6 Extensions of complete valued fields

Theorem 6.1. Assuming that:

Then
  • (i) || extends uniquely to an absolute value ||L on L defined by
    |y|L=|NLL(y)|1nyL.
  • (ii) L is complete with respect to ||L.

Recall: If LK is finite, NLK:LK is defined by NLK(y)=detK(mult(y) where mult(y):LL is the K-linear map induced by multiplication by y.

Facts:

Definition 6.2 (Norm). Let (K,||) be a non-archimedean valued field, V a vector space over K. A normon V is a function :V0 satisfying:

  • (i) x=0x=0.
  • (ii) λx=λx for all λK, xV.
  • (iii) x+ymax(x,y) for all x,yV.

Example. If V is finite dimensional and e1,,en is a basis of V. The supremum sup on V is defined by

xsup=maxi|xi|,

where x=i=1nxiei.

Exercise: sup is a norm.

Definition 6.3 (Equivalent norms). Two norms 1 and 2 on V are equivalent if there exists C,D>0 such that

Cx1x2Dx1xV.

Fact: A norm defines a topology on V, and equivalent norms induce the same topology.

Proposition 6.4. Assuming that:

Then V is complete with respect to sup.

Proof. Let (vi)i=1 be a Cauchy sequence in V, and let e1,,en be a basis for V.

Write vi=j=1nxjiej. Then (xji)i=1 is a Cauchy sequence in K. Let xjixjK, then viv:=j=1nxjej.

Theorem 6.5. Assuming that:

Then any two norms on K are equivalent. In particular, V is complete with respect to any norm (using Proposition 6.4).

Proof. Since equivalence defines an equivalence relation on the set of norms, it suffices to show that any norm is equivalent to sup.

Let e1,,en be a basis for V, and set D:=maxiei>0. Then for x=i=1nxiei, we have

xmaxixiei=maxi|xi|eiDmaxi|xi|=Dxsup.

To find C such that Csup, we induct on n=dimV.

For n=1: x=x1e1=|x1|e1, so take C=e1.

For n>1: set Vi=spane1,,ei1,ei+1,,en. By induction, Vi is complete with respect to , hence closed.

Then ei+Vi is closed for all i, and hence

S:=i=1nei+Vi

is a closed subset not containing 0. Thus there exists c>0 such that B(0,C)S= where B(0,C)={xV|x<C}.

Let 0x=i=1nxiei and suppose |xj|=maxi|xi|. Then xsup=|xj|, and 1xjS. Thus xixjC, and hence

xC|xj|=Cxsup.

V is complete since it is complete with respect to sup (see Proposition 6.4).

Definition 6.6 (Integral closure). Let R be a subring of S. We say sS is integral over R if there exists a monic polynomial f(X)R[X] such that f(s)=0.

The integral closure Rint(S) of R inside S is defined to be

Rint(S)={sS|s integral over R}.

We say R is integrally closed in S if Rint(S)=R.

Proposition 6.7. Rint(S) is a subring of S. Moreover, Rint(S) is integrally closed in S.

Proof. Example Sheet 2.

Lemma 6.8. Assuming that:

Then OK is integrally closed in K.

Proof. Let xK be integral over OK. Without loss of generality, x0. Let f(X)=Xn+an1Xn1++a0OK[X] such that f(x)=0. Then

x=an11xa01xn1.

If |x|>1, we have |an11xa01xn1|<1. Thus |x|1xOK.

Lemma 6.9. OL is the integral closure of OK inside L.

Proof. Let 0yL and let

f(X)=Xd+ad1Xd1++a0K[X]

be the minimal (monic) polynomial of y.

Claim: y integral over OK if and only if f(X)K[X].

  • Clear.
  • Let g(X)OK[X] monic such that g(y)=0. Then f|g (in K[X]), and hence every root of f is a root of g. So every root of f in K¯ is integral over OK, so ai are integral over OK for i=0,,d1.

Hence aiOk (by Lemma 6.8). By Corollary 4.5, |ai|max(|a0|,1) for i=0,,d1. By property of NLK, we have NLK(y)=±a0m for m1.

Hence

yOL|NLK(y)|1|a0|1Corollary 4.5|ai|1i,i.e. aiOK

Thus msubint(L)=OL and proves the Lemma.

Proof of Theorem 6.1. We first show ||L=|NLK()|1n satisfies the three axioms in the definition of absolute value.

  • (i) |y|L=0|NLK(y)|1n=0NLK(y)=0y=0
  • (ii) |y1y2|L=|NLK(y1,y2)|1n=|NLK(y1)NLK(y2)|1n=|NLK(y1)|1n|NLK(y2)|1n=|y1|L|y2|L
  • (iii) Set OL={yL||y|L1}.

    Claim: OL is the integral closure of OK inside L.

    Assuming this, we prove (iii). Let x,yL, and without loss of generality assume |x|L|y|L. Then |xy|L hence xyOL. Since 1OL and OLis a ring, we have 1+xyOL and hence |1+xy|L1. Hence |x+y|L|y|L=max(|x|L,|y|L) thus (iii) is satisfied.

So we have proved that ||L is an absolute value on L.

Since NLK(x)=xn for xK, |x|L extends || on K.

If ||L is another absolute value on L extending ||, then ||L,||L are norms on L.

Theorem 6.5 tells us that ||L,||L induce the same topology on L. Hence ||L=||Lc for some c>0 (by Proposition 1.4) since ||L extends ||, we have c=1.

Now we show that L is complete with respect to ||L: this is immediate by Theorem 6.5.

Let (K,||) be a complete discretely valued field.

Corollary 6.10. Let LK be a finite extension. Then

  • (i) L is discretely valued with respect to ||L.
  • (ii) OL is the integral closure of OK in L.

Proof.

  • (i) v a valuation on K, vL valuation on L such that vL extends v. Let n=[L:K], and let yL×. Then |y|L=|NLK(y)|1n hence vL(y)=1nv(NLK(y)), hence vL(L×)1nv(K×), so vL is discrete.
  • (ii) Lemma 6.9.

Corollary 6.11. Let K¯K be an algebraic closure of K. Then || extends to a unique absolute value ||K¯ on K¯.

Proof. Let xK¯, then xL for some LK finite. Define |x|K¯=|x|L. Well-defined, i.e. independent of L by the uniqueness in Theorem 6.1.

The axioms for ||K¯ to be an absolute value can be checked over finite extensions.

Uniqueness: clear.

Remark. ||K¯ on K¯ is never discrete. For example K=p, pnp¯ for all n>0. Then

vp(pn)=1nv(p)=1n.

p¯ is not complete with respect to ||p.

Example Sheet 2: p:= completion of p¯ with respect to ||p¯, then p is algebraically closed.

Proposition 6.12. Assuming that:

  • LK finite extension of complete discretely valued fields.

  • (i): OK is compact.

  • (ii): The extension of residue fields kLk is finite and separable.

Then there exists αOL such that OL=OK[α].

Later we’ll prove that the (i) implies (ii).

Proof. We’ll choose αOL such that:

  • there exists βOL[α] a uniformiser for OL

  • OK[α]kL surjective

kLk separable tells us that there exists α¯kL such that kL=k(α¯).

Let αOL a lift of α¯, and g(X)OK[X] a monic lift of the minimal polynomial of α¯.

Fix πLOL a uniformiser. Then g¯(X)k[X] irreducible and separable, hence g(α)0modπL and g(α)0modπL.

If g(α)0modπL2, then

g(α+πL)g(α)+πLg(α)modπL2.

Thus

vL(g(α+πL))=vL(πLg(α))=vL(πL)=1.

(vL normalised valuation on L).

Thus either vL(g(α))=1 or vL(g(α+πL))=1. Upon possibly replacing α by α+πL, we may assume vL(g(α))=1.

Set β=g(α)OK[α] a uniformiser. Then OK[α]L is the image of a continuous map:

OKnL(x0,,xn1)i=0nxiai

where n=[K(α):K]. Since OK is compact, OK[α]L is compact, hence closed. Since kL=k(α¯), OK[α] contians a set of coset representatives for kL=OLβOL.

Let yOL. Then Proposition 3.4 gives us

y=i=0λiβi,λiOK[α]

Then ym=i=0mλiβiOK[α]. Hence yOK[α], since Ok[α] is closed.