Algebraic function fields Places Definition An algebraic function field F/K of one variable over K is an extension field F ⊇ K such that F is a finite algebraic extension of K(x) for some element x ∈ F which is transcendental over K. K = {z ∈ F : z is algebraic over K} K is called the field of constants of F/K. K is algebraically closed in F if K = K. We have 1. K ⊆ K ⊂ F 2. F/K is an algebraic function field over K (why?) 1 Definition F/K is called rational, if F = K(x) for some x ∈ F transcendental over K. Definition A valuation ring of the function field F/K is a ring O ⊆ F with the following properties. 1. K ⊂ O ⊂ F 2. For any z ∈ F , z ∈ O or z −1 ∈ O Example Let p(x) ∈ K[x] be an irreducible polynomial. Then f (x) : f (x), g(x) ∈ K[x] and p(x) 6 | g(x) Op(x) = g(x) is a valuation ring of K(x)/K. 2 Proposition Let O be a valuation ring of the function field F/K. Then (a) O is a local ring, i.e., O has a unique maximal ideal P = O \ O ∗ , where O ∗ is the group of units of O. (b) For 0 6= x ∈ F , x ∈ P if and only if x−1 6∈ O. (c) K ⊂ O and K ∩ P = {0}. Proof (a) It suffices to show that P = O \ O ∗ is an ideal of O. • Let x ∈ P and let z ∈ O. Then xz 6∈ O ∗ (why?), implying xz ∈ P . • Let x, y ∈ P . We may assume that x/y ∈ O. Then 1 + x/y ∈ O, implying x + y = y(1 + x/y) ∈ P . (b) ... 3 (c) Let z ∈ K and assume to the contrary that z 6∈ O. Then z −1 ∈ O. Since z −1 is algebraic over K, there are elements a1 , . . . , ar ∈ K such that ar (z −1 )r + · · · + a1 z −1 + 1 = 0, implying z −1 (ar (z −1 )r−1 + · · · + a1 ) = −1. Thus, z = −(ar (z −1 )r−1 + · · · + a1 ) ∈ K[z −1 ] ⊆ O which contradicts our assumption. 2 4 Theorem Let O be a valuation ring of the function field F/K and let P be its unique maximal ideal. Then (a) P is a principal ideal. (b) If P = tO, then any 0 6= z ∈ F has a unique representation of the form z = tn u for some n ∈ Z and u ∈ O ∗ . (c) O is a principal ideal domain, and if P = tO and {0} 6= I ⊆ O is an is an ideal, then I = tn O for some n ∈ N. Definition A ring possessing the above properties is called a discrete valuation ring. Lemma Let O be a valuation ring of the algebraic function field F/K, P be its maximal ideal, and let 0 6= x ∈ P . Let x1 , . . . , xn ∈ P be such that x1 = x and xi ∈ xi+1 P , i = 1, . . . , n − 1. Then n ≤ [F : K(x)] < ∞. 5 Proof The inequality [F : K(x)] < ∞ is immediate (why?) and for the proof of n ≤ [F : K(x)] it suffices to show that x1 , . . . , xn are linearly independent over K(x). Assume to the contrary that n P ϕi xi = 0, where ϕi ∈ K(x), not all equal i=1 zero. We may assume that all ϕi ∈ K[x] and that x does not divide all of them (why). Let j = max{i : x 6 | ϕi }. Then for all i > j there exists a polynomial n P ϕi xi = 0 that gi ∈ K[x] such that ϕi = xgi , and it follows from i=1 P xi ϕi . Therefore, −xj ϕj = X xi X x i6=j ϕi + xi g i , −ϕj = x x i<j j i>j j implying ϕj ∈ P (why?). However, since ϕj = ϕj (0)+xgj with gj ∈ K[x] ⊆ O, ϕj (0) = ϕj −xgj ∈ 2 P ∩ K (why?), which is impossible, because ϕj (0) 6= 0 (why?). 6 Corollary If t ∈ P , then ∞ T tn O = {0}. n=1 Proof Assume to the contrary that for some z 6= 0, z ∈ n−1 ∞ T n=1 n−2 tn O. Then for all n ∈ N, the sequence x1 = z, x2 = t , x3 = t , . . . , xn = t satisfies the lemma prerequisite xi ∈ xi+1 P , i = 1 . . . , n − 1. However, this is impossible for n > [F : K(x)]. 2 7 Proof of the theorem (a) Assume to the contrary that P is not principal and let 0 6= x1 ∈ P . Since P 6= x1 O, there is x2 ∈ P \ x1 O, implying x2 x−1 1 6∈ O. Therefore, x1 x−1 2 ∈ P and x1 ∈ x2 P . We proceed by induction and obtain an infinite sequence x1 , x2 , . . . ∈ P such that xi ∈ xi+1 P for all i = 1, 2, . . .. However, this contradicts the lemma. (b) The uniqueness of the representation is immediate (why?), and for the proof of the existence we may assume that z ∈ O (why?). If z is a unit, then z = t0 z, i.e., u = z. Otherwise, z ∈ P (why?) and, by the corollary to the lemma, there is a maximal non-negative integer n such that z ∈ tn O. Therefore, z = tn u, where u is a unit (why?). 8 (c) Let {0} 6= I ⊆ O be an ideal an let A = {r ∈ N : tr ∈ I}. Then A 6= ∅ (why?) and let n = min(A). Since tn O ⊆ I, the proof of part (c) of the theorem will be complete if we show the converse inclusion. Let 0 6= x ∈ I. Since x = ts u for some u ∈ O ∗ and s ≥ n, ts ∈ I as well. Therefore, x = tn ts−n u ∈ tn O. 2 Definition (a) A place P of the function field F/K is the maximal ideal of some valuation ring O of F/K. Any element t ∈ P such that P = tO is called a prime element for P (or local parameter or uniformizing variable). (b) PF = {P : P is a place of F/K}. 9 Remark If O is a valuation ring of F/K and P is its maximal ideal, then O is uniquely determined by P : O = {z ∈ F : z −1 6∈ P }. We call Op = O the valuation ring of the place P . Definition (The arithmetic of Z ∪ {∞}) For all n ∈ Z, • ∞+∞=n+∞=∞+n • ∞>n 10 Definition A discrete valuation of F/K is a function v : F → Z ∪ {∞} with the following properties. (1) v(x) = ∞ if an only if x = 0. (2) v(xy) = v(x) + v(y) for all x, y ∈ F . (3) v(x + y) ≥ min{v(x), v(y)} for all x, y ∈ F . (4) There exists an element z ∈ F with v(z) = 1. (5) v(a) = 0 for all 0 6= a ∈ K. It follows from (2) and (4) that v : F → Z ∪ {∞} is surjective (how?). Also it follows from the definition that v(x) = v(−x) for all x ∈ F (how?). 11 Lemma (Strict Triangle Inequality) Let v be a discrete valuation of F/K and let x, y ∈ F be such that v(x) 6= v(y). Then v(x + y) = min{v(x), v(y)}. Proof Since v(x) 6= v(y), we may assume that v(x) < v(y). Now, assume to the contrary that v(x + y) 6= min{v(x), v(y)}. Then v(x + y) > v(x) (why?), implying a contradiction v(x) = v((x + y) − y) ≥ min{v(x + y), v(y)} > v(x). 2 Definition To a place P ∈ PF we associate a function vP : F → Z∪{∞} such that vP (0) = ∞ and for 0 6= z ∈ F vP (z) is defined as follows. Let t be a prime element for P . Then for some u ∈ O ∗ and some n ∈ Z, z = tn u and this representation of z is unique (for a given t). We put vP (z) = n. This function vP is well-defined (why?). 12 Theorem Let F/K be a function field. (a) For any place P ∈ PF , the function vP is a discrete valuation of F/K. In addition, OP OP∗ = {z ∈ F : vp (z) ≥ 0}, = {z ∈ F : vp (z) = 0}, P = {z ∈ F : vp (z) > 0}. An element z of F is a prime element for P if and only if vP (z) = 1. (b) Conversely, if v is a discrete valuation of F/K, then • Pv = {z ∈ F : v(z) > 0} is a place of F/K, and • OPv = {z ∈ F : v(z) ≥ 0} is the corresponding valuation ring. (c) Any valuation ring O of F/K is a maximal proper subring of F . 13 Proof (a) We shall prove the triangle inequality only. Let x, y ∈ F , vP (x) = n, and vP (y) = m, where n ≤ m < ∞. Then x = tn u1 and y = tm u2 , where u1 , u2 ∈ OP∗ and x + y = tn (u1 + tm−n u2 ) = tn z, where z ∈ OP . If z = 0, then vP (x + y) = ∞ > min{n, m}. Otherwise, z = tk u, where k ≥ 0 and u ∈ OP∗ , implying vP (x + y) = vP (tn+k u) = n + k ≥ n = min{vP (x), vP (y)}. The rest of (a) ... (b) ... 14 (c) Let O be a valuation ring of F/K and let z ∈ F \ O. We shall prove that O[z] = F . Let y ∈ F and let P be the maximal ideal of O. Then, for sufficiently large k, vP (yz −k ) ≥ 0 (why?). Consequently, w = yz −k ∈ O, implying y = wz k ∈ O[z]. 2 Since P is a maximal ideal of OP , OP /P is a field, and for z ∈ OP we define z(P ) to be the residue class of z modulo P . If z ∈ F \ OP , we put z(P ) = ∞.1 Sometimes we shall also write z + P instead of z(P ) for z ∈ OP . Since K ⊂ OP (K ⊂ OP ) (why?) and K∩P = {0} (K∩P = {0}) (why?), the residue class map OP → OP /P induces a canonical embedding of K (K) into OP /P . We shall always consider K (K) as a subfield of OP /P via this embedding. 1 This ∞ is not the same as in discrete valuations. 15 Definition Let P ∈ PF . (a) FP = OP /P is called the residue class field of P and the map x 7→ x(P ) from F to FP ∪ {∞} is called the residue class map with respect to P . (b) deg P = [FP : K] is called the degree of P . What happens when deg P = 1? Proposition Let P be a place of F/K and let 0 6= x ∈ P . Then deg P ≤ [F : K(x)] < ∞. Proof We have already seen that [F : K(x)] < ∞ (where?), and for the proof of deg P ≤ [F : K(x)] it suffices to show that any elements z1 , . . . , zn of OP , whose residue classes z1 (P ), . . . , zn (P ) are linearly independent over K, are linearly independent over K(x). 16 So, let z1 (P ), . . . , zn (P ) be linearly independent over K, and assume to n P ϕi z i = 0 the contrary that there is a non-trivial linear combination i=1 with ϕi ∈ K(x). We may assume that the ϕi are polynomials in x and not all of them are divisible by x (why?). That is, ϕ = ai + xgi , where ai ∈ K and gi ∈ K[x], and not all ai = 0. Applying the residue class map to n P ϕi zi = 0 we obtain i=1 0 = 0(P ) = n X ϕi (P )zi (P ) = i=1 n X ai zi (P ), i=1 which contradicts the linear independence of z1 (P ), . . . , zn (P ) over K. 2 17 Definition Let z ∈ F and let P ∈ Pf . We say that P is a zero (respectively, a pole) of z if and only if vP (z) > 0 (respectively, vP (z) < 0). If vP (z) = m > 0 (respectively, vP (m) = −m < 0), then P is a zero (respectively, pole) of P of order m. Theorem Let F/K be a function field, K ⊂ R be a subring of F , and let I 6= {0} be a proper ideal of R. Then there is a place P ∈ PF such that I ⊆ P and R ⊆ OP . Proof Let F = {S : S is a subring of F with R ⊆ S and IS 6= S}.2 F 6= ∅, because R ∈ F, and we shall prove that F is inductively ordered by inclusion (what does it mean?). 2 IS is the ideal of S generated by I. 18 S Indeed, if H ⊆ F is a totally ordered subset of F, then T = {S : S ∈ H} is a subring of S with R ⊆ T . We have to show that IT 6= T . Pn For this, assume to the contrary that i=1 ai si = 1, with ai ∈ I and si ∈ T . Since H is totally Pn ordered, there is an S ∈ H such that s1 , . . . , sn ∈ S. Therefore, 1 = i=1 ai si ∈ S as well, in contradiction with the definition of F (which contradiction?). By Zorn’s lemma, F contains a maximal element O that, as we prove below, is a valuation ring of F/K. We have K ⊂ O ⊂ F (why?). Assume to the contrary that there exists an element z of F with z 6∈ O and z −1 6∈ O. Then IO[z] = O[z] and IO[z −1 ] = O[z −1 ] (why?). Consequently, there are a0 , . . . , an , b0 , . . . , bm ∈ IO such that 1 = a 0 + a1 z + · · · + a n z n 1 = b0 + b1 z −1 + · · · + an z −m 19 It follows that m, n ≥ 1 (how?) and we may assume that m and n are chosen minimally and m ≤ n (why?). Multiplying the first equality by 1 − b0 and the second by an z n we obtain 1 − b0 = (1 − b0 )a0 + (1 − b0 )a1 z + · · · + (1 − b0 )an z n 0 = (b0 − 1)an z n + b1 an z n−1 + · · · + bm an z n−m Adding these equalities yields 1 = c0 + c1 z + · · · + cn z n−1 with coefficients ci in IO (why?). This contradicts the minimality of n. That is, we have shown that for any z ∈ F , z ∈ O or z −1 ∈ O. 20 2 Corollary Let F/K be a function field and let z ∈ F be transcendental over K. Then z has at least one zero and one pole.3 Proof Let R = K[z]. Then I = zK[z] is a proper ideal of R. Thus, there is a place P ∈ PF with z ∈ P (hence P is a zero of z). To prove that z has a pole we start with ... 2 Corollary The field K of constants of F/K is a finite extension of K. Proof Let P ∈ PF . Since K embeds into FP via the residue class map OP → FP , it follows that [K : K] ≤ [FP : K] (< ∞). 2 3 In particular, PF 6= ∅. 21