)=0f'(0) = 0. The condition v2(f(0))>2v2(f′(0))v_2(f(0)) > 2v_2(f'(0)) becomes ∞>∞\infty > \infty, which fails. Indeed, 00 is a "simple" root modulo any power of 22, but there's no unique lift: both x=0x = 0 and x=2kx = 2^k (for any k≥1k \geq 1) satisfy x2≡0(mod22k)x^2 \equiv 0 \pmod{2^{2k}}. Without the derivative condition, we lose uniqueness and even existence. For instance, f(x)=x2+xf(x) = x^2 + x has f(0)=0f(0) = 0, f′(0)=1f'(0) = 1, so the derivative is good. But f(1)=0f(1) = 0 and f′(1)=1f'(1) = 1 as well mod 22. The equation x2+x≡0(mod2n)x^2 + x \equiv 0 \pmod{2^n} has multiple solutions for all nn, demonstrating that the valuation condition is essential for Hensel's conclusion.

🧠 Intuition

Mental Model: Think of Zp\mathbb{Z}_p as an infinite tree where each node has pp children, labeled 0,1,…,p−10, 1, \ldots, p-1. A p-adic integer is an infinite path from the root, specifying digits a0,a1,a2,…a_0, a_1, a_2, \ldots. The p-adic metric measures how long two paths stay together: dp(x,y)=p−nd_p(x,y) = p^{-n} where nn is the first level where the paths diverge. This topology makes "nearby" elements share many initial digits. Balls are clopen (both closed and open), and any point in a ball is its center—there's no distinguished origin for a ball! This is radically different from Euclidean space. Hensel's Lemma says: if you have an approximate root modulo pnp^n with a good derivative, you can systematically improve it, like focusing in on a specific branch of the tree. The non-vanishing derivative ensures you're not at a "branch point" where multiple paths could work. Geometrically, Qp\mathbb{Q}_p is totally disconnected (every point is a connected component) but locally compact. It's the p-adic analogue of the real line, but with a fractal, tree-like structure rather than a continuum.

Historical Development

Kurt Hensel introduced p-adic numbers in 1897, motivated by analogy with Laurent series and power series expansions in algebraic geometry. Just as complex functions can be studied via Taylor expansions around points, Hensel wanted to study algebraic integers via expansions in powers of primes. His key insight was that Modular Arithmetic modulo increasing powers of pp naturally led to a complete metric space. Hensel's lemma first appeared in his 1908 book Theorie der algebraischen Zahlen. The result was initially viewed as a technical tool for algebraic number theory, particularly for studying ramification in field extensions. Helmut Hasse later recognized p-adic numbers as fundamental objects in their own right, developing p-adic analysis systematically in the 1920s. The modern categorical perspective emerged with Grothendieck's development of scheme theory in the 1960s. The spectrum Spec Zp\text{Spec } \mathbb{Z}_p became a fundamental example of a local scheme, and Hensel's lemma was reinterpreted as a formal smoothness criterion. The étale cohomology framework revealed deep connections between p-adic analysis and arithmetic geometry. Contemporary formulations emphasize Hensel's lemma as an example of inverse limits and lifting properties. The Henselization of a local ring abstractly captures the essence of Hensel's lemma, while rigid analytic geometry (Tate, 1962) extends p-adic analysis to a full geometric theory, complete with analytic spaces and coherent sheaves.

Connections to Other Areas

connection

<span class="wiki-link-missing">Iwasawa Theory</span> studies Galois representations and L-functions using infinite field extensions of Qp\mathbb{Q}_p. The Iwasawa algebra Zp[[Gal(Q∞/Q)]]\mathbb{Z}_p[[\text{Gal}(\mathbb{Q}_\infty/\mathbb{Q})]] for the cyclotomic extension Q∞\mathbb{Q}_\infty is a complete local ring where Hensel-like techniques control the arithmetic of ideal class groups. The Main Conjecture of Iwasawa theory (proved by Mazur-Wiles for cyclotomic fields) relates p-adic L-functions to Selmer groups, fundamentally relying on p-adic completeness and analytic continuation in the p-adic domain. This connection bridges classical number theory, algebraic K-theory, and modular forms, showing how p-adic topology enables deep arithmetic results.

What You Need First

What This Opens Up

See Also

Valuation Theory · Local Fields · Adèles and Idèles · Newton's Method · Complete Metric Spaces · Discrete Valuation Rings · Formal Groups · Rigid Analytic Geometry