Let us now study a different area of algebraic number theory, introduced by Kurt Hensel in an attempt to apply power series expansions in connection with numbers. While trying to explain the properties of (rational) integers mathematicians started embedding in bigger and bigger structures, richer and richer in properties. came in a natural attempt to form quotients, and for some time people believed that that is all about reality. Pythagoras was seemingly the first to locate and prove the irrationality of a number, namely, . It took humankind centuries for completing the picture of the real line. One possibility is to look as the completion of . A sequence an, , of rational numbers is called a Cauchy sequence if for every real ε > 0, there exists such that |am – an| ≤ ε for all m, , m, n ≥ N. Every Cauchy sequence should converge to a limit and it is (and not ) where this happens. Seeing convergence of Cauchy sequences, people were not wholeheartedly happy, because the real polynomial X2 + 1 did not have—it continues not to have—roots in . So the next question that arose was that of algebraic closure. was invented and turned out to be a nice field which is both algebraically closed and complete.
Throughout the above business, we were led by the conventional notion of distance between points (that is, between numbers)—the so-called Archimedean distance or the absolute value. For every rational prime p, there exists a p-adic distance which leads to a ring strictly bigger than and containing . This is the ring of p-adic integers. The quotient field of is the field of p-adic numbers. is complete in the sense of convergence of Cauchy sequences (under the p-adic distance), but is not algebraically closed. We know anyway that a (unique) algebraic closure of exists. We have , that is, it was necessary and sufficient to add the imaginary quantity i to to get an algebraically closed field. Unfortunately in the case of the p-adic distance the closure is of infinite extension degree over . In addition, is not complete. An attempt to make complete gives an even bigger field Ωp and the story stops here, Ωp being both algebraically closed and complete. But Ωp is already a pretty huge field and very little is known about it.
In the rest of this section, we, without specific mention, denote by p an arbitrary rational prime.
There are various ways in which p-adic integers can be defined. A simple way is to use infinite sequences.
A p-adic integer is defined as an infinite sequence of elements with the property that an+1 ≡ an (mod pn) for every . Each an, being an element of , can be represented as a (rational) integer unique modulo pn. Thus, if bn, , define another sequence of integers with bn ≡ an (mod pn) for every n, the p-adic integers (an) and (bn) are treated the same. In particular, if 0 ≤ bn < pn for every n, then (bn) is called the canonical representation of (an). The set of all p-adic integers is denoted by .[20] A sequence (an) of integers with an+1 ≡ an (mod pn) for every n is called a p-coherent sequence.
|
See Exercise 2.144 for another way of defining p-adic integers. We now show that is a ring. Before doing that, we mention that the ring is canonically embedded in by the injective map , a ↦ (a).
It turns out that is an integral domain. In order to see why, let us focus our attention on the units of . Let us plan to denote (the multiplicative group of units of ) by Up. The next result characterizes elements of Up.
For , the following conditions are equivalent:
Proof [(a)⇒(b)] Let (an)(bn) = (anbn) = 1 = (1) for some . Then for every we have anbn ≡ 1 (mod pn), that is, an is invertible modulo pn and hence modulo p as well, that is, p an. [(b)⇒(c)] Obvious. [(c)⇒(a)] Let us construct a p-coherent sequence bn, , of (rational) integers with anbn ≡ 1 (mod pn). This (bn) would be the desired inverse of (an) in . Since p a1 and an ≡ a1 (mod p), it follows that p an as well and, therefore, the congruence anx ≡ 1 (mod pn) has a unique solution modulo pn, namely (mod pn). We also have an+1bn+1 ≡ 1 (mod pn), that is, anbn+1 ≡ 1 (mod pn), that is, . |
Every can be written uniquely as x = pry for some and for some . Proof If p a1, take r := 0 and y := x. So assume that p|a1. Choose such that [an]pn = [0]pn for 1 ≤ n ≤ r, whereas [ar+1]pr+1 ≠ [0]pr+1. Such an r exists, since x ≠ 0 by hypothesis. For , we have ar+n ≡ ar ≡ 0 (mod pr), that is, pr|ar+n, whereas ar+n ≡ ar+1 ≢ 0 (mod pr+1), that is, pr+1 ar+n, that is, vp(ar+n) = r. Define bn := ar+n/pr. Since ar+n+1 ≡ ar+n (mod pr+n), division by pr gives bn+1 ≡ bn (mod pn), that is, . Moreover, prbn = ar+n ≡ an (mod pn), that is, x = pry. Finally, since p b1, we have . This establishes the existence of a factorization x = pry. The uniqueness of this factorization is left to the reader as an easy exercise. |
is an integral domain. Proof Let x1 and x2 be non-zero elements of . By Proposition 2.53, we can write x1 = pr1 y1 and x2 = pr2 y2 with r1, and y1, . Then (an) := x1x2 = pr1+r2 y1y2. Now and hence no bn is divisible by p. Therefore, ar1+r2+1 = pr1+r2 br1+r2+1 ≢ 0 (mod pr1+r2+1), that is, (an) = x1x2 ≠ 0. |
The quotient field of is called the field of p-adic numbers. |
Every non-zero can be expressed uniquely as x = pry with and . Proof One can write x = a/b for some a, . Then a = psc and b = ptd for some s, , c, and so x = ps–t(c/d) with . The proof for the uniqueness is left to the reader. |
The canonical inclusion naturally extends to the canonical inclusion . We can identify with the rational a/b and say that is contained in . Being a field of characteristic 0, contains an isomorphic copy of . The map gives this isomorphism explicitly. Note that the ring is strictly bigger than and the field is strictly bigger than the field (Exercise 2.147).
Proposition 2.55 leads to the notion of p-adic distance between pairs of points in . Let us start with some formal definitions.
A metric on a set S is a map such that for every x, y, we have:
A set S together with a metric d is called a metric space (with metric d). |
A norm on a field K is a map such that for all x, we have:
It is an easy check that for a norm ‖ ‖ on K the function , d(x, y) := ‖x – y‖, defines a metric on K. A norm ‖ ‖ on a field K is called non-Archimedean (or a finite valuation), if ‖x + y‖ ≤ max(‖x‖, ‖y‖) for all x, (a condition stronger than the triangle inequality). A norm which is not non-Archimedean is called Archimedean (or an infinite valuation). |
|
The p-adic norm on is defined as:
|
The p-adic norm | |p is a non-Archimedean norm on . Proof Non-negative-ness, non-degeneracy and multiplicativity of | |p are immediate. For proving the triangle inequality, it is sufficient to prove the non-Archimedean condition. Take x, . If x = 0 or y = 0 or x + y = 0, we clearly have |x + y|p ≤ max(|x|p, |y|p). So assume that each of x, y and x + y is non-zero. Write x = pru and y = psv with r, and u, . Without loss of generality, we may assume that r ≥ s. Then, x + y = psz, where . Since x + y ≠ 0, we have z ≠ 0; so we can write z = ptw for some and . But then |x + y|p = p–(s+t) ≤ p–s = max(p–r, p–s) = max(|x|p, |y|p). |
Two metrics d1 and d2 on a metric space S are called equivalent if a sequence (xn) from S is Cauchy with respect to d1 if and only if it is Cauchy with respect to d2. Two norms on a field are called equivalent if they induce equivalent metrics. |
For every , the field is canonically embedded in and thus we have a notion of a p-adic distance on . We also have the usual Archimedean distance | |∞ on . We now state an interesting result without a proof, which asserts that any distance on must be essentially the same as either the usual Archimedean distance or one of the p-adic distances.
The notions of sequences and series and their convergences can be readily extended to under the norm | |p. Since the p-adic distance assumes only the discrete values pr, , it is often customary to restrict ourselves only to these values while talking about the convergence criteria of sequences and series, that is, instead of an infinitesimally small real ε > 0 one can talk about an arbitrarily large with p–M ≤ ε.
Let x1, x2, . . . be a sequence of elements of . We say that this sequence converges to a limit , if given there exists such that |xn – x|p ≤ p–M for all n ≥ N. We write this as x = lim xn or as xn → x. Consider the partial sums for each . If there exists with sn → s, we say that the sum converges to s and write . A sequence x1, x2, . . . of elements of is said to be a Cauchy sequence if for every , there exists an such that |xm – xn|p ≤ p–M for all m, n ≥ N. |
A field K is called complete under a norm ‖ ‖ if every sequence of elements of K, which is Cauchy under ‖ ‖, converges to an element in K. |
For example, is complete under | |∞. We shortly demonstrate that is complete under | |p.
Consider a field K not (necessarily) complete under a norm ‖ ‖. Let C denote the set of all Cauchy sequences from K. Define addition and multiplication in C as (an) + (bn) := (an + bn) and (an)(bn) := (anbn). Under these operations C becomes a commutative ring with identity having a maximal ideal . The field is called the completion of K with respect to the norm ‖ ‖. K is canonically embedded in L via the map . The norm ‖ ‖ on K extends to elements of L as limn→∞ ‖an‖. L is a complete field under this extended norm. In fact, it is the smallest field containing K and complete under ‖ ‖.
is the completion of with respect to the Archimedean norm | |∞. On the other hand, turns out to be the completion of with respect to the p-adic norm | |p. Before proving this let us first prove that itself is a complete field under the p-adic norm. Let us start with a lemma.
A sequence (an) of p-adic numbers is a Cauchy sequence if and only if the sequence (an+1 – an) converges to 0. Proof [if] Take any . Since an+1–an → 0 by hypothesis, there exists such that |an+1 –an|p ≤ p–M for all n ≥ N. But then for all m, n ≥ N with m = n+k, , we have . Thus (an) is a Cauchy sequence. [only if] Take any . Since (an) is a Cauchy sequence by hypothesis, there exists such that |am – an|p ≤ p–M for all m, n ≥ N. In particular, |an+1 – an|p ≤ p–M for all n ≥ N, that is, an+1 – an → 0. |
The field is complete with respect to | |p. Proof Let (an) be a Cauchy sequence in . By Lemma 2.18, an+1 – an → 0. Therefore, there exists such that |an+1 – an|p ≤ 1 for all n ≥ N. For n = N + k, , we have
It then follows that |an|p ≤ p–m for all , where satisfies p–m = max(1, |a1|p, . . . , |aN |p). If m ≥ 0, then each (Exercise 2.148). Otherwise consider the sequence (p–man) which is clearly Cauchy and in which each , since |p–man|p ≤ pmp–m = 1. Thus, without loss of generality, we may assume that the given sequence (an) itself is one of p-adic integers. Let an = an,0+an,1p+an,2p2+· · · be the p-adic expansion of an (Exercise 2.145). Since (an) is Cauchy, for every there exists such that |am – an|p ≤ p–(M+1) for all m, n ≥ NM: that is, an, i = am, i for 0 ≤ i ≤ M, m, n ≥ NM. Define xM := an, M for any n ≥ NM and . It then follows that an → x. |
is the completion of with respect to the norm | |p. Proof Let C denote the ring of Cauchy sequences from (under the p-adic norm), the maximal ideal of C consisting of sequences that converge to 0, and . We now show that . If has the p-adic expansion a = a–rp–r +· · ·+a–1p–1 +a0+a1p+a2p2 + · · · (Exercise 2.145), then αn := a–rp–r + · · · + a–1p–1 + a0 + a1p + · · · + anpn, , define a sequence of elements of . We have |αn – a|p ≤ p–(n+1), that is, αn → a. Moreover, the sequence (αn) of rational numbers is Cauchy with respect to | |p, since for every we have |αm – αn|p ≤ p–(M+1) for all m, n≥ M. Thus , , is a well-defined field homomorphism. Being a field homomorphism is injective. What remains is to show that the map is surjective. Take any . Since (βn) is a Cauchy sequence, by Theorem 2.61 it converges to a point . We construct the sequence (αn) corresponding to a as described in the last paragraph. Then αn → a as well and hence using the triangle inequality (or the non-Archimedean condition) we have αn – βn = (αn – a) – (βn – a) → 0, that is, , that is, . |
The p-adic series (with ) converges if and only if |an|p → 0. Proof The only if part is obvious. For the if part, take a sequence (an) of p-adic numbers with |an|p → 0. Define . Since an+1 = sn+1 – sn → 0 by hypothesis, Lemma 2.18 guarantees that (sn) is a Cauchy sequence, that is, (sn) converges in . |
This is quite unlike the Archimedean norm | |∞. For example, with respect to this norm , whereas the series diverges.
Let us conclude our short study of p-adic methods by proving an important theorem due to Hensel. This theorem talks about the solvability of polynomial equations f(X) = 0 for . Before proceeding further, let us introduce a notation. Recall that every has a unique p-adic expansion of the form a = a0 + a1p + a2p2 + · · · with 0 ≤ an < p (Exercises 2.144 and 2.145). If a0 = a1 = · · · = an–1 = 0, then a = anpn + an+1pn+1 + an +2pn+2 + · · · = pnb, where . Thus pn|a in . We denote this by saying that a ≡ 0 (mod pn). Notice that a ≡ 0 (mod pn) if and only if |a|p ≤ p–n. We write a ≡ b (mod pn) for a, , if a – b ≡ 0 (mod pn). Since pn can be viewed as the element of , this congruence notation conforms to that for a general PID. ( is a PID by Exercise 2.148.)
Since by our assumption any ring A comes with identity (that we denote by 1 = 1A), it makes sense to talk for every about an element n = nA in A, which is the n-fold sum of 1. More precisely:
Given any , one can define the formal derivative of f as . Properties of formal derivatives of polynomials are covered in Exercise 2.61.
Let . Suppose that there exist and satisfying:
Then there exists a unique such that f(α) = 0 and |α – α0|p ≤ p–(M+1) (that is, α ≡ α0 (mod pM+1)). Proof Let us inductively construct a sequence α0, α1, α2, · · · of p-adic integers with the properties that |f(αn)|p ≤ p–(2M+n+1) and |f′(αn)|p = p–M for every . The given α0 provides the starting point (induction basis). For the inductive step, assume that n ≥ 1 and that α0, α1, . . . , αn–1 have been constructed with the desired properties. we now explain how to construct αn from αn–1. Put
We want to find a suitable kn for which |f(αn)|p ≤ p–(2M+n+1). Taylor expansion gives f(αn) = f(αn–1) + knpM+nf′(αn–1) + cnp2(M+n) for some . Since by induction hypothesis p2M+n |f(αn–1) and pM |f′(αn – 1), we can write
Since pM+1 f′(αn–1), the element and, therefore, there is a unique solution for kn of the congruence
This value of kn yields f (αn) = p2M + n(bnp + cnpn) ≡ 0 (mod p2M+n+1) for some . The Taylor expansion of f′ gives f′(αn) = f′(αn–1) + dnpM+n (for some ) which implies that f′(αn) ≡ f′(αn–1) (mod pM), that is, |f′(αn)|p = p–M. Since |αn – αn–1|p ≤ p–(M+n), it follows that αn – αn–1 → 0, that is, (αn) is a Cauchy sequence (under | |p). By the completeness of , we then have an such that αn → α. Similarly f(αn) – f(αn–1) → 0, that is, the sequence (f(αn)) is Cauchy and hence converges to f(α). Also |f(αn)|p ≤ p–(2M+n+1), that is, f(αn) → 0, that is, f(α) = 0. Finally, each αn ≡ α0 (mod pM+1), so that α ≡ α0 (mod pM+1). This establishes the existence of a desired . For proving the uniqueness of α, let satisfy f(β) = 0 and |β – α0|p ≤ p–(M+1). By Taylor expansion, f(β) = f(α) + (β – α)f′(α) + (β – α)2c for some , that is, (β – α)(f′(α) + (β – α)c) = 0. Now β – α = (β – α0) – (α – α0) and so |β – α|p ≤ max(|β – α0|p, |α – α0|p) ≤ p–(M+1), whereas f′(αn) → f′(α), so that |f′(α)|p = p–M. Therefore, f′(α)+(β –α)c ≢ 0 (mod pM+1) and, in particular, f′(α) + (β – α)c ≠ 0. Thus we must have β – α = 0. |
Note that αn in the last proof satisfies the congruence
f(αn) ≡ 0 (mod p2M+n+1)
for each . We are given the solution α0 corresponding to n = 0. From this, we inductively construct the solutions α1, α2, . . . corresponding to n = 1, 2, . . . respectively. The process for computing αn from αn–1 as described in the proof of Hensel’s lemma is referred to as Hensel lifting. The given conditions ensure that this lifting is possible (and uniquely doable) for every , and in the limit n → ∞ we get a root of f. Since each kn is required modulo p, we can take . So α admits a p-adic expansion of the form α = α0 + k1pM+1 + k2pM+2 + k3pM+3 + · · ·.
The special case M = 0 for Hensel’s lemma is now singled out:
Let . Suppose that there exists an satisfying:
Then there exists a unique such that f(α) = 0 and |α – α0|p < 1 (that is, α ≡ α0 (mod p)). |
For this special case, we compute solutions αn of f(x) ≡ 0 (mod pn+1) inductively for n = 1, 2, 3, . . . , given a suitable solution α0 of this congruence for n = 0. The lifting formula is now:
Equation 2.21
is canonically embedded in and so is in . Thus it makes sense to carry out the lifting process for a polynomial and for some solution α0 of f(X) ≡ 0 (mod p) in . One solves Formula (2.21) in and obtains each . The limit α belongs to and is a solution of f(X) = 0 in . For example, let p be an odd prime and . Let be a solution of X2 ≡ a (mod p). Here f(X) = X2 – a, so that f′(X) = 2X, that is, f′(α0) = 2α0 ≢ 0 (mod p). Thus the conditions of Corollary 2.30 are satisfied and we get a unique square root of α in with α ≡ α0 (mod p). This α has a p-adic expansion of the form α = α0 + k1p + k2p2 + k3p3 + · · ·. As a specific numerical example, take p = 7, a = 2 and α0 = 3. Using Formula (2.21), we compute k1 = 1, α1 = 10, k2 = 2, α2 = 108, k3 = 6, α3 = 2166, and so on. Thus a square root of 2 in is 3 + 1 × 7 + 2 × 72 + 6 × 73 + · · ·. The other square root of α in can be obtained by starting with α0 = 4. |
2.144 |
|
2.145 | In view of Exercise 2.144, every admits a unique expansion of the form x = x0 + x1p + x2p2 + · · · , where each . This notion of p-adic expansion can be extended to the elements of .
|
2.146 | Let p be an odd prime and with . From elementary number theory we know that the congruence x2 ≡ a (mod pn) has two solutions for every . Let x1 be a solution of x2 ≡ a (mod p). We know that a solution xn of x2 ≡ a (mod pn) lifts uniquely to a solution xn+1 of x2 ≡ a (mod pn+1). Thus we can inductively compute a sequence x1, x2, x3, · · · of integers. Show that (xn) is a p-adic integer and that (xn)2 = (a). |
2.147 |
|
2.148 | Prove the following assertions:
|
2.149 | Compute the p-adic expansion of 1/3 in and of –2/5 in . |
2.150 | Show that is dense in under the p-adic norm | |p, that is, show that given any and real ε > 0, there exists with |x – a|p < ε. Show also that is dense in . |
2.151 | Prove the following assertions that establish that is the closure of in under | |p.
|
2.152 | Show that:
|
2.153 | Prove that for any non-zero . [H] |
2.154 | Prove that for any the sequence (apn) converges in . [H] |
2.155 | Let p, , p ≠ q. Show that the fields and are not isomorphic. |
2.156 | Let a be an integer congruent to 1 modulo 8. Show that there exists an such that α2 = a and . |
2.157 | Compute with α2 + α + 223 = 0 and α ≡ 4 (mod 243). |
2.158 | Let p be an odd prime and . Show that the polynomial X2 – a has exactly root in . |
2.159 | Show that the polynomial X2 – p is irreducible in . |
2.160 | Teichmüller representative Let . Show that there exists a unique such that αp = α and α ≡ a (mod p). |
2.161 | Show that the algebraic closure of is of infinite extension degree over . [H] |
3.144.41.229