• Aucun résultat trouvé

Relative Spectral Norm on Algebraic Numbers

N/A
N/A
Protected

Academic year: 2022

Partager "Relative Spectral Norm on Algebraic Numbers"

Copied!
12
0
0

Texte intégral

(1)

Relative Spectral Norm on Algebraic Numbers

ANGELPOPESCU(*) - SOBIASULTANA(**)

ABSTRACT- LetK be a subfield ofQ, a fixed algebraic closure ofQ, the field of rational numbers. LetGKˆGal(Q=K) be the absolute Galois group ofK. For anyx2Q, we consider theK-spectral norm:kxkKˆmaxfjs(x)j:s2GKg. Let ebe the conjugation automorphism ofQand letC(GK) be the Banach algebra of all continuous mappings defined on the compact groupGKwith values inC. Let Csym(GK) be the Banach algebra overKeof all symmetric functions ofC(GK). Here KeˆRorCis the completion ofK relative to the usual absolute valuej:j. A function f is said to be symmetric if f(es)ˆe(f(s)) for any s2GK (when e2GK). LetQeKbe the completion ofQwith respect tok:kK. In this paper we prove thatQeK Csym(GK) ife2GKandQeK C(GK) ife2=GK. These last iso- morphisms are K-isomorphisms of Banach algebras. We give some othere properties of the closed subalgebras ofC(GK) in connection with some subfields of algebraic numbers.

1. The notion of a spectral norm on the algebraic number field was in- troduced by A. Popescu, N. Popescu and A. Zaharescu in [13]. This is a normk:konQ, a fixed algebraic closure ofQ, the field of rational numbers, defined byx ! kxk ˆmaxfjs(x)j:s2Gal(Q=Q)g. The completion Qe of Qrelative to this norm is a ring with zero divisors but with some interesting arithmetical and topological properties (see [8], [10], [11] and [13] - [17]).

The main property ofQe is that the absolute Galois groupGˆGal(Q=Q) of Qacts continuously on it, which is not the case ofC(only the identityeand the conjugation automorphismeofGact continuously onC). So that, if we

(*) Indirizzo dell'A.: Technical University of Civil Engineering Bucharest, Department of Mathematics, B-ul Lacul Tei 124, 72302 Bucharest 38, Romania.

E-mail: [email protected]

(**) Indirizzo dell'A.: Abdus Salam School of Mathematical Sciences, 68-B, New Muslim Town, Lahore, Pakistan.

E-mail: [email protected]

1991 Mathematics Subject Classification. Primary 11R99, 12J10; Secondary 46J10, 11R32.

(2)

want to get information onGby its action on different topological spaces, a good candidate isQ. Such type of researches were also done for the locale absolute Galois group GpˆGal(Qp=Qp), where Qp is a fixed algebraic closure of thep-adic number fieldQp(see [4], [9], and [12]). Some ideas of these researches come directly or indirectly from [2], [3].

In a letter to the first author [19], Prof. A. Zaharescu had the idea to generalize the usual spectral normk:kto aK-spectral norm onQ:

x ! kxkKˆmaxfjs(x)j:s2GKˆGal(Q=K)g;

whereKis an arbitrary subfield ofQ.

In this paper we study the completionQeKofQwith respect to theK- spectral norm defined above. Since the techniques of [13] - [17] are sometimes not useful in this new frame, we introduce some new ones. Let C(GK) be theC-Banach algebra of all continuous mappings defined on the compact groupGKˆGal(Q=K) (relative to its Krull topology) with values inC(with its usual absolute value topology). An elementf 2 C(GK) is said to be symmetric (only in the case whene2GK) iff(es)ˆe(f(s)) for any s2GK. Ife2= GKwe cannot define such a notion. LetCsym(GK) be theK- subalgebra of C(GK) which consists of all symmetric mappings ofC(GK).

LetLeKbe the topological closure inQeK of a subfieldLofQ,LK, with respect tok:kK. In Theorem 1 and Remark 2 we prove that ifKR\Q, then QeK is isomorphic and isometric with Csym(GK) as K-algebras. Ife K6R\Q( i.e.e2= GK), thenQeK is isometric and isomorphic asC-alge- bras with C(GK) (Theorem 2). In Theorem 3 we prove that always QeK[ei] C(GK), whereeiis the constant functionf(s)ˆiˆ 

p 1

for anys inGK. SoW(QeK) has the codimension 1 or 2 inC(GK). HereW:Q ! C(GK), is the following map: W(a)ˆWa, where Wa(s)ˆs(a) and W is the unique extension by continuity of W to QeK. In Theorems 4, 5 and 6 we obtain analogous results as in Theorems 1, 2 and 3 for a subextension L=K of Q=K. Theorem 7 is the direct generalization of Theorem 2.2 of [13], namely LeKˆK[x] for anyg KLQand for anx2LeK (a generalization of the primitive element theorem from the classical field theory). In Lemma 2 and Corollary 2 we give a general Archimedean frame for the classical idea of Krasner (Krasner's Lemma) [1] or [7]. We apply this new ideas to prove that iff;gare locally constant and nonconstant mapping fromC(GK) such thatgis ``sufficiently close'' tof;thenC[f]C[g] (see Theorem 8). Then we apply this last result in order to get some information on the closedK-e subalgebras ofC(GK) (see Corollary 3, Proposition 2 and Proposition 3).

2. Let Qbe the rational number field and let C be a fixed complex number field,QC. LetQbe the algebraic closure ofQinCand letK,

(3)

QKQbe a fixed subfield ofQ. If [K:Q]51, usuallyKis called an algebraic number field. In this paper we call any subfield ofQan algebraic number field. LetGKˆGal(Q=K) be the absolute Galois group ofK. We denote byGˆGal(Q=Q) the absolute Galois group ofQ.Gis a compact totally disconnected group relative to the Krull topology on it [1]. The Krull topology on GKGis exactly the topology ofGinduced toGK. IfGK is finite, the Krull topology becomes the discrete topology.

Let us fix a tower of subfields:

K ˆK0 K1K2. . .Kn . . .Q;

such that all extensionsKn=Kare normal and finite and[1nˆ0KnˆQ. Then the Galois groupsHn ˆGal(Q=Kn);nˆ0;1;2;. . .are closed and open and form a fundamental system of neighborhoods of the identity automorphism e of GK. Let denote by j:j the usual absolute value on C. For any s2GK;x ! js(x)j ˆ jxjsis an Archimedean absolute value onQ. Letebe the conjugation automorphism ofQ. Ife2GKthenj:jesˆ j:js. So any two automorphismss;m2GKgive rise to one and the same topology onQ(j:js andj:jmare dependent) if and only ifsˆem. For anyx2Qone considers:

kxkKˆmaxfjs(x)j:s2GKg:

It is not difficult to prove the following result.

LEMMA 1. The map x ! kxkK is a norm onQand(Q;k:kK)is a K- normed algebra, i.e.:

i) kxkK ˆ0 if and only ifxˆ0;for anyx2Q, ii) kxykK kxkKkykK;for anyx;y2Q, iii) kx‡ykK kxkK‡ kykK;for any x;y2Q, iv) kaxkKˆ jajkxkK;fora2K;x2Q, and

v) k1kKˆ1:

We call this norm theK-spectral normonQ:It extends toQthe usual absolute value ofK. In general, in ii), one has no equality. For instance, if KˆQ and xˆ1‡ 

p2

;yˆ1 

p2

2Q, then kxykKˆ16ˆ(1‡ 

p2 )2 ˆ

ˆ k1‡ 

p2

kKk1 

p2

kK. Taking count of the Artin-Schreir theory [6] we see that theK-spectral normx ! ks(x)kKis an absolute value onQequal to the usual absolute value if and only ifKis a ``real closed'' subfield inQ, i.e. if K(i)ˆQ. In this last caseGK Z2 and this case is the only case whenGKis finite. LetQeKbe the completion ofQrelative to theK-spectral norm. The topological closureKeofKinQeKisRorCwheneverKR\Q or not. HenceKis dense inC(relative toj:j) if and only ifK6R\Q. In

(4)

generalQeKhas zero divisors. It is a field if and only ifKis a ``real closed'' subfield inQ. In all the other casesQeKhas zero divisors. But alwaysQeKis aK-Banach algebra.e

Let us now consider theC-Banach algebra

C(GK)ˆ ff :GK !Cjf continuousg;

with the usual sup-norm:kfk ˆsupfjf(s)j:s2GKg. Since GK is a com- pact group, this last quantity is always finite. For anya2Qone defines :

Wa:GK !C;Wa(s)ˆs(a):

The mapWahas only a finite number of values namely, the image ofWais the orbit of a (relative to GK), i.e. the set of all conjugates of a in Q. If a2K;Wa(s)ˆafor anys2GK. So we can embedKe intoC(GK) by asso- ciating toa2K, the constant mapWa inC(GK).

PROPOSITION1. The mapW:Q ! C(GK)defined byW(a)ˆWa;Wa(s)ˆ

ˆs(a)is an isometry and aK-algebra morphism.So it is aK-embedding of (Q;k:kK)into(C(GK);k:k).

PROOF. It is clear that Wa‡WbˆWa‡b; WaWbˆWab and lWaˆWla for any a;b2Q and l2K. Now kakKˆmaxfjs(a)j:s2GKg ˆ

ˆsupfjWa(s)j:s2GKg ˆ kWak and the proof is complete. p For some later applications we fixa2Qand as02GK. We are going to describe Wa1(s0(a)). If a2K, Wa1(s0(a))ˆGK. If a2= K, let us take LˆK(a)Qand let us consider HL the corresponding subgroup ofL:

HLˆ fm2GK:m(a)ˆag. ThenWa1(s0(a))ˆs0HL, the left coset ofs0in GK=HL. Since HL has the index mˆdegKa;HL is closed and open in GK: Hence s0HL is also closed and open. It follows that Wa is a continuous map which has the same value on each siHL, where

GKˆeHL[s2HL[. . .[smHL is the partition of GK, corresponding to

the left classes ofGK=HL:

COROLLARY 1. There is one and only one isometry and K-algebrae morphismWwhich extend by continuityWtoQeK:W(f)ˆ lim

n!1Wan, where f 2QeK, f ˆ lim

n!1an relative tok:kKandan2Q;nˆ1;2;. . ..

REMARK 1. Since ks(a)kKˆ kakK for any s2GK and a2Q, any s2GKis a continuous automorphism ofQrelative tok:kK. This is the main reason to consider the K-spectral norm on Q instead of usual absolute

(5)

valuej:j. Iffangnis a Cauchy sequence inQrelative tok:kK, thenfs(an)gn is also a Cauchy sequence inQrelative to the same norm. So the element s(Aˆ lim

n!1an)ˆ lim

n!1s(an) does exist in QeK. We denote it by s(A). If fn!k k: f inC(GK) thenkfk ˆ lim

n!1kfnk, soks(A)kKˆ kAkKfor anyA2QeK. MoreoverkAkKˆ lim

n!1k kan K.

Sometimes we identifyQeK with its imageW(QeK) inC(GK).

DEFINITION1. (see also [13]). Assume thate2GK. A mapf :GK !C is said to be symmetric iff(es)ˆf(s), wheref(s) is the usual conjugate of f(s) inCandsis any element ofGK.

The K-subalgebra of all symmetric maps ofe C(GK) is denoted by Csym(GK). It is closed inC(GK), so it is aK-Banach subalgebra ofe C(GK).

THEOREM1. Let K be a proper subfield ofR\Q.ThenQeK isK-iso-e morphic and isometric withCsym(GK).

PROOF. Sincee2GKin this case (e2GKif and only ifKR\Q), we can use the same technique as in [13]. However we prefer here a direct proof, which could be useful in other future researches. First of all we must remark thatWa, for anya2Q, is a symmetric continuous map. Since any x2QeK is a limit ofxn 2Qwith respect to k:kK, one has that Wxn!Wx in C(GK); where Wx(s)ˆs(x) (see Remark 1). One can use the fact that the conjugation automorphism e is continuous relative to j:j on C. So

n!1lim Wxn

(es)ˆe lim

n!1Wxn(s)

ˆ lim

n!1Wxn(s), i.e.Wx2 Csym(GK). We prove now that any element f of Csym(GK) can be approximated by ``algebraic numbers''Wx, wherex2Q. Letf b e inCsym(GK) and lete>0 be a small positive real number. LetHbe a finite index subgroup ofGK;e2= Hand such that ifGK ˆs1H[. . .[smH; s1ˆe, is the corresponding partition ofGKinto left cosets, one hasjf(s) f(m)j5e3whenevers;m2siHfor an i2 f1;2;. . .;ng. Let us writeliˆf(si);foriˆ1;2;. . .;n. Sincee2GKwe can divides1;. . .;sminto two classes relative to the subgroupfe;eg:

fs1;. . .;smg ˆ fs1;. . .;st;st‡1ˆes1;. . .;s2t ˆestg; mˆ2t Since f(esj)ˆf(sj);jˆ1;2;. . .;t, one has that lt‡hˆlh for any

hˆ1;2;. . .;t. We focus only on the points l1;l2;. . .;lt in the com-

plex plane C. Since K6ˆR\Q, the codimension of K in R\Q is 1 (see Artin-Schreier theory [6]). So one can take a subfield L of Q;LK;[L:K]ˆlt. We can enlarge L to L0Q such that iˆ 

p 1

2L0andL0=Kbe a normal extension. So this lastL0is dense inC.

(6)

We can find x1;x2;. . .;xt2L0 such that jsi(xi) lij5e

3;iˆ1;2;. . .;t.

One can now use the Approximation Theorem (see [1]) to find x2L0 such thatjx xijsi ˆjsi(x) si(xi)j5e

3for anyiˆ1;2;. . .;t. This means thatjsi(x) lij52e

3 for anyiˆ1;2;. . .;t. Butjesi(x) lij52e

3 implies that jsi(x) lij52e

3 for all iˆ1;2;. . .;m. Let us take s2GK and choose j2 f1;2;. . .;mgsuch thatjf(s) ljj5e

3. Thenjf(s) sj(x)j5efor thisj.

We can enlargeL0such thatHˆGal(Q=L0) be good enough for oure>0.

Thens2sjHmeans thats(x)ˆsj(x) b ecausex2L0. Sojf(s) s(x)j5e orjf(s) Wx(s)j5efor anys2GK. Hencekf Wxk5e for anx2Qand

the theorem is completely proved. p

REMARK 2. IfKˆR\Q, then GK ˆ fe;eg because Qˆ(R\Q)(i).

So C(GK)ˆCC and Csym(GK)ˆCˆ f(a;a)ja2Cg CC. But QeKˆR[i]ˆC. Hence Theorem 1 is true even if K ˆR\Q. If KR\Q and [R\Q:K]51, then [Q:K]51 and K would be real closed in Q. Then QˆK(i)ˆ(R\Q)(i) and [Q:K]ˆ[Q:R\Q]ˆ2 which impliesKˆR\Q(see Artin-Schreier theory [6]).

THEOREM2. Let K6R\Qbe a subfield ofQ.ThenQeK is isometric and isomorphic as aC-algebra withC(GK).

PROOF. In fact K6R\Q means that e2= GK. So the ideas of [13]

cannot be applied in this case. But in this caseKis dense inQand inC. Since any continuous function f :GK !C can be uniformly approximated by locally constant functions with a finite number of values, one has only to prove that such a mapg:GK !C;g(s)ˆli for anys2siH;i2 f1;2;. . .;mg, whereGK ˆs1H[s2H[. . .[smH;s1ˆe;[GK :H]ˆm, is a partition of GKrelative to a closed and open subgroupHofGK, can be approximated by elementsWx;withx2Q. Lete>0 be a small real positive number such that

D(li;e)ˆ fy2C:jy lij5eg;iˆ1;2;. . .;m

be a disjoint set of balls centered in li. Since K is dense in C, we can take x1;x2;. . .;xm2K and xi2D li;e

2

for all iˆ1;2; ::;m. Let Lˆ fy2Q:m(y)ˆy for anym2Hg be the fixed field of H. Since

e2= GK; fs1;s2;. . .;smg give rise to independent absolute values

:

j js1; :j js2;. . .; :j jsmonL:We use now the Approximation Theorem [1] and

(7)

find x2L such that

jx xijsiˆjsi(x) si(xi)j5e

2 for anyiˆ1;2;. . .;m:

Butsi(xi)ˆxibecausexi2K, so we get jsi(x) xij5e

2;iˆ1;2;. . .;m:

This last inequality together withjxi lij5e 2imply jsi(x) lij5e for any iˆ1;2;. . .;m:

Let us take m2GK and j2 f1;2;. . .;mg such that m2sjH, then m(x)ˆsj(x) and g(m)ˆlj. Hence jWx(m) g(m)j5e for anym2GK; so kWx gk5e for an x2Qand the proof of the theorem is complete. p THEOREM 3. Let KR\Q be a subfield of Q (e2GK).Then C(GK)QeK[ei], whereei is the constant functionei(s)ˆiˆ 

p 1

for any s2GK.

This is an isomorphism ofR- algebras.If on QeK[ei]we consider the componentwise topology induced by QeK, this last isomorphism is also a homeomorphism.

PROOF. The idea of the proof is the same like in [13]. For anyf 2GKwe consider:

f1(s)ˆRe f(s)‡Re f(es) 2

Im f(s) Im f(es)

2i ; and

f2(s)ˆRe f(s) Re f(es)

2i ‡Im f(s)‡Im f(es)

2 ;

where Re(a‡ib)ˆa and Im(a‡ib)ˆb. It is not difficult to see that f1;f22 Csym(GK) andf ˆf1‡eif2. p The following three theorems are the natural generalizations of Theo- rems 1, 2 and 3. We do not give the proofs of these theorems here because the ideas of the proofs are the same like those used to prove the above last three theorems.

THEOREM4. Assume e2GKand let LQ;such that L=K be a sub- extension ofQ=K.LetLeKbe the completion of L relative to the K-spectral norm.Let Hˆ fs2GK:s(x)ˆx for all x2Lg be the subgroup of GK which fixes L.Then the natural embedding WH:L ! C(GK=H), where

(8)

GK=H is the quotient topological space of the left classes of H in G, WH(x)ˆWx, whereWx(sH)ˆs(x)can be extended to aK-isomorphism ande isometryWH:LeK ! Csym(GK=H).

THEOREM5. If e2= GK, then the same isomorphism and isometry like that in Theorem 2, give rise to a K-Banach algebra isomorphism (iso-e metry)LeK C(GK=H).

THEOREM6. If e2GK, then the same isomorphism and isometry like that in Theorem 2, give rise to a K-Banach algebra isomorphism (iso-e metry)C(GK=H)eLK[ei].

Theorem 2.2 of [13] can be rewrite in this new general context. We do not prove it because the ideas of the proof are the same like in [13].

THEOREM7. For any LQ;KLQ,LeKhas a topological generic element x, i.e. LeKˆK[x].In particular, there exist zg 2QeK such that QeKˆK[z].g

This last result can be considered as a generalization of the classical primitive element theorem from field theory. It will be useful for future research on the structure of the groupAutcont(QeK):

3. LetKbe a subfield ofQand letC(GK) be theC-Banach algebra of all continuous functions defined onGK ˆGal(Q=K), with values inC. For any f 2 C(GK) one denotes byv(f)ˆinffjf(s) f(m)j:f(s)6ˆf(m);s;m2GKg orv(f)ˆ0 iff is a constant function. It is easy to see thatv(f)>0 if and only iff has a finite number of distinct values and at least two (f is not constant). Here is a Krasner type lemma [1] forC(GK):

LEMMA 2. Let f;g2 C(GK)such that f is a locally constant map and v(f)>0.Let us assume that kf gk5v(f)

2 and that g(s)ˆg(m) for s;m2GK.Then f(s)ˆf(m).

PROOF. Let us suppose thatf(s)6ˆf(m). Thenv(f) jf(s) f(m)j jf(s) g(s)j‡jg(m) f(m)j 2kf gk5v(f), a contradiction. So f(s)ˆ

ˆf(m). p

Using this result one can obtain an Archimedean version of Krasner Lemma [1] forQ.

(9)

COROLLARY 2. Let a;b2Q, a2= K such that ka bkK5v(a) 2 , where v(a)ˆinffjaj aij:aj6ˆaigandai;ajare all the conjugates ofaˆa1inQ over K.Then K(a)K(b).

PROOF. Let Wa and Wb be the corresponding locally constant maps in C(GK) of a and b respectively (Wa(s)ˆs(a);s2GK, etc.). Then v(Wa)ˆinffjs(a) m(a)j:s(a)6ˆm(a);s;m2GKg ˆv(a). Since kWa Wbk ˆ

ˆ ka bkK5v(Wa)

2 we can apply Lemma 2 and obtain that Wb(s)ˆWb(m)

ˆ)Wa(s)ˆWa(m). Let us takesˆe, the identity ofGK. This last implication becomesm(b)ˆb)m(a)ˆaorm2Gal(Q=K(b)))m2Gal(Q=K(a)), i.e.

K(a)K(b). p

This last result can be generalized as follows.

THEOREM 8. Let f;g be two locally constant maps in C(GK) with v(f)>0(f not constant).LetVCbe a subfield ofCsuch that f(s)and g(s)2V for all s2GK.We assume that kf gk5v(f)

2 .Then V[f] V[g] in C(GK).Here V[f]ˆ fa0‡a1f‡. . .‡atftja0;a1;. . .;at 2Vg and fi(s)ˆ[f(s)]i for any s2GK and i2N.Moreover, if kf gk51

2minfv(f);v(g)g and ifv(g)>0, thenV[f]ˆV[g].

PROOF. SinceGK is a compact groupf andghave a finite number of values. Let m be the number of distinct values of g:g(s1)ˆl1;

g(s2)ˆl2;. . .;g(sm)ˆlm2V. Then we can solve the linear system ofm

equations andmunknownsA0;A1;. . .;Am 1:

f(s1)ˆA0‡A1l1‡A2l12‡. . .‡Am 1l1m 1 f(s2)ˆA0‡A1l2‡A2l22‡. . .‡Am 1l2m 1

: : : : : :

: : : : : :

: : : : : :

f(sm)ˆA0‡A1lm‡A2lm2‡. . .‡Am 1lmm 1 The determinant D of the system is equal to Q

i5j(li lj)6ˆ0. So AiˆDi

D 2V for all iˆ0;1;. . .;m 1. Now, if g(s)ˆg(si) for an

iˆ1;2;. . .;m, thenf(s)ˆf(si) (see Lemma 2) for the samei. Hence, for

(10)

thesefAig;iˆ1;2;. . .;mwe can write:

f(s)ˆA0‡A1g(s)‡A2g2(s)‡. . .‡Am 1gm 1(s); i:e:

f ˆA0‡A1g‡A2g2‡. . .‡Am 1gm 1; or

f 2V[g]. ThereforeV[f]V[g]:The last statement is simply obtained be

changingf andg;one to each other. p

COROLLARY 3. Let B be a C-subalgebra of C(GK) and let f be a nonconstant locally map inB, the topological closure of B ine C(GK).Then f 2B.

PROOF. SinceBis dense inB, lete gb e inBsuch thatkf gk5v(f) 2 . By Theorem 8 we obtain thatC[f]C[g]B, i.e.f 2B. p PROPOSITION2. LetKLQbe subfields of the algebraic closure of Q.LetQeKbe theK-spectral closure ofQand letLeKQeKthe topological closure ofLinQeK(relative to theK-spectral norm).ThenLeK\QˆL.

PROOF. Let x2LeK\Q. If x2K; then x2L: If x2= K; then v(x)ˆminfjxj xij:xj6ˆxigwherexiare all theK-conjugates ofxˆx1

in Q, is not zero. Let y2L such that kx ykK5v(x)

2 . Then K(x) K(y)L(see Corollary 2). Hencex2Land soLeK\QˆL. p We recall that the mappingW:Q! C(GK);defined byW(x)ˆWx;where Wx(s)ˆs(x) for any x2Qands2G;is an isometric K-embedding ofQ intoC(GK):

PROPOSITION 3. Let A be a closed K-subalgebra of C(GK).Then A\W(Q)is a subfield ofW(Q).

PROOF. Let BˆA\W(Q) and let x2Bnf1g. Since QB we may suppose thatkxkK51. SinceC(GK) is a Banach algebra overC(see [5] or [18]) and because A is closed, the series 1‡x‡x2‡. . .ˆy2A. But yˆ 1

1 x, so 1 x is invertible inB(because inW(Q) it is !). So we just proved that for any x2B;x6ˆ1;1 x has an inverse inB. Letm b e a maximal ideal ofBand letz6ˆ0;z2m. Thenzˆ1 (1 z) has an inverse inB, i.e.mˆB, thusBis a subfield ofW(Q). p Propositions 2 and 3 can be viewed as the first steps to a Galois theory in QeK:

(11)

Acknowledgments. We express our gratitude to the referee for his patience in reading our paper and for his (or her) opinions which directly led us to improve the first version of the article (in particular Theorem 8).

REFERENCES

[1] E. ARTIN, Algebraic numbers and functions, Gordon and Breach, Science Publications, New York, Paris, London, 1967.

[2] V. ALEXANDRU- N. POPESCU- A. ZAHARESCU,On the closed subfields ofCp, J.

Number Theory,68, no. 2 (1998), pp. 131-150.

[3] V. ALEXANDRU - N. POPESCU - A. ZAHARESCU, Trace on Cp, J. Number Theory,88(2001), pp. 13-48.

[4] A. D. R. CHOUDARY- A. POPESCU- N. POPESCU,On the structure of compact subsets ofCp;Acta Arith.,123, no. 3 (2006), 253-266.

[5] L. GILMAN - M. JERISON, Rings of continuous functions, D. van Nostrand Company, Inc., 1960.

[6] N. JACOBSON,Lectures in Abstract Algebra III.Theory of Fields and Galois Theory, D. van Nostrand Company, Inc., Princeton, New York, 1964.

[7] J. NEUKIRCH,Algebraic number theory, Grund. Math. Wiss., Vol.322(Spring- er Verlag, 1999).

[8] A. POPESCU, On some topological representations of the absolute Galois group, JP of Algebra, N. Th. & Appl.,4(1) (2001), pp. 103-112.

[9] E. L. POPESCU,Spectral extensions, Anal. Univ. Bucuresti Mat.,51(2002), pp.

159-168.

[10] N. POPESCU,Galois action on plane compacts,in: Proc. 35th Sympos. on Ring TheoryandRepresentationTheory(Okayama,2002),Okayama,2003,pp.113-120.

[11] V. PASOL- A. POPESCU- N. POPESCU,Spectral norms on valued fields, Math.

Z.,238(2001), pp. 101-114.

[12] L. POPESCU- N. POPESCU- C. VRACIU,Completion of the spectral extension of a p-adic valuation, Rev. Roumaine Math. Pures Appl.,46(2001), pp. 805-817.

[13] A. POPESCU- N. POPESCU- A. ZAHARESCU,On the spectral norms of algebraic numbers, Math. Nach.,260(2003), pp. 78-83.

[14] A. POPESCU - N. POPESCU - A. ZAHARESCU, Galois structures on plane compacts, J. of Algebra,270(1) (2003), pp. 238-248.

[15] A. POPESCU- N. POPESCU- A. ZAHARESCU,Trace series onQeK, Results Math., 43(2003), pp. 331-342.

[16] A. POPESCU- N. POPESCU- A. ZAHARESCU,Transcendental divisors and their critical functions, Manuscripta Math.,110(2003), pp. 527-541.

[17] A. POPESCU- N. POPESCU- A. ZAHARESCU,A Galois theory for the Banach algebra of continuous symmetric functions on absolute Galois groups, Results Math.,45(2004), pp. 349-358.

[18] W. RUDIN,Real and Complex Analysis, McGraw-Hill International editions Mathematics Series (New York, Dusseldorf), 1987.

[19] A. ZAHARESCU, Some notes on Q, a letter to N. Popescu and A. Popescu,e January 6, 2000.

Manoscritto pervenuto in redazione l'11 novembre 2007.

(12)

Références

Documents relatifs

In this note (Theorem 3.1), we show how to deduce from our results on families of diophantine equations [3] some results on the approximation of an algebraic number by products

We refer to [11] for several proofs involving (5.8):.. When K contains the p-th roots of unity it can also be proved by using arguments from Kummer’s theory as follows.. The

A strong form of the Four Exponentials Conjecture states that a 2 × 2 matrix whose entries are in L e is regular, as soon as the two rows as well as the two columns are

We examine one of Toën and Vezzosi’s [63] topologies (which we call the homotopy Zariski topology) restricted to the opposite category of affinoid algebras over a non-Archimedean

Pour l’unique arbre infiniment prolongeable de K-hauteiir zéro, , il suffit de démontrer que n’importe quel nombre k d’arbres. corrects et non-équivalents

In their proposal for an algebraic K-theory Karoubi and Villamayor attach functorially to every short exact sequence of rings.. 0-> A-^ B-^

be the category of finite cell A -modules, which we regard as a Waldhausen category with w-cofibrations the maps that are isomorphic to inclusions of cell subcomplexes, and with

Colorie le dessin en respectant l’orthographe des