This paper investigates the rank of the 2-class group in certain imaginary triquadratic fields formed by adjoining an 8th root of unity and a square root of an odd square-free integer.
Contribution
It provides new insights into the structure of 2-class groups in specific imaginary triquadratic fields, extending understanding of their algebraic properties.
Findings
01
Determined conditions affecting the 2-class group rank
02
Identified patterns in the 2-class group structure for these fields
03
Extended existing theories on class groups of number fields
Abstract
Let d be an odd square-free integer and ζ8 a primitive 8-th root of unity. The purpose of this paper is to investigate the rank of the 2-class group of the fields Ld=Q(ζ8,d).
v=\left\{\begin{array}[]{cl}n(2^{n-1}-1);&\text{ if }k\text{ is real, }\\
(n-1)(2^{n-2}-1)+2^{n-1}-1&\text{ if }k\text{ is imaginary.}\end{array}\right.
v=\left\{\begin{array}[]{cl}n(2^{n-1}-1);&\text{ if }k\text{ is real, }\\
(n-1)(2^{n-2}-1)+2^{n-1}-1&\text{ if }k\text{ is imaginary.}\end{array}\right.
Cl2(Lp)=(2,2) if and only if p≡1(mod16) and (p2)4=(2p)4.
Cl2(Lp)=(2,2) if and only if p≡1(mod16) and (p2)4=(2p)4.
\left\{\begin{array}[]{ccc}a\pm 1&=&b_{1}^{2}\\
a\mp 1&=&pqb_{2}^{2}\end{array}\right.\text{ and }\left\{\begin{array}[]{ccc}x\pm 1&=&y_{1}^{2}\\
x\mp 1&=&2pqy_{2}^{2}.\end{array}\right.
\left\{\begin{array}[]{ccc}a\pm 1&=&b_{1}^{2}\\
a\mp 1&=&pqb_{2}^{2}\end{array}\right.\text{ and }\left\{\begin{array}[]{ccc}x\pm 1&=&y_{1}^{2}\\
x\mp 1&=&2pqy_{2}^{2}.\end{array}\right.
Cl2(Ld)=(2,2) if and only if (qp)=−1,
Cl2(Ld)=(2,2) if and only if (qp)=−1,
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Full text
On the rank of the 2-class group of some imaginary triquadratic number fields
Abdelmalek Azizi
Abdelmalek Azizi: Mohammed First University, Mathematics Department, Sciences Faculty, Oujda, Morocco
Let d be an odd square-free integer and ζ8 a primitive 8-th root of unity. The purpose of this paper is to investigate the rank of the 2-class group of the fields Ld=Q(ζ8,d).
Key words and phrases:
2-group rank; 2-class group; imaginary triquadratic number fields; real quadratic field.
2010 Mathematics Subject Classification:
11R11; 11R16; 11R18; 11R27; 11R29.
1. Introduction
Let k be a number field and p a prime integer. Let Clp(k) denote the p-class group of k, that is the p-Sylow subgroup of its ideal class group Cl(k) in the wide sense. Class groups of number fields have been studied for a long time, and
there are many very interesting (and very difficult) problems concerning their behavior.
An interesting invariant is the p-rank of Clp(k), i.e. the dimension of Cl(k)/Cl(k)p as a vector space over the field with p elements Fp.
In this work, we investigate the 2-rank of the class groups of some imaginary triquadratic number fields. To the best of our knowledge, there is no study in this setting and it
would be an interesting task to develop such investigations. Note that the odd part of the class group of a number field is much better understood, since it is known to be isomorphic to the direct product of the odd parts of the class groups of its quadratic subfields (see [22]).
In the literature, there are many studies that investigated the 2-rank of the class group of a number field k, let us quote some. For a quadratic field k, using Gauss’s genus theory, one can easily deduce the rank of Cl2(k). For biquadratic fields, the authors of [1, 3] determined all positive integers d such that Cl2(k) is isomorphic to Z/2Z×Z/2Z, where k=Q(d,−ℓ) and ℓ=1,2. The papers [4, 5] investigated the rank of Cl2(k) for the fields
k=Q(d,m), where m is a prime and d a positive square-free integer. In the same direction,
[7] classified all fields k=Q(d,i) such that Cl2(k) is isomorphic to Z/2Z×Z/4Z or (Z/2Z)3.
In [9, 11], E. Brown and C. Parry determined some imaginary quartic cyclic number fields k such that
the rank of Cl2(k) is at most 3. Finally, [24] determined all imaginary biquadratic fields whose 2-class group is cyclic.
Let d be an odd square-free integer and ζ8 a primitive 8-th root of unity. In the present work, we are interested in the rank of the 2-class group of the imaginary triquadratic number fields Ld:=Q(i,2,d)=Q(ζ8,d).
Methods and techniques used are based on genus theory, ambiguous class number formula, the properties of the norm residue symbol, units of some number fields and some other results of algebraic number theory.
The structure of this paper is the following. In § 2, we recall the ambiguous class number formula and the relations allowing to calculate the rank of the 2-class group of Ld. Next, we compute the number of prime ideals of K=Q(ζ8) ramified in Ld=K(d). Thereafter, in § 3, we recall the definition and some properties of the quadratic norm residue symbol. In § 4, we investigate the rank of Cl2(Ld) according to classes (mod8) in which the prime divisors of d lie. As applications, in § 5 we determine the integers d such that Cl2(Ld) is trivial, cyclic or of rank 2, and we thus deduce the integers d satisfying Cl2(Ld)≃(Z/2Z)2, we shall say Cl2(Ld) is of type (2,2). We end this section by giving the 2-part of the class number of Ld in terms of those of its biquadratic subfields.
Notations
Let k be a number field. Throughout this paper, we use the following notations.
∙
d: An odd square-free integer,
2. ∙
ζn: An n-th primitive root of unity,
3. ∙
K:=Q(ζ8),
4. ∙
Ld:=K(d),
5. ∙
δk: The discriminant of k,
6. ∙
Ok: The ring of integers of k,
7. ∙
Cl(k): The class group of k,
8. ∙
Cl2(k): The 2-class group of k,
9. ∙
h(k): The class number of k,
10. ∙
h2(k): The 2-class number of k,
11. ∙
h2(m): The 2-class number of a quadratic field Q(m),
12. ∙
N: The norm map for the extension Ld/K,
13. ∙
Ek: The unit group of Ok,
14. ∙
ed: The integer defined by (EK:EK∩N(Ld))=2ed,
15. ∙
(pα,d): The quadratic norm residue symbol over K,
16. ∙
(⋅⋅)4: The biquadratic residue symbol,
17. ∙
α: The coset of α in EK/(EK∩N(Ld)),
18. ∙
r2(d): The rank of the 2-class group of Ld,
19. ∙
εm: The fundamental unit of Q(m), where m is a positive square-free integer,
20. ∙
Wk: The set of roots of unity contained in k,
21. ∙
ωk: The cardinality of Wk,
22. ∙
k+: The maximal real subfield of k,
23. ∙
Qk: The Hasse’s index, that is (Ek:WkEk+), if k/k+ is CM,
24. ∙
Q(k′/k): The unit index of a biquadratic extension k′/k,
25. ∙
q(k):=(Ek:∏iEki), where ki are the quadratic subfields of a multiquadratic field k.
2. Preliminaries
Let K:=Q(ζ8) and Ld:=K(d), where d is an odd square-free integer. Note that
OK is a principal ideal domain. Denote by Am(Ld/K) the group of ambiguous ideal classes
of Ld/K, that are classes of Cl(Ld) fixed under any element
of Gal(Ld/K), and by Am2(Ld/K) its 2-Sylow subgroup. In fact, Am2(Ld/K)={c∈Cl(Ld):c2=1} is an elementary 2-group of order
2r2(d).
It is well known, according to the ambiguous class number formula for a cyclic extension of prime degree (cf. [17]), that
[TABLE]
It follows that
[TABLE]
where td is the number of finite and infinite primes of K which ramify in Ld/K and ed is defined by (EK:EK∩N(Ld))=2ed. So the 2-rank of Cl(Ld) verifies the relation:
[TABLE]
Let us now determine the ideals of K that ramify in Ld. Let δLd/K denote the generator of the relative discriminant of the extension Ld/K. We have:
Proposition 2.1**.**
Let d be an odd square-free integer. Then the relative discriminant of Ld/K is generated by δLd/K=d.
Proof.
Assume that d≡1(mod4). Thus δQ(d)=d, δK=28 and OQ(d)=Z[21+d].
Since δQ(d) and δK are coprime, then OLd=OK[21+d]. Hence δLd/K=discLd/K(1,21+d)=d.
If d≡3(mod4), then −d≡1(mod4). As Ld=L−d, the previous case completes the proof.
∎
The following result is easily deduced.
Corollary 2.2**.**
Let d be an odd square-free integer.
The discriminant of Ld is δLd=216⋅d4, and thus the primes that ramify in Ld are exactly 2 and the prime divisors of d.
2. 2.
The ring of integers of Ld is given by
[TABLE]
Next, we compute the number of prime ideals of K that ramify in Ld.
Theorem 2.3**.**
Let d be an odd square-free integer. Then the number of prime ideals of K that ramify in Ld is 2(q+r), where r is the number of prime integers dividing d and q is the number of those which are congruent to 1(mod8).
Proof.
Using the theorem of the cyclotomic reciprocity law ([27, Theorem 2.13]), one can verify what follows:
∙
If p is a prime integer such that p≡1(mod8), then there are exactly 4 prime ideals of K lying over p.
2. ∙
If p≡1(mod8), then there are exactly 2 primes of K above p.
By Proposition 2.1, the prime ideals of K that ramify in Ld are exactly the divisors of d in OK. So the number of those primes is 4q+2(r−q)=2(q+r) as desired.
∎
Since K is a biquadratic field, ed∈{0,1,2}. The next corollary follows directly from the previous theorem and the formula (2).
Corollary 2.4**.**
Let d be an odd square-free integer. Then h(Ld) is even in the following two cases:
∣d∣* is composite.*
2. 2.
∣d∣* is a prime congruent to 1(mod8).*
The next lemma provides the unit group of K.
Lemma 2.5**.**
The unit group of K is given by
EK=⟨ζ8,ε2⟩.
Proof.
The field K=Q(ζ8)=Q(i,2) is a Galois extension of Q whose Galois group is an elementary 2-group of order 4.
We have, Q(2) is the real quadratic subfield of K with fundamental unit ε2=1+2, which is also a fundamental unit of K. Thus by [16, Theorem 42], the Hasse’s unit index is equal to 1, i.e.,
(EK:EQ(2)WK)=1, where WK is the set of roots of unity contained in K.
∎
3. Quadratic norm residue symbol
To compute the index ed appearing in the formula (2), we will use the quadratic norm residue symbol, so we have to recall its definition and some of its properties (cf. [18, Chapter II, Theorem 3.1.3]).
Let k be a number field and β∈k∗ a square-free element in k. Let f be the conductor of k(β)/k. For any prime
p of k (finite or infinite), we denote by fp the largest power of p dividing f. Let α∈k∗, according to the approximation theorem there exists α0∈k such that
[TABLE]
If (α0)=pnP with n∈Z and (P,p)=1 (n=0 if p is infinite), set
[TABLE]
where (Pk(β)) is the Artin map applied to P. For α∈k∗ and a prime (finite or infinite) p of k, the quadratic norm residue symbol
is defined by
[TABLE]
If the prime p is unramified in k(β)/k, we set
[TABLE]
Note that the norm residue symbol may be defined more generally for an extension k(mβ)/k, where m∈\mathdsN∗, k is a number field containing the m-th root of unity and β∈k∗.
The quadratic norm residue symbol verifies the following properties that we shall use later.
(pα1α2,β)=(pα1,β)(pα2,β).
2. 2.
(pα,β)=(pβ,α).
3. 3.
If p is unramified in k(β)/k and appears with exponent e in the decomposition of (α), then
(pα,β)=(pβ)e.
4. 4.
If p is unramified in k(β)/k and does not appear in the decomposition of (α), then (pα,β)=1.
5. 5.
∏p∈Pl(pα,β)=1, where Pl is the set of all finite and infinite primes of k.
6. 6.
Let k1 be a finite extension of k, α∈k1∗ and β∈k∗. Denote by p a prime ideal of k and by B a prime ideal of k1 above p. Thus
[TABLE]
4. The rank of the 2-class group of Ld
In the present section we compute, r2(d), the rank of the 2-class group of Ld.
Since Ld=L−d=L2d, then without loss of generality, we suppose that d is an odd positive square-free integer. We shall compute r2(d) distinguishing two cases according to the classes (mod8) in which the prime divisors of d lie. Let us start with the following results.
Let k be a number field, d a positive square-free integer and α∈k∗ such that the ideal αOk is the norm of a fractional ideal of k(d).
Then, α is norm in k(d)/k if and only if (pα,d)=1 for all primes p of k ramified in k(d).
Thus, to compute ed, it suffices to compute the quadratic norm residue symbols (pζ8,d) and (pε2,d)
for the prime ideals p of K dividing d (see Proposition 2.1 and Lemma 2.5 ).
The next lemma follows directly from properties 1, 4 and 5 of the
norm residue symbol.
Lemma 4.2**.**
Let d be an odd positive square-free integer and p a prime of K dividing d. Denote by α a unit of K. Then
(pα,d)=(pα,p),* where p is the prime contained in p.*
2. 2.
p∣d∏(pα,d)=1.**
4.1. Case 1: The prime divisors of d are in the same coset of Z/8Z
Let d be an odd positive square-free integer such that the primes p∣d are in the same coset (mod8).
Lemma 4.3**.**
Let p be a prime such that p≡1(mod8) and denote by pK any prime ideal of K above p.
If p≡3(mod8), then
(pKε2,p)=−1 and (pKζ8,p)=−1.
2. 2.
If p≡5(mod8), then
(pKζ8,p)=−1 and (pKε2,p)=1.
3. 3.
If p≡7(mod8), then
(pKε2,p)=1 and (pKζ8,p)=1.
Proof.
Suppose that p is congruent to 3(mod8).
Let pQ(i) and pQ(2) be the prime ideals of Q(i) and Q(2) respectively above p. Note that these two prime ideals are totally decomposed in K. As
p is inert in both Q(i) and Q(2), then
[TABLE]
[TABLE]
We similarly prove the other cases.
∎
Theorem 4.4**.**
Let p>2 be a prime such that p≡1(mod8).
If p≡3 or 5(mod8), then the 2-class group of Lp
is trivial, i.e., r2(p)=0.
2. 2.
If p≡7(mod8), then the 2-class group of Lp
is cyclic nontrivial, i.e., r2(p)=1.
Proof.
By Theorem 2.3, we have r2(p)=2−1−ep=1−ep,
thus ep∈{0,1}. According to the previous Lemma 4.3, we have ep=1 if p≡3 or 5(mod8), and ep=0 otherwise. Hence, the results.
∎
Theorem 4.5**.**
Let d>2 be a composite odd square-free integer. Denote by r the number of distinct primes dividing d.
*If all the primes dividing d are congruent to 3(mod8) *(resp. 5\pmod{8}$$), then r2(d)=2r−2.
2. 2.
If the primes dividing d are congruent to 7(mod8), then r2(d)=2r−1.
Proof.
Assume that all the primes p∣d are congruent to 3(mod8).
Let pK be a prime ideal of K above p. By Lemmas 4.2 and 4.3, we have
[TABLE]
thus (pKζ8ε2,d)=1.
It follows that ζ8=ε2 in EK/(EK∩N(Lp)). We infer that EK/(EK∩N(Lp))={1,ζ8} and ed=1. By Theorem 2.3, the number of prime ideals of K ramified in Ld is 2r. Hence,
r2(d)=2r−1−1=2r−2.
The other cases are similarly proved.
∎
In what follows, we will compute r2(d) when all the primes p∣d are congruent to 1(mod8). For this, we need the following lemmas.
Lemma 4.6**.**
Let p be a prime such that p≡1(mod8). Then ep=0 or 1. More precisely,
ep=0* if and only if p≡1(mod16) and (p2)4=(2p)4.*
2. 2.
ep=1* if and only if
p≡9(mod16) or (p2)4=(2p)4.*
Proof.
Let p∈pQ(i)⊂pK be two prime ideals of Q(i) and K respectively. We shall calculate the two symbols (pKζ8,p) and (pKε2,p).
We have
[TABLE]
where a and b are two integers such that p=a2+b2 (see [23, page 154]). Since (pKε22,p)=(pKε22)=(pQ(2)ε22)=(−1)8p−1, where pQ(2) is an ideal of Q(2) lying over p (see [11, page 21]), then
[TABLE]
On the other hand, by [19, page 323] and [23, page 160], we have
[TABLE]
Hence
[TABLE]
If ζ8 and ε2 are not norms in Lp/K, we claim that ζ8=ε2 in EK/(EK∩N(Lp)). Indeed (pKζ8,p) and (pKε2,p) do not depend on pK so (pKζ8ε2,p)=1 for all pK. It follows that ζ8ε2=1, then ζ8=ε2 as claimed.
∎
From the previous proof, we have the following remarks.
Remark 4.7*.*
Let p be a prime such that p≡1(mod8). Then, for any prime ideal pK of K above p, we have
(pKζ8,p)=(−1)8p−1and (pKε2,p)=(p2)4(2p)4.
Lemma 4.8**.**
Let d=p1...pr>2 be an odd composite square-free positive integer such that pi≡1(mod8) for all i. Consider the following assertions:
(a)
ζ8* and ε2 are not norms in Lpi/K and Lpj/K respectively for some i=j.*
2. (b)
ζ8* is a norm in Lpj/K or ε2 is a norm in Lpi/K.*
3. (c)
ζ8* and ε2 are not norms in Ld/K.*
4. (d)
ζ8=ε2* in EK/(EK∩N(Ld)).*
Then (a) and (b) hold if and only if (c) and (d) hold.
Proof.
Assume that (a) and (b) hold, then
(piζ8,pi)=−1=(pjε2,pj),
where pi (resp. pj ) is a prime ideal of K lying above pi (resp. pj).
So by Lemma 4.2 we get
[TABLE]
and (c) follows.
If (piε2,pi)=1, then
(piζ8ε2,d)=(piζ8ε2,pi)=(piζ8,pi)(piε2,pi)=−1. Thus ζ8ε2=1. Hence ζ8=ε2. (We similarly treat the case (piζ8,pi)=1). So the assertion (d).
Conversely, suppose (c) and (d) hold. Since ζ8 and ε2 are not norms in Ld/K, then there exist i and j such that (piζ8,pi)=−1 and
(pjε2,pj)=−1. Suppose this is true only for i=j, then for all k, we get (pkζ8,pk)(pkε2,pk)=(pkζ8ε2,pk)=1. So ζ8ε2 is a norm in Ld/K, which contradicts (d). Thus (a) holds. Suppose that (b) is not verified, then for all i=j satisfying (a) and not verifying (b), we have
[TABLE]
Thus for all k, (pkζ8ε2,pk)=1, i.e., (pζ8ε2,d)=1 for all prime p of K ramified in Ld. It follows that, ζ8ε2 is a norm in Ld/K, which contradicts (d). Hence (b) holds too, which ends the proof of the lemma.
∎
Now we give an analogous of Lemma 4.6 for a composite square-free integer d>2.
Lemma 4.9**.**
Let d=p1...pr>2 be a composite odd square-free integer such that pi≡1(mod8) for all i∈I={1,⋯,r}. Then
ed=0⟺∀i∈I,pi≡1(mod16) and (pi2)4=(2pi)4.**
2. 2.
ed=1* if and only if one of the following assertions holds*
i.
∀i∈I,pi≡1(mod16) and ∃j∈I,(pj2)4=(2pj)4,**
2. ii.
∃i∈I,pi≡9(mod16) and ∀j∈I,(pj2)4=(2pj)4,**
3. iii.
∃(i,j)∈I2,[pi≡9(mod16) and (pj2)4=(2pj)4]* and
all the couples (i,j) satisfying the last condition satisfy also [(pi2)4=(2pi)4 and pj≡9(mod16)].*
3. 3.
ed=2* if and only if there exist i=j∈I such that
pi≡9(mod16), (pj2)4=(2pj)4 and [(pi2)4=(2pi)4 or pj≡1(mod16)].*
Proof.
We have:
∙
The first assertion is deduced from Lemma 4.2, Remark 4.7 and the fact that ed=0 if and only if both ζ8 and ε2 are norms in Ld/K.
2. ∙
The second assertion is deduced too from Lemmas 4.2, 4.8, Remark 4.7 and the fact that ed=1 if and only if {(ζ8 is a norm and ε2 is not) or (ζ8 is not a norm and ε2 is) or (both ζ8 and ε2 are not norms and ε2=ζ8)}.
3. ∙
The last assertion is a result of Lemmas 4.2, 4.8 and the fact that ed=2 if and only if ζ8 and ε2 are not norms and ε2=ζ8.
∎
Now, we can easily deduce the following theorem.
Theorem 4.10**.**
Let d>2 be an odd square-free integer such that all its prime divisors are congruent to 1(mod8). If r is the number of these distinct primes divisors, then
We close this subsection with some numerical examples.
Example* 4.11**.*
For d=73⋅89⋅97, we have 73≡89≡9(mod16), 97≡1(mod16), (273)4=(289)4=−(297)4=−1 and
(732)4=(892)4=−(972)4=1. Thus, by Theorem 4.10, the rank of the 2-class group of Ld:=Q(2,i,73⋅89⋅97) equals
4⋅3−3=9 (see the third item of Lemma 4.9), and the class group of Ld by PARI/GP is of type (1224,8,4,4,4,2,2,2,2).
2. 2.
For d=73⋅89⋅113, we have 73≡89≡9(mod16), 113≡1(mod16), (273)4=(289)4=−(2113)4=−1 and (732)4=(892)4=(1132)4=1. So by Theorem 4.10, the rank of the 2-class group of Ld:=Q(2,i,73⋅89⋅113) (resp. Q(2,i,73⋅113) ) is 4⋅3−2=10 (resp. 4⋅2−2=6) (see the second item of Lemma 4.9), and the class group of Ld by PARI/GP is of type (384,32,2,2,2,2,2,2,2,2) (resp. (912,2,2,2,2,2)).
3. 3.
For d=353⋅257⋅113, we have 353≡257≡113≡1(mod16) and (2353)4=(2257)4=(2113)4=(3532)4=(2572)4=(1132)4=1. So by Theorem 4.10, the rank of the 2-class group of Ld:=Q(2,i,353⋅257⋅113) (resp. Q(2,i,257⋅113)) is 4⋅3−1=11 (resp. 4⋅2−1=7) (see the first item of Lemma 4.9), and the class group of Ld by PARI/GP is of type (408,204,2,2,2,2,2,2,2,2,2) (resp. (4368,8,2,2,2,2,2)).
4.2. Case 2: The prime divisors of d are not in the same coset of Z/8Z
In this subsection, we will make use of Lemmas 4.2, 4.3 and Remark 4.7 to determine r2(d) for any odd composite square-free integer d>2 for which the prime divisors are not in the same coset of Z/8Z. Since the number td of prime ideals of K ramified in Ld is determined by Theorem 2.3, we shall give the rank of the 2-class group of Ld in terms of td.
Theorem 4.12**.**
Let d>2 be an odd composite square-free integer such that the primes dividing d are not in the same coset (mod8).
If there exist two prime divisors p1 and p2 of d such that p1≡−p2≡5(mod8), then
r2(d)=td−3.
2. 2.
If d is divisible by a prime congruent to 3(mod8) and none of the other prime divisors of d is congruent to 5(mod8), then r2(d)=td−2 or td−3. More precisely,
r2(d)=td−3 if and only if there exists a prime p≡1(mod8) dividing d such that (p2)4=−1.
3. 3.
If d is divisible by a prime congruent to 5(mod8) and none of the other prime divisors of d is congruent to 3(mod8), then r2(d)=td−2 or td−3. More precisely,
r2(d)=td−3 if and only if there exists a prime p≡1(mod8) dividing d such that (p2)4=(2p)4.
4. 4.
If all the primes dividing d are congruent to ±1(mod8), then
r2(d)=td−1−ed1, where d1 is the product of all the primes p∣d such that p≡1(mod8).
Note that ed1 is given by Lemmas 4.6 and 4.9.
Proof.
By formula (2), the rank of the 2-class group of Ld is
r2(d)=td−1−ed. As K is a biquadratic number field, then ed∈{0,1,2}. On the other hand, the symbols (pζ8,d) and (pε2,d) are trivial for any prime ideal p of K lying over a prime p≡7(mod8) (see Lemmas 4.2, 4.3) so we will ignore them.
Let p1 be a prime of K above p1. By Lemmas 4.2 and 4.3, the units ζ8 and ε2 are not norms in Ld/K and
(p1ζ8ε2,d)=(p1ε2,p1)(p1ζ8,p1)=−1, so
ζ8=ε2 in EK/(EK∩N(Ld)). Thus ed=2. Hence the first item.
2. 2.
Let p1 be a prime dividing d such that p1≡3(mod8). Assume that d is divisible by prime p2≡1(mod8). Denote by p1 and p2 two prime ideals of K lying over p1 and p2 respectively.
By Lemmas 4.2, 4.3 and Remark 4.7, we have
(p1ζ8,d)=(p1ζ8,p1)=−1, (p1ε2,d)=(p1ε2,p1)=−1,(p2ζ8,d)=(p2ζ8,p2)=(−1)8p2−1 and similarly (p2ε2,d)=(p22)4(2p2)4.
Hence ζ8 and ε2 are not norms in Ld/K, and so ed=0, which implies that ed∈{1,2}. We have ed=2 if and only if ζ8=ε2 in EK/(EK∩N(Ld)) and this, by Lemma 4.9, can only happen for primes ≡1(mod8). Since (p2ζ8ε2,d)=(−1)8p2−1(p22)4(2p2)4=(p22)4, then ed=2 if and only if (p2)4=−1 for some prime p∣d such that p≡1(mod8).
The second item follows.
We similarly prove the third item. The fourth item is immediate.
∎
We close this subsection with the following numerical examples.
Example* 4.13**.*
For d=7⋅3⋅113 (resp. d=7⋅3⋅17), we have (2113)4=(1132)4=−(172)4=1. So by the second item of the previous theorem the rank of the 2-class group of Ld=Q(i,2,7⋅3⋅113) (resp. Ld=Q(i,2,7⋅3⋅17)) is 8−2=6 (resp. 8−3=5), and the class group of Ld by PARI/GP is of type (42,2,2,2,2,2) (resp. (12,2,2,2,2)).
2. 2.
For d=7⋅5⋅17 (resp. d=7⋅5⋅113), we have (217)4=−(172)4=(2113)4=(1132)4=1. So, by the third item of the previous theorem the rank of the 2-class group of Ld=Q(i,2,7⋅5⋅17) is 8−3=5 (resp. Ld=Q(i,2,7⋅5⋅113) is 8−2=6), and the class group of Ld by PARI/GP is of type (20,2,2,2,2) (resp. (42,6,2,2,2,2)).
3. 3.
For d=7⋅17 (resp. d=7⋅113), we have (217)4=−(172)4=(2113)4=(1132)4=1. Then, by the last item of the previous theorem the rank of the 2-class group of Ld=Q(i,2,7⋅17) is 6−1−1=4 (resp. Ld=Q(i,2,7⋅113) is 6−1−0=5), and the class group of Ld by PARI/GP is of type (20,2,2,2) (resp. (64,2,2,2,2)).
5. Applications
In this section, we will determine the integers d such that the 2-class group of Ld is trivial, cyclic or isomorphic to Z/2Z×Z/2Z. For this, we have to recall the following results.
Let S be the set of odd primes that are ramified in Q(d) and S0 its subset
consisting of those primes that are congruent to 1(mod4). Let s and s0 be the cardinality of S and S0 respectively. Then
the rank of the 2-class group of k=Q(d,i) equals
•
s+s0* if d is even and p≡1(mod8) for all p∈S0(the case S0=∅ is included here).*
•
s+s0−1* if d is even and there exists p∈S0 satisfying p≡5(mod8),*
or d is odd and p≡1(mod8), for all p∈S0(the case S0=∅ is included here).
•
s+s0−2* if d is odd and there exists p∈S0 satisfying p≡5(mod8).*
5.1. Fields Ld with trivial or cyclic 2-class group
Using Theorem 2.3 and the theorems in the previous sections, one can easily deduce the following results.
Theorem 5.4**.**
The 2-class group of Ld is trivial if and only if d is a prime congruent to either 3 or 5(mod8).
Theorem 5.5**.**
The 2-class group of Ld is cyclic nontrivial if and only if d takes one of the following forms
d=q≡7(mod8)* is a prime,*
2. 2.
d=qp, where q≡3(mod8) and p≡5(mod8) are primes.
5.2. Fields Ld with 2-class group of rank 2
By Theorem 2.3 and the theorems in the previous sections, we easily deduce the following results.
Theorem 5.6**.**
The rank of the 2-class group of Ld equals 2 if and only if d takes one of the following forms
d=q1q2, with q1≡q2≡3(mod8),
2. 2.
d=p1p2, with p1≡p2≡5(mod8),
3. 3.
d=q1q2, with q1≡3(mod8) and q2≡7(mod8),
4. 4.
d=pq, with p≡5(mod8) and q≡7(mod8),
5. 5.
d=p≡1(mod8)* is a prime satisfying [p≡9(mod16) or (p2)4=(2p)4],*
where pi,qi,p and q are prime integers.
5.3. Fields Ld with 2-class group of type (2,2)
Now we shall determine the integers d for which the 2-class group of Ld is of type (2,2). The main theorem of this subsection is the following. For the proof see Propositions 5.8, 5.9, 5.10, 5.13 and 5.14 below.
Theorem 5.7**.**
Let d>2 be an odd square-free integer. The 2-class group of Ld=Q(d,2,i) is of type (2,2) if and only if d takes one of the following forms
d=p, with p≡1(mod16) and (p2)4=(2p)4.
2. 2.
d=pq, with p≡5(mod8), q≡7(mod8) and (qp)=−1.
3. 3.
d=q1q2, with q1≡3(mod8), q2≡7(mod8) and (q2q1)=−1.
where qi,q and p are primes.
To prove this theorem, we shall check all the items of Theorem 5.6.
Proposition 5.8**.**
Let p be a prime such that p≡1(mod8). Then
[TABLE]
Moreover, h2(Lp)=h2(−2p) if and only if (p2)4=(2p)4.
Proof.
Let p≡1(mod8) be a prime.
Set Lp+=Q(2,p), K=Q(2,i) and K′=Q(2,−p). By applying Lemma 5.1 to the extension Lp/Q(2), we have
[TABLE]
We have h(Q(2))=h(K)=1. By [6, Théorème 3], QLp=1 and by Lemma 2.5QK=1. Since ωLp=ωK=8 and ωK′=2, then by passing to the 2-part in the above equality we get
[TABLE]
As ε2 has a negative norm, so by the item (2) of Section 3 of [2] we obtain that
EK′=⟨−1,ε2⟩. This in turn implies that q(K′)=QK′=1. From which we infer, by Proposition 5.2, that
h2(K′)=21⋅1⋅h2(2)h2(−p)h2(−2p)=21h2(−p)h2(−2p). It follows, by the equality (3), that
[TABLE]
Keep the notations of [10, Theorem 2], by this theorem we have
h2(−p)=4 if and only if (pe)=−1.
From the proof of [10, Theorem 1], one deduces easily that
(pe)=(p2)4(2p)4. Therefore, h2(−p)=4 if and only if (p2)4=(2p)4.
Thus by Theorem 5.6, we have only two cases to check.
∙
If (p2)4=(2p)4=(−1)8p−1, then by
[20, Theorem 2],
h(Lp+) is odd. So
[TABLE]
As h2(−p)=4, then
h2(Lp)=h2(−2p).
From [25] we deduce that h2(−2p)=4 if and only if (2p)4=1.
2. ∙
If p≡9(mod16) and (p2)4=(2p)4=−1, then by [6, Théorème 10], the rank of the 4-class group is equal to 1 and from Theorem 4.10, we infer that h2(Lp) is divisible by 8. Thus Cl2(Ld)=(2,2). Which achieves the proof.
∎
Proposition 5.9**.**
Let d=p1p2 where p1 and p2 are two primes such that p1≡p2≡5(mod8). Then h2(Ld)≡0(mod8) and Cl2(Ld)=(2,2).
Proof.
Consider the following diagram
By Kuroda’s class number formula (see [22]), we have
[TABLE]
Using Lemma 5.3, we get 8 and 4 divide h2(K1) and h2(K2) respectively. Hence, 8 divides h2(Ld) and so Cl2(Ld) is not elementary by Theorem 5.6.
∎
Proposition 5.10**.**
Let d=q1q2 where q1 and q2 are two primes such that q1≡q2≡3(mod8). Then h2(Ld)≡0(mod8) and Cl2(Ld)=(2,2).
Proof.
By Kuroda’s class number formula (see [22]), we have
[TABLE]
where k1=Q(q1q2,i) and k2=Q(2q1q2,i). Note that h2(k1) is divisible by 2 (see Lemma 5.3). On the other hand by Lemma 5.3 the rank of the 2-class group of k2 is 2 and
by [1, 7] its 2-class group is not of type (2,2) or (2,4), so
h2(k2) is divisible by 16.
Hence h2(Ld) is divisible by 8. So the result.
∎
To continue, we need the following two lemmas.
Lemma 5.11**.**
Let d=pq with p≡5(mod8) and q≡7(mod8) are primes. Then {ε2,εpq,εpqε2pq} is a fundamental system of units of both Ld and Ld+=Q(pq,2). Moreover,
ELd=⟨ζ8,ε2,εpq,εpqε2pq⟩.**
2. 2.
QLd=1* and q(Ld)=4.*
Proof.
Let ε2pq=x+y2pq and εpq=a+bpq be the fundamental units of Q(2pq) and Q(pq) respectively. It is well known that N(ε2pq)=N(εpq)=1. Then a2−1=b2pq and x2−1=2y2pq. Hence the unique prime factorization in Z implies that one of the numbers: x±1 (resp. (a±1)), p(x±1) (resp. p(a±1)) and 2p(x±1) (resp. 2p(a±1)) is a square in \mathdsN. We claim that x+1, a+1, x−1 and a−1 are not squares in \mathdsN, otherwise we get for
b=b1b2 and y=y1y2
[TABLE]
Hence 1=(pb12)=(pa±1)=(pa∓1±2)=(p±2)=(p2)=−1 and
1=(py12)=(px±1)=(px∓1±2)=(p±2)=(p2)=−1,
which is absurd.
Thus by [8, Proposition 3.3] {ε2,εpq,εpqε2pq} is the fundamental system of units of both Ld and Ld+. So ELd=⟨ζ8,ε2,εpq,εpqε2pq⟩ and QLd=1. As ∏i=17Eki=⟨i,ε2,εpq,ε2pq⟩, then q(Ld)=4.
∎
Lemma 5.12**.**
Let d=q1q2 with q1≡3(mod8) and q2≡7(mod8) are primes. Then {ε2,εq1q2,εq1q2ε2q1q2} is a fundamental system of units of both Ld and Ld+=Q(q1q2,2). Moreover,
Let d=pq where p and q are two primes such that p≡5(mod8) and q≡7(mod8). Then
[TABLE]
otherwise, h2(Ld)≡0(mod16).
Proof.
By Lemma 5.11 we have q(Ld)=4. It follows by Proposition 5.2 that we have
[TABLE]
∙
Assume that (qp)=−1, then by [15, Corollaries 19.6 and 19.7] h2(pq)=h2(−pq)=h2(2pq)=2 and by [19, p. 353] h2(−2pq)=4.
Therefore the above equation gives
[TABLE]
2. ∙
Assume that (qp)=1. By [15, Corollary 19.7] we have h2(pq)=h2(2pq)=2.
Then as above we have
[TABLE]
By Proposition B10′ of [19, p. 353], h2(−2pq) is divisible by 8. By [15, Corollaries 19.6 and 18.4 ] h2(−pq) is divisible by 4. Therefore in this case h2(Ld)
is divisible by 16.
Let d=q1q2 where q1, q2 are two primes such that q1≡3(mod8) and q2≡7(mod8). Then
[TABLE]
otherwise, h2(Ld) is divisible by 16.
Proof.
The proof is similar to the one of Proposition 5.13.
∎
Example* 5.15**.*
The examples are given by using PARI/GP software and they confirm our results.
We have (172)4=−(217)4=−1 and the 2-class group of Q(17,2,i) is of type (2,2).
2. 2.
The 2-class groups of the fields Q(21,2,i) and Q(35,2,i) are of type (2,2).
3. 3.
The 2-class group of the field Q(19⋅23,2,i) is of type (2,2). Whereas, that of Q(19⋅31,2,i) is not, since the 2-part of its class number is divisible by 16. In fact (3119)=−(2319)=1.
5.4. The 2-part of the class number of some biquadratic fields
In this subsection we use the previous results to give the 2-class number of Ld in terms of that of Q(2,−d).
Theorem 5.16**.**
Let d=pq where p, q are two primes such that p≡5(mod8) and q≡7(mod8). Then
[TABLE]
with k=Q(2,−d). Moreover,
[TABLE]
Proof.
Applying Lemma 5.1 to the extension Ld/Q(2), we get
[TABLE]
By Proposition 5.2, Lemma 5.11 and the settings on values of class numbers of quadratic fields given in the proof of Proposition 5.13 we obtain
[TABLE]
By Lemmas 2.5 and 5.11 (resp. [2, p. 19]), we have QLd=QK=1 (resp. Qk=1). Thus
h2(Ld)=h2(k).
Since by [24, Proposition 2], the rank of the 2-class group of k is 2, we have the equivalence by Proposition 5.13.
∎
Theorem 5.17**.**
Let d=q1q2 where q1, q2 are two primes such that q1≡3(mod8) and q2≡7(mod8). Then
As a continuation of this work we interested in the cases where Cl2(Ld) is of type (2n,2m) or (2,2,2), where n≥1 and m>1, and based on this work, we studied the problem of the Hilbert 2-class field tower of these fields. Note also that, we generalized some of our results on Q(ζ8,d) to the fields Q(ζ2m,d), for m≥4 (cf. [12, 13, 14]).
Acknowledgment
The authors are very grateful to the reviewer for his/her careful and meticulous reading of the paper.
Bibliography27
The reference list from the paper itself. Each links out to its DOI / PubMed record.
1[1] A. Azizi, Sur le 2 2 2 -groupe de classes d’idéaux de ℚ ( d , i ) ℚ 𝑑 𝑖 \mathbb{Q}(\sqrt{d},i) , Rend. Circ. Mat. Palermo, (2) 48 (1999), 71–92.
2[2] A. Azizi, Unités de certains corps de nombres imaginaires et abéliens sur ℚ ℚ \mathbb{Q} , Ann. Sci. Math. Québec, 23 (1999), 15–21.
3[3] A. Azizi et I. Benhamza, Sur la capitulation des 2 2 2 -classes d’idéaux de ℚ ( d , − 2 ) ℚ 𝑑 2 \mathbb{Q}(\sqrt{d},\sqrt{-2}) , Ann. Sci. Math. Québec., 29 (2005), 1-20.
4[4] A. Azizi et A. Mouhib, Le 2 2 2 -rang du groupe de classes de certains corps biquadratiques et applications, Internat. J. Math., 15 (2004), 169–182.
5[5] A. Azizi et A. Mouhib, Sur le rang du 2 2 2 -groupe de classes de ℚ ( m , d ) ℚ 𝑚 𝑑 \mathbb{Q}(\sqrt{m},\sqrt{d}) où m = 2 𝑚 2 m=2 ou un premier p ≡ 1 ( mod 4 ) 𝑝 annotated 1 𝑝𝑚𝑜𝑑 4 p\equiv 1\pmod{4} . Trans. Amer. Math. Soc., 353 (2001), 2741–2752.
6[6] A. Azizi et M. Taous, Capitulation des 2 2 2 -classes d’idéaux de k = ℚ ( 2 p , i ) 𝑘 ℚ 2 𝑝 𝑖 k=\mathbb{Q}(\sqrt{2p},i) , Acta Arith., 131 (2008), 103–123.
7[7] A. Azizi and M. Taous, Déterminations des corps k = ℚ ( d , − 1 ) 𝑘 ℚ 𝑑 1 k=\mathbb{Q}(\sqrt{d},\sqrt{-1}) dont les 2 2 2 -groupes de classes sont de type ( 2 , 4 ) 2 4 (2,4) ou ( 2 , 2 , 2 ) 2 2 2 (2,2,2) , Rend. Istit. Mat. Univ. Trieste, 40 (2008), 93–116.
8[8] A. Azizi, A. Zekhnini and M. Taous, On the strongly ambiguous classes of some biquadratic number fields, Math. Bohem., 141 (2016), 363–384.