CLASS FIELDS FROM THE FUNDAMENTAL THOMPSON SERIES OF LEVEL N = o(g)
So Young Choi and Ja Kyung Koo
Abstract. Thompson series is a Hauptmodul for a genus zero group which lies between
0(N ) and its normalizer in P SL
2(R) ([1]). We construct explicit ring class elds over an imaginary quadratic eld
K from the Thompson series Tg(α) (Theorem 4), which would be an extension of [3], Theorem 3.7.5 (2) by using the Shimura theory and the standard results of complex multiplica- tion. Also we construct various class elds over
K, over a CM-eld K(ζN+ ζ
N−1), and over a eld
K(ζN). Furthermore, we nd an explicit formula for the conjugates of T
g(α) to calculate its minimal polynomial where α(∈ H) is the quotient of a basis of an integral ideal in K.
1. Introduction
The main purpose of this paper is to study the class elds generated by singular values of Thompson series T
gat imaginary quadratic argu- ments in the complex upper half plane H, over K, CM-eld K(ζ
N+ ζ
N−1) and K(ζ
N). To this end, we recall the classical results on singular mod- uli of the elliptic modular function j for SL
2( Z) evaluated at imaginary quadratic arguments ([2], [5], [9], [11]). Let K be an imaginary quadratic eld over Q of discriminant d
Kand O be an order of K of conductor f , discriminant f
2d
Kand class number h( O). Let α ∈ H ∩ O be an imaginary quadratic argument. Then a singular modulus j(α) generates the ring class eld L of O (the Hilbert class eld if O is the maximal order of K).
Helling showed in [6] that the group
0(N )
∗generated by {
0(N ), (
0 −1N 0
) } has a genus zero exactly for N = 1 ∼ 21, 23 ∼ 27, 29, 31, 32, 35, 36, 39, 41, 47, 49, 50, 59, 71. Moreover, for all such N but 49 and 50,
Received August 6, 2003.
2000 Mathematics Subject Classication: 11F11, 11R04, 11R37.
Key words and phrases: modular functions, Thompson series, class elds.
This work was Supported by KOSEF Research Grant 98-0701-01-01-3.
0
(N )
∗has a fundamental Thompson series T
N∗(α) corresponding to itself, that is, T
N∗(α) is dened by an element of order N of the Monster group ([4], Table 2). Throughout this paper, we denote α to be a root in H of a quadratic equation az
2+ bz + c = 0 with (a, b, c) = 1 and a > 0, and K as an imaginary quadratic eld Q(α). Chen-Yui ([3], Theorem 3.7.5 (2)) showed by using the Shimura reciprocity law that when (a, N ) = 1, T
N∗(α) generates a ring class eld over K.
In §2, we rst construct some sort of class elds with
0(N )
∗by means of Shimura’s ideas.
Next in §3, we generate a ring class eld K(T
N∗(α)) over K under the condition (a, b, N ) ̸= N or (
Na, N ) ̸= 1. This further generalizes Chen-Yui’s result under the assumption (a, N ) = 1.
On the other hand, it has been known that there exists a funda- mental Thompson series T
gof level N = o(g) exactly when N = 1 ∼ 36, 38, 39, 41, 42, 44 ∼ 47, 49, 50, 51, 54 ∼ 56, 59, 60, 62, 66, 69 ∼ 71, 78, 87, 92, 94, 95, 105, 110, 119 ([4], table 2) where o(g) is the order of el- ement g of the Monster group. In §4, we will construct, from such T
g, ζ
N+ ζ
N−1or ζ
N, not only various class elds over K which are nei- ther ray class elds of conductor N nor ring class elds of order O with discriminant N
2d
K, but also ray class elds of conductor N by applying Chen-Yui’s method. We also demonstrate what sort of class eld T
g(α) generates over a CM-eld K(ζ
N+ ζ
N−1) and over a eld K(ζ
N).
In §5, we explore an explicit formula for the conjugates of T
g(α) to calculate its minimal polynomial.
Throughout the article we adopt the following notations:
• = SL
2( Z)
•
0(N ) = { (
a bc d
) ∈ SL
2( Z)|c ≡ 0 mod N}
•
0(N )
∗= ⟨
{
0(N ), (
0 −1N 0
) } ⟩
• Z
pthe ring of p-adic integers
• Q
pthe eld of p-adic numbers
• H upper half complex plane
• ζ
N= e
2πi/N• i = √
−1
• T
N∗a fundamental Thompson series for a genus zero group
0(N )
∗• T
ga fundamental Thompson series of level N = o(g)
• x ∼
SL2(Z)
y means that x = γy for some γ ∈ SL
2( Z)
• G
αthe stabilizer of α for a group G
2. Class elds obtained by applying Shimura’s method Let be a Fuchsian group of the rst kind. Then X() = \H
∗is a compact Riemann surface. Hence there exists a projective nonsingular algebraic curve V , dened over C, biregularly isomorphic to \H
∗. We specify a -invariant holomorphic map φ of H
∗to V which gives a biregular isomorphism of \H
∗to V . In that situation, we call (V , φ ) a model of \H
∗. Through this article we always assume that the genus of \H
∗is zero. Then it’s function eld K(X()) is equal to C(J
′) for some J
′∈ K(X()) and the pair ( P
1( C), J
′) is a model of \H
∗([7], Lemma 14).
Let G
Abe the adelization of an algebraic group G = GL
2dened over Q.
Put
G
p= GL
2( Q
p) (p : rational prime), G
∞= GL
2( R),
G
∞+= {x ∈ G
∞| det(x) > 0}, G
Q+= {x ∈ GL
2( Q)| det(x) > 0}.
We dene the topology of G
Aby taking U = ∏
p
GL
2(Z
p) × G
∞+to be an open subgroup of G
A. Let K be an imaginary quadratic eld and ξ
zbe an embedding of K into M
2( Q). We call ξ
znormalized if it is dened by a (
z1) = ξ
z(a) (
z1) for a ∈ K where z is the xed point of ξ
z(K
×) ( ⊂ G
Q+) in H. Observe that the embedding ξ
zdenes a continuous homomorphism of K
A×into G
A+, which we denote again by ξ
z. Here G
A+is the group G
0G
∞+with G
0the non-archimedean part of G
Aand K
A×is the idele group of K. Let Z be the set of open subgroups S of G
A+containing Q
×G
∞+such that S/Q
×G
∞+is compact. For S ∈ Z, we see that det(S) is open in Q
×A. Therefore the subgroup Q
×det(S) of Q
×Acorresponds to a nite abelian extension of Q, which we write k
S. Put
S= S ∩ G
Q+for S ∈ Z. As is well known ([11], Proposition 6.27),
S/Q
×is a Fuchsian group of the rst kind commensurable with P SL
2( Z). Let
U
0= {x = (x
p) ∈ U|x
p∈ U
p0for all nite p } and U
∗0= U
0∪U
0( N ),
where U
p0= { (
a bc d
) ∈ GL
2( Z
p) |c ≡ 0 mod NZ
p}, ( N ) = (x
p) ∈ G
A+and x
p= (
0 −1N 0
) .
Then we have
Lemma 1. (i) Q
×U
∗0∈ Z, (ii) k
S= Q,
(iii)
S= Q
× 0(N )
∗if S = Q
×U
∗0.
Proof. Since Q
×U
∗0= Q
×U
0∪ Q
×U
0( N ) and Q
×U
0is an open subgroup in G
A+, Q
×U
∗0is also an open subgroup in G
A+. Observing that Q
×U
0/ Q
×G
∞+is compact, we obtain Q
×U
∗0∈ Z. As for (ii), we see that Q corresponds to the norm group Q
×Q
×∞Awith Q
×∞A= R
×∏
p
Z
×pand det ( U
∗0)
= N det(U
0). But det(U
0) = Q
×∞Aand hence by the class eld theory k
S= Q. Indeed, clearly det(U
0) is contained in Q
×∞A. Conversely, for any element (α
p) ∈ Q
×A∞, take y
p= (
1 00 αp
) , then (y
p) ∈ U
0and det(y
p) = (det y
p) = (α
p). Lastly, we readily get that
S
= Q
×U
∗0∩ G
Q+= Q
×(
U
∗0∩ G
Q+)
= Q
× 0(N )
∗Theorem 2. Let K be an imaginary quadratic eld. For xed z ∈ K ∩ H, let ξ
zbe the normalized embedding. Then T
N∗(z) belongs to the maximal abelian extension K
abof K and K(T
N∗(z)) is the class eld of K corresponding to the subgroup K
×· ξ
z−1( Q
×U
∗0) of K
A×.
Proof. We have k
S= Q and
S= Q
× 0(N )
∗by Lemma 1. Since T
N∗gives a model of X (
0(N )
∗), the assertion follows from [11] Proposition 6.31 and Proposition 6.33.
3. Ring class elds generated by singular values of T
N˜In this section, we obtain Theorem 4 which would be an extension of [3], 3.7.5 Theorem(2) by using Shimura’s method and the standard results of complex multiplication. To this end, we need the following fact.
Theorem 3. Let F
Nbe the eld of modular functions of level N rational over Q(e
2πi/N), and let K be an imaginary quadratic eld. Let O
Kbe the maximal order of K and a be an O
K-ideal such that a = [z
1, z
2] and α = z
1/z
2∈ H. Then the eld KF
N(α) generated over K by all values f (α) with f ∈ F
Nand f dened at α, is the ray class eld K
(N )over K with modulus N .
Proof. [9], Ch. 10, Corollary of Theorem 2.
By class eld theory, the reciprocity map induces an isomorphism
[ ·, K] : K
A×/K
×U
(N )−→ Gal(K
∼ (N )/K)
where U
(N )is the subgroup of K
A×given by
U
(N )= {s ∈ K
A×|s
p∈ O
p×and s
p≡ 1 mod NO
pfor all nite primes p}.
For a subeld L of K
(N ), let
L/K
: I
K(N ) −→ Gal(L/K)
denote the Artin map. Then Ker(
K(N )/K) = P
K,1(N ), where I
K(N ) is the ideal group generated by all fractional ideals in K prime to N and P
K,1(N ) denotes the subgroup of I
K(N ) generated by the principal ideals β O
Kwith β ∈ O
Kand β ≡ 1 mod NO
K. Of course, O
Kis the ring of integers in K.
Theorem 4. Let T
N∗be a fundamental Thompson series for a genus zero group
0(N )
∗. Let α be a root in H of a quadradic equation az
2+ bz + c = 0 such that a > 0, (a, b, c) = 1, and b
2− 4ac = m
2d
K<
0 (m > 0). Let K = Q(α) and O (= Z[aα]) be an order in K of discrimi- nant m
2d
K. Assume that (a, b, N ) ̸= N or (
Na, N ) ̸= 1. Then K(T
N∗(α)) is a ring class eld of an imaginary quadratic order O
′of discriminant f
2d
Kwhere f = mN/(a, N ) and d
Kis the discriminant of K.
Proof. First of all, we describe the action of arbitrary prime deal ℘ of O
Kon T
N∗(α) under the Artin map
K(T∗N(α))/K
which is guaranteed by Theorem 4. Take a rational prime p which does not divide 2abcmN and splits in K. Then p O
K= ℘℘
′, where ℘ = (p,
−r+m√dK
2
), r
2= m
2d
Kmod p and r ∈ Z with b − r and b + r both even (since aα is a root of the polynomial z
2+ bz + ac = 0 and p splits in K, there exists r ∈ Z such that r
2= m
2d
Kmod p, z
2+ bz + ac = (z +
b−r2)(z +
b+r
2
) mod p and b − r, b + r ∈ 2Z, and aα +
b±r2=
−b+m2√dK+
b±r2=
±r+m√ dK
2
). We dene v =
r2−m4p2dK( ∈ Z) and hence ord
pv = 0.
We now take an idele s corresponding to an ideal ℘ to be s = (1, · · · , −r + m √
d
K2 , 1, · · · ),
where we put 1 at all places except the place corresponding to p. Under the embedding ξ
α: K
A×→ G
A+, s is sent to
ξ
α(s) = (I
2, · · · , (
−b−r2 −c
a b−r2
) , · · · )
because
−r+m2√dK= aα +
−r+b2and
−r+m2√dKα =
−r−b2α − c. This implies
ξ
α(s
−1) = ( I
2, · · · , 1 pv
(
b−r2 c
−a −r−b2
)
, · · · ).
Let p
−1a, k, p
−1N, l ∈ Z be such that ak + pp
−1a= 1 and N l + pp
−1N= 1.
Then
ξ
α(s
−1) = (
(
p −(r+b)kNl20 1
)
, · · · , 1 v
(
b−r2 v
−a 0a
) (
1 (r+b)(12ap−kNla)0 1
) , · · · )
· (
1p
(r+b)kN l 2p
0 1
)
= u · g.
Since 1 − kNla = p(akp
−1N+ p
−1aN l + pp
−1ap
−1N), u = (u
q) ∈ U and g ∈ GL
+2(Q). Notice that u
q≡ (
p 00 1
) mod N M
2(Z
q) for every nite prime q and dene
U f
N= (
p 00 1
) and g A
N= (
1 00 p−1
) U f
N∈ SL
2( Z/NZ).
Lift g A
Nto a matrix A
Nin
0(N ). By the Shimura reciprocity law, we get
T
N∗(α)
[℘, Kab/K]= T
N∗τ(ξα(s−1))(α) = T
N∗τ(u)τ(g)(α)
= T
N∗(A
Ngα) = T
N∗(gα)
= T
N∗(
(
1 (r+b)kN l/20 p
) α).
Let A be a matrix (
1 (r+b)kN l/2
0 p
)
and suppose that T
N∗(α)
[℘, Kab/K]= T
N∗(α). Then
A ∈
0(N )
∗GL
2( Q)
+α. If A is contained in
0(N ) (
0 −1N 0
) GL
+2( Q)
α, then A = γ (
0 −1N 0
) (
x Ny
z wN
) for some x, y, z, w ∈ Z and γ ∈
0(N ) with x
Nα+
Ny= (zα + w)α. We also see that
N za=
N wb−x=
−yc:= λ ∈ Z\{0} and (λ,N)=1 ( because if λ = 0, then w = ± √
p/N / ∈ Z. This implies that λ is not zero. We now set λ =
△˜with , △ ∈ Z\{0}, (, △) = 1, then △|(a, b, c) = 1 and hence λ ∈ Z. Also, (N, p) = 1 and xw − yz=(Nw − bλ)w + cλz = p imply (λ, N ) = 1). Here, the fact that z=
aλNis an integer means that N divides a. Moreover, since N divides x and bλ = N w − x, N divides b and hence (
Na, N ) = 1. These give a contradiction to our assumption (a, b, N ) ̸= N or (
Na, N ) ̸= 1. Therefore, A ∈
0(N )M
2( Z)
+αand we can take γ ∈
0(N ) and d, λ ∈ Z such that
A = (
1 (r+b)kN l/2
0 p
)
= γ (
d−bλ −cλaλ d
)
and (d − bλ)α − cλ = α(aλ + d). Observing p = acλ
2− bdλ + d
2=
α
′(d − aαλ − bλ)=α
′α
′where α
′:= d + aαλ ∈ O(= Z[aα]) and α
′∈ O
is the complex conjugate of α
′, we obtain p O
K=(α
′)(α
′) =℘℘
′. That is,
℘ is a principal ideal generated by an element α
′(or α
′) in O .
On the other hand, for each a ∈ I
K(N ), we can take a prime ideal
℘ ∈ I
K(2abcmN ) and a principal ideal (x) ∈ P
K,1(N ) with a = (x)℘.
Indeed, if an ideal a in K is prime to N , then a = a
∗(β) for some a
∗∈ I
K(2abcmN ) and (β) ∈ P
K,1(N ) ([10], Corollary 3.16 and 3.18), and the ideal a
∗just given can be factored into (x)℘ for some prime ideal
℘ ∈ I
K(2abcmN ) and (x) ∈ P
K,1(2abcmN ) ([8] VIII §4 ,Corollary of Theorem 8). Consequently, ker(
K(T∗N(α))/K
) is contained in a subgroup P
K,1(N )(P ( O)∩I
K(N )) of I
K(N ), where P ( O) is the group of principal O-ideals. ( Of course, we consider elements in P (O) as O
K-ideals.)
Let (β) be a principal ideal of O relatively prime to N and write β = naα + l ∈ Zaα + Z(= O). We see from Lemma 5 below that the action of (β) on T
N∗(α) is represented by a matrix A
β∈ SL
2( Z) whose image in SL
2( Z/NZ) is equal to (
−bn+l −cn anN (β)−1 lN (β)−1
)
. Thus, we derive that (β) xes T
N∗(α) if and only if A
β∈
0(N )
∗GL
2(Q)
α. First, suppose that N does not divide a. By similar arguments as we used to show that A ∈
0(N )
∗GL
2( Q)
+αimplies A ∈
0(N )M
2( Z)
+α, we can verify that (β) xes T
N∗(α) if and only if A
β∈
0(N )SL
2( Z)
α. But we know that SL
2( Z)
αis trivial unless α is SL
2( Z)-equivalent to e
2πi/rwith r ∈ {3, 4}.
Assuming SL
2( Z)
α= {±1}, we see that A
β∈ ±
0(N ) if and only if N |an. Therefore the principal ideals in O which x T
N∗(α) are of the form (β) with disc(β) dividing (mN/(a, N ))
2d
K. But these principal ideals are all in P ( O
′). The cases α ∼
SL2(Z)
e
2πi/rwith r=3 or 4 can be treated similarly. These prove our assertion in the case that N does not divide a. Now suppose that N divides a. Then A
βxes T
N∗(α) and hence T
N∗(α) generates the ring class eld of O
′.
Lemma 5. Let f be a modular function of level N with rational Fourier coecients and (β) a principal ideal of O relatively prime to N . Write β = n(aα) + l ∈ Z(aα) + Z(= O). And let A
βbe a matrix in SL
2( Z) whose image in SL
2( Z/NZ) is equal to (
−bn+l −cn anN (β)−1 lN (β)−1
) . Then the action of (β) on f (α) is given by
f (α)
[(β), Kab/K]= f (A
β· α) where [ · , K
ab/K] is the Artin map.
Proof. Let (β) correspond to an idele s = (1, . . . ; β, . . . ) ∈ K
A×where
we put 1 on each place dividing N and β on the other places. Under the
embedding ξ
α: K
A×→ G
A+, s is sent to ξ
α(s) = (
I
2, (
−bn+l −cnan l
) , . . . )
because β · α = (naα + l)α = (−bn + l)α − cn and β · 1 = na · α + l. Then ξ
α(s
−1) =
( I
2, 1
N (β)
(
l cn−an −bn+l
) , . . . )
= ((
−bn+l −cnan l
) , I
2, . . . )
· 1 N (β)
(
l cn−an −bn+l
)
= u · g where u = ((
−bn+l −cnan l
) , I
2, . . . )
and g =
N (β)1(
l cn−an −bn+l
) ∈ G
Q+. Since det (
−bn+l −cnan l
) = N (β) is relatively prime to N , u belongs to U . Write u = (u
p). Note that u
pis congruent to (
−bn+l −cnan l
) modulo N M
2(Z
p) for any nite prime p. Put f U
N= (
−bn+l −cnan l
) ∈ GL
2(Z/NZ) and A
β=
(
1 00 det(gUN)−1
) (
−bn+l −cnan l
) =
(
−bn+l −cn anN (β)−1 lN (β)−1) ∈ SL
2( Z/NZ).
We now lift A
βto a matrix A
βin SL
2(Z). Then we derive by Shimura reciprocity law
f (α)
[(β),Kab/K]= f
τ (ξα(s−1))(α) = f
τ (u)τ (g)(α)
= f ( A
β· gα) since f has rational Fourier coecients
= f (A
βα) since g = ξ
α(β
−1).
4. Class elds over an imaginary quadratic eld K, CM- elds K(ζ
N+ ζ
N`1) and K(ζ
N)
Theorem 6. Let T
gbe a fundamental Thompson series of level N = o(g). Let α be a root in H of a quadratic equation az
2+ bz + c = 0 such that a > 0, (a, b, c) = 1, and b
2− 4ac = d
K< 0. Let K = Q(α) and O(N) be an order Z + NO
Kin K of discriminant N
2d
Kand put K
′be a CM-eld K(ζ
N+ ζ
N−1). Under the assumption (a, N ) = 1 the following assertions hold.
(1) K
′(T
g(α)) is a class eld over K with
Gal(K
′(T
g(α))/K) ∼ = I
K(N )/ e P
K(N ) and
[K
′(T
g(α)) : K] = h( O(N))ψ(N)
2[ e P
K(N ) : P
K,1(N )]
where e P
K(N ) is a subgroup of I
K(N ) generated by all principal ideals β O
Ksuch that β = naα + l ∈ Zaα + Z(= O
K), N |n and l
2≡ ±1modN, and ψ(1) = ψ(2) = 2 and ψ is the Euler φ-function for N ≥ 3.
(2) T
g(α) generates a class eld K
′(T
g(α)) over K
′with Gal(K
′(T
g(α))/K
′) ∼ = I
K′(N )/N
−1K′/K
( e P
K(N )) and
[K
′(T
g(α)) : K
′] = h( O(N)) [ e P
K(N ) : P
K,1(N )]
where h( O(N)) is the class number of O(N) and I
K′(N ) is the ideal group generated by all fractional ideals in K
′prime to N .
Proof. In the proof of Theorem 4, take m = 1, give a condition (N, a) = 1 and replace T
N∗(α) by T
g(α). Then the arguments from the beginning to the step of letting a matrix A = (
1 (r+b)kN l2
0 p
)
in M
2( Z) are exactly the same as those in Theorem 4. We assume that T
g(α)
[p, Kab/K]= T
g(α). Then A ∈
0(N )M
2( Z)
+α. Indeed, if A ∈ W
eGL
+2( Q)
α, then A
β= (
Ae BCN De
) (
d−b△2 −c△2 a△2 d)
for Ae
2−BcN = e, (2, △) = 1, e|△, △ ̸= 1 and A, B, C, D, e, λ, △, 2 ∈ Z. A matrix (
Ae BCN De
)
−1A
βimplies that a
△2is an integer, and hence e divides a. This is absurd. Here W
eis a non- trivial Atkin-Leher involution. Therefore, likewise as in the proof of The- orem 4, ker(
K′(Tg(α))/K) is contained in a subgroup P
K,1(N )(P (O
K) ∩ I
K(N )) of I
K(N ).
Let (β) be a principal ideal of O
Krelatively prime to N and write β = naα + l ∈ Zaα + Z(= O
K). Since we see from Lemma 5 that the action of (β) on T
g(α) is represented by a matrix A
β∈ SL
2( Z) whose image in SL
2( Z/NZ) is equal to (
−bn+l −cn anN (β)−1lN (β)−1
)
, we get that (β) xes T
g(α) if and only if A
β∈
0(N )SL
2(Z)
αby replacing A by A
βin the above argument. This implies that (β) ∈ ker(
K′(Tg(α))/K) if and only if A
β∈
0(N )SL
2( Z)
αand N (β) = ±1 mod N. But we know that SL
2( Z)
αis trivial unless α is SL
2( Z)-equivalent to e
2πirwith r ∈ {3, 4}.
Assuming SL
2(Z)
α= {±1}, we obtain that
(β) ∈ ker(
K′(Tg(α))/K) if and only if N |an and N(β) ≡ ±1 mod N
if and only if N |n and l
2≡ ±1 mod N
if and only if (β) ∈ e P
K(N ).
The case where α is SL
2( Z)-equivalent to e
2πirwith r ∈ {3, 4} can be treated similarly. These prove (1).
For any a ∈ I
K′(N ), [a, K
′(T
g(α))/K
′] = [N
K′/K(a), K
′(T
g(α))/K]
and N
K′/K(I
K′(N )) ⊂ I
K(N ) implies
Gal(K
′(T
g(α))/K
′) ∼ = I
K′(N )/N
−1K′/K
( e P
K(N )).
Since P
K,1(N ) is contained in e P
K(N ), we obtain from the eld tower below
[K
′(T
g(α)) : K
′] = ψ(N )h( O(N))
2[K
(N ): K
′(T
g(α))][K
′: K] = h( O(N)) [ e P
K(N ) : P
K,1(N )] .
K Q(ζ
N+ ζ
N−1) = K
′K ∩ Q(ζ
N+ ζ
N−1) = Q
K Q(ζ
N+ ζ
N−1)
Q Q Q Q Q
Q Q Q Q Q
Corollary 7. Under the same assumptions and notations as in Theorem 6, we have that if x ∈ Z and x
2≡ ±1 mod N implies x ≡
±1 mod N, then
(1) K
′(T
g(α)) = K
(N )is the ray class eld over K with modulus N and
[K
(N ): K] = ψ(N )h(O(N))
2 ,
(2) T
g(α) generates a class eld K
′(T
g(α)) over K
′with Gal(K
′(T
g(α))/K
′) ∼ = I
K′(N )/N
−1K′/K
(P
K,1(N )) and
[K
′(T
g(α)) : K
′] = h( O(N)).
Proof. These are clear from Theorem 6
Examples 1. For N = 2, 3, 4, 6, 7, 9, 11, 14, 18, 19, 22, 23, 27, 31, 38, 46, 47, 54, 59, 62, 71, 94, x ∈ Z and x
2≡ ±1 mod N imply x ≡ ±1 mod N so that K(ζ
N+ ζ
N−1, T
g(α)) = K
(N )is the ray class eld over K with modulus N . On the other hand, observing that
P e
K(13)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 13|n and l ≡ 1 or 5 (mod 13)} ⟩ , [ e P
K(13) : P
K,1(13)] = 2, and
[K(ζ
13+ ζ
13−1, T
g(α)) : K(ζ
13+ ζ
13−1)] = h( O(13)) 2
we see that K(ζ
13+ ζ
13−1, T
g(α)) is neither a ring class eld of an order Z + 13O
Knor a ray class eld K
(13)with moduluds 13. We also have similar examples:
P e
K(15)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 15|n and l ≡ 1 or 4 mod 15}⟩ , P e
K(16)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 16|n and l ≡ 1 or 7 mod 16}⟩ , P e
K(17)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 17|n and l ≡ 1 or 4 mod 17}⟩ , P e
K(20)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 20|n and l ≡ 1 or 9 mod 20}⟩ , P e
K(21)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 21|n and l ≡ 1 or 8 mod 21}⟩ , P e
K(25)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 25|n and l ≡ 1 or 7 mod 25}⟩ , P e
K(26)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 26|n and l ≡ 1 or 5 mod 26}⟩ , P e
K(28)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 28|n and l ≡ 1 or 13 mod 28}⟩ , P e
K(29)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 29|n and l ≡ 1 or 12 mod 29}⟩ ,
P e
K(30)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 30|n and l ≡ 1 or 11 mod 30}⟩ , P e
K(32)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 32|n and l ≡ 1 or 15 mod 32}⟩ , P e
K(33)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 33|n and l ≡ 1 or 10 mod 33}⟩ , P e
K(34)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 34|n and l ≡ 1 or 13 mod 34}⟩ , P e
K(35)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 35|n and l ≡ 1 or 6 mod 35}⟩ , P e
K(36)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 36|n and l ≡ 1 or 17 mod 36}⟩ , P e
K(39)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 39|n and l ≡ 1 or 14 mod 39}⟩ , P e
K(41)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 41|n and l ≡ 1 or 9 mod 41}⟩ , P e
K(42)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 42|n and l ≡ 1 or 13 mod 42}⟩ , P e
K(44)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 44|n and l ≡ 1 or 21 mod 44}⟩ , P e
K(45)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 45|n and l ≡ 1 or 19 mod 45}⟩ , P e
K(50)
= ⟨{βO
K| β = naα + l ∈ Zaα + Z, 50|n and l ≡ 1 or 7 mod 50}⟩ , P e
K(51)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 51|n and l ≡ 1 or 16 mod 51}⟩ , P e
K(55)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 55|n and l ≡ 1 or 21 mod 55}⟩ , P e
K(66)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 66|n and l ≡ 1 or 23 mod 66}⟩ ,
P e
K(69)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 69|n and l ≡ 1 or 22 mod 69}⟩ , P e
K(70)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 70|n and l ≡ 1 or 29 mod 70}⟩ , P e
K(78)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 78|n and l ≡ 1 or 25 mod 78}⟩ , P e
K(87)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 87|n and l ≡ 1 or 28 mod 87}⟩ , P e
K(92)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 92|n and l ≡ 1 or 45 mod 92}⟩ , P e
K(95)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 95|n and l ≡ 1 or 39 mod 95}⟩ , P e
K(110)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 110|n and l ≡ 1 or 21 mod 110}⟩ , P e
K(119)
= ⟨{βO
K|β = naα + l ∈ Zaα + Z, 119|n and l ≡ 1 or 50 mod 119}⟩ , [ e P
K(N ) : P
K,1(N )] = 2, and [K(ζ
N+ ζ
N−1, T
g(α)) : K(ζ
N+ ζ
N−1)] =
h(O(N))
2
for N = 13, 15, 16, 17, 20, 21, 25, 26, 28, 29, 30, 32, 33, 34, 35, 36, 39, 41, 42, 44, 45, 50, 51, 55, 66, 69, 70, 78, 87, 92, 95, 110, 119. Meanwhile
P e
K(56) =< {βO
K|β = naα + l ∈ Zaα + Z, 56|n and l ≡ 1 or 13 or 15 or 27 (mod 56) }t >,
P e
K(60) =< {βO
K|β = naα + l ∈ Zaα + Z, 60|n and l ≡ 1 or 11 or 19 or 29 (mod 60) } >,
P e
K(105) =< {βO
K|β = naα + l ∈ Zaα + Z, 105|n and l ≡ 1 or 29 or 34 or 41 (mod 105) } >,
[ e P
K(N ) : P
K,1(N )] = 4, and [K(ζ
N+ ζ
N−1, T
g(α)) : K(ζ
N+ ζ
N−1)] =
h(O(N))
4
for N = 56, 60, 105.
Theorem 8. Under the same assumptions and notations as in The- orem 6, we let E = K(ζ
N) = Q(α, ζ
N). Then the followings hold : (1) E(T
g(α)) is a class eld over K with
Gal(E(T
g(α))/K) ∼ = I
K(N )/ e P e
K(N ) and
[E(T
g(α)) : K] = h(O(N))ψ(N) 2[ e P e
K(N ) : P
K,1(N )]
,
where e P e
K(N ) is a subgroup of I
K(N ) generated by all principal ideal β O
Ksuch that β = naα + l ∈ Zaα + Z(= O
K), N |n and l
2≡ 1 mod N.
(2) T
g(α) generates a class eld E(T
g(α)) over E with Gal(E(T
g(α))/E) ∼ = I
E(N )/N
E/K−1( e P e
K(N )) and
[E(T
g(α)) : E] =
h(O(N))
[ ePeK(N ):PK,1(N )]
if K ⊂ Q(ζ
N),
h(O(N))
2[ ePeK(N ):PK,1(N )]
otherwise
where I
E(N ) is the ideal group generated by all fractional ideals in E prime to N .
Proof. We can show (1) by the same arguments as in Theorem 6.
Considering the eld tower bellow we have (2).
E = K(ζ
N)
K ∩ Q(ζ
N)
K Q(ζ
N)
Q Q Q Q Q
Q Q Q Q Q
. Q
Corollary 9. Under the same assumptions and notations as in The- orem 8, We derive that if x
2≡ 1 mod N implies x ≡ ±1 mod N then
(1) E(T
g(α)) = K
(N )is a ray class eld over K with modulus N and [K
(N ): K] = ψ(N )h(O(N))
2 ,
(2) T
g(α) generates a class eld E(T
g(α)) over E with Gal(E(T
g(α))/E) ∼ = I
E(N )/N
E/K−1(P
K,1(N )) and
[E(T
g(α)) : E] =
{ h(O(N)) if K ⊂ Q(ζ
N),
h(O(N))
2
otherwise.
Proof. They are clear from Theorem 8.
Examples 2. For N = 5, 10, 13, 17, 25, 26, 29, 41, 50, x
2≡ 1 mod N implies x ≡ ±1 mod N and hence K(ζ
N, T
g(α))= K
(N )is the ray class eld over K with modulus N .
On the other hand, taking α as a root in H of the equation z
2+ 2 = 0 we have
K = Q( √
−2)(̸∈ Q(ζ
6)), [Q( √
−2, ζ
6)(T
6∗( √
−2)) : Q( √
−2, ζ
6)]
= h( O(N))
2 and Q( √
−2, ζ
6, T
6∗( √
−2))
= Q( √
−2, ζ
6+ ζ
6−1, T
6∗( √
−2)) = Q( √
−2)
(6).
Taking α as a root H of the equation z
2+ 3 = 0 we also get that K = Q( √
−3)(∈ Q(ζ
6)), [ Q( √
−3, ζ
6)(T
6∗( √
−3)) : Q( √
−3, ζ
6)]
= h( O(N)) and Q( √
−3, ζ
6, T
6∗( √
−3))
= Q( √
−3, ζ
6+ ζ
6−1, T
6∗( √
−3)) = Q( √
−3)
(6).
Remark. Under the same assumptions and notations as in Theorem 6,8 let β be a root in H of the equation a
′z
2+ b
′z + c
′= 0 such that a
′> 0, (a
′, b
′, c
′) = 1 and b
′2−4a
′c
′= m
2d
K. In fact, if (a
′, N ) = 1, then K
′(T
g(α))=K
′(T
g(β)) and E(T
g(α))=E(T
g(β)).
By using the same arguments as in Theorem 6, 8 we obtain the fol- lowing two theorems.
Theorem 10. Under the same assumptions and notations as in The-
orem 4, let K
′be a CM-eld K(ζ
N+ ζ
N−1). Then the following assertions
hold.
(1) K
′(T
N∗(α)) is a class eld over K with
Gal(K
′(T
N∗(α))/K) ∼ = I
K(N )/P
K,1(N ) e P
K(N, α) and
[K
′(T
N∗(α)) : K] = h( O(N))ψ(N)
2[P
K,1(N ) e P
K(N, α) : P
K,1(N )] ,
where e P
K(N, α) is a subgroup of I
K(N ) generated by all principal ideal βO
Ksuch that β = naα + l ∈ Zaα + Z(= O), N|an and l
2− bnl ≡
±1 mod N, and ψ(1) = ψ(2) = 2 and ψ is the Euler φ-function for N ≥ 3.
(2) T
N∗(α) generates a class eld K
′(T
N∗(α)) over K
′with Gal(K
′(T
N∗(α))/K
′) ∼ = I
K′(N )/N
−1K′/K