Abstract
In this paper, we apply Qin’s theorem for the 4-rank of \(K_2O_F\) to establish the relation between the 4-rank of the ideal class group of \(F=\mathbb {Q}(\sqrt{d})\) and the 4-rank of \(K_2O_F\) provided that all odd prime factors of d are congruent to 1 mod 8. As an application, we give a concise and unified proof of two conjectures proposed by Conner and Hurrelbrink (Acta Arith 73:59–65, 1995).
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let F be a number field, \(O_F\) the ring of integers in F. Let \(K_2O_F\) denote the Milnor group of \(O_F\), which coincides with the tame kernel of F. As a finite abelian group, \(K_2O_F\) contains rich arithmetical information. In particular, the Birch–Tate conjecture and the Lichtenbaum conjecture establish a relation between the order of \(K_2O_F\) and the value \(\zeta _F^*(-1)\), where \(\zeta _F(\cdot )\) is the Dedekind zeta-function of F. To understand \(K_2O_F\), we need to know the \(p^n\)-rank of \(K_2O_F\) for any prime p and any positive integer n. If F contains a primitive \(p^n\)-th root of unity, the \(p^n\)-rank of \(K_2O_F\) formula is given by Tate [13], in particular, the 2-rank of \(K_2O_F\) formula is given for any number field. See also Keune [6].
When \(F=\mathbb {Q}(\sqrt{d})\) is a quadratic field, Browkin and Schinzel [3] give an explicit formula for the 2-rank of \(K_2O_F\). Qin [9,10,11] obtains formulas for the 4-rank of \(K_2O_F\) and the 8-rank of \(K_2O_F\). See also [8, 12]. In this paper, we apply Qin’s theorem for the 4-rank of \(K_2O_F\) to establish a relation between the 4-rank of ideal class group and the 4-rank of \(K_2O_F\) provided that all odd prime factors of d are congruent to 1 mod 8. As an application, we give a concise and unified proof of two conjectures proposed by Conner and Hurrelbrink in [4]. Both conjectures are proved by Vazzana [14, 15], but the proofs, which use graph theory, are somewhat involved.
2 Statement of main theorems
We first state main theorems of this paper. For any abelian group A, let \(r_4(A)\) denote the 4-rank of A.
Theorem 2.1
Let \(E=\mathbb {Q}(\sqrt{d})\), where \(d = p_1\ldots p_k\) with \(p_i\equiv 1\pmod {8}\) and let C(E) be the class group of E. Assume that
- (i):
-
the norm of the fundamental unit of E is \(-1,\) and
- (ii):
-
an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 32y^2.\) Then
$$\begin{aligned} r_4(K_2O_E)=r_4(C(E)). \end{aligned}$$
Theorem 2.2
Let \(F=\mathbb {Q}(\sqrt{2d})\) and \(E=\mathbb {Q}(\sqrt{d})\), where \(d = p_1\ldots p_k\) with \(p_i\equiv 1\pmod {8}\) and let C(E) be the class group of E. Assume that
- (i):
-
the norm of the fundamental unit of E is \(-1,\) and
- (ii):
-
an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 64y^2.\) Then
$$\begin{aligned} r_4(K_2O_F)=r_4(C(E)). \end{aligned}$$
Theorem 2.3
Let \(F=\mathbb {Q}(\sqrt{2d})\), where \(d = p_1\ldots p_k\) with \(p_i\equiv 1\pmod {8}\) and let C(F) be the class group of F. Assume that
- (i):
-
the norm of the fundamental unit of F is \(-1,\) and
- (ii):
-
an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 64y^2.\) Then
$$\begin{aligned} r_4(K_2O_F)=r_4(C(F))-1. \end{aligned}$$
As corollaries of Theorems 2.1 and 2.2, we have the following two theorems.
Theorem 2.4
The 2-primary part of \(K_2(\mathcal {O}_E)\) is elementary abelian if and only if
- (i):
-
the 2-primary part of the ideal class group C(E) is elementary abelian and the norm of the fundamental unit of E is \(-1,\) and
- (ii):
-
an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 32y^2.\)
Theorem 2.4 gives an explicit set of conditions under which the 2-primary part of \(K_2(\mathcal {O}_E)\) is elementary abelian. A similar theorem for the field \(F= \mathbb {Q}(\sqrt{2d})\) is as follows.
Theorem 2.5
The 2-primary part of \(K_2(\mathcal {O}_F)\) is elementary abelian if and only if
- (i):
-
the 2-primary part of the ideal class group C(E) is elementary abelian and the norm of the fundamental unit of E is \(-1,\) and
- (ii):
-
an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 64y^2.\)
Both of these two theorems were conjectured by Conner and Hurrelbrink in [4]. Vazzana proved the two theorems in [14] and [15], respectively. In particular, his proof for Theorem 2.5 is complicated. He made use of a graph associated with the primes \(p_1,\ldots , p_k\) and studied its relationship to a new graph associated with the primes lying over \(p_1,\ldots , p_k\) in \(\mathbb {Q}(\sqrt{-2}).\) Our alternative proof is not only unified but also very concise.
3 Prelimilary results
In this section, we review some known results which are useful in this paper. The following result is classical. One can find a proof in [5].
Lemma 3.1
(Legendre’s Theorem) Suppose that a, b, c are square-free,
and a, b, c do not have the same sign. Then the Diophantine equation
has nontrivial solutions if and only if for every odd prime p|abc, say, p|a, \(\left( \frac{-bc}{p}\right) =1.\)
We adopt the notation from [11]. Let \(d\ne 0\) be an integer. Put
For any abelian group A, set
For a number field F, we use NF for the set of norms from F over \(\mathbb {Q}.\)
Let F be a number field with \(r_2\) complex places. By Tate [13], every element of order two in \(K_2F\) is of the form \(\{-1,\ a\}\), where \(a\in F^{*}\). Let \(\Delta \) denote the Tate kernel of F, i.e., the group of elements \(x\in F^{*}\) such that \(\{-1,\ x\}=1\). If \(\sqrt{2}\not \in F\), then by [13],
In particular, if \(F\not =\mathbb {Q}(\sqrt{2})\) is a real quadratic field, then \(\Delta =F^*\cup 2F^*.\)
Theorem 3.2
(Browkin and Schinzel) Let \(F=\mathbb {Q}(\sqrt{d}), d\in \mathbb {Z}\) square-free. Then \(_{2}K_2(O_F)\) can be generated by
together with
if \(\{-1, \pm 2\}\cap NF\ne \emptyset ,\) where \(u_i\in \mathbb {Z}\) is such that \(u_i^2-d=c_iw_i^2\) for some \(w_i\in \mathbb {Z}\) and \(c_i\in \{-1, \pm 2\}\cap NF.\)
Remark
The general \(r_2(K_2O_F)\) formula for an arbitrary number field F is given by Tate [13].
For any integer n, put \(\nabla ^n=\{\alpha \in K_2O_F| \alpha =\beta ^n \;\mathrm{for} \;\mathrm{some} \; \beta \in K_2O_F\}.\)
Our main tool is the following theorem of Qin, which completely determines the 4-rank of \(K_2O_F\) for any quadratic field (see [9, 10]).
Theorem 3.3
(Qin) Let \(F=\mathbb {Q}(\sqrt{d}), d\in \mathbb {Z}\) square-free. Suppose that m|d \((m>0 \;{if} \;d>0,\) ,but m also takes on negative values if \(d< 0)\) and write \(d=u^2-2w^2\) with \(u, w\in \mathbb {Z}\) if \(2\in NF.\) Then \(\{-1,m\}\in \nabla ^2\) if and only if one can find an \(\epsilon \in S(d)\) such that
- (i):
-
\(\left( \frac{\frac{d}{m}}{p}\right) =\left( \frac{\epsilon }{p}\right) \) for every odd prime p|m;
- (ii):
-
\(\left( \frac{m}{p}\right) =\left( \frac{\epsilon }{p}\right) \) for every odd prime \(p|\frac{d}{m};\)and \(\{-1,m(u+\sqrt{d})\}\in \nabla ^2\) if and only if one can find an \(\epsilon \in S(d)\) such that
- (iii):
-
\(\left( \frac{\frac{d}{m}}{p}\right) =\left( \frac{\epsilon (u+w)}{p}\right) \) for every odd prime p|m;
- (iv):
-
\(\left( \frac{m}{p}\right) =\left( \frac{\epsilon (u+w)}{p}\right) \) for every odd prime \(p|\frac{d}{m}.\)
Corollary 3.4
Let \(d = p_1\ldots p_k\) be a product of different primes congruent to 1 mod 8, and let \(F=\mathbb {Q}(\sqrt{d})\) or \(F=\mathbb {Q}(\sqrt{2d})\). If \(m\mid d\), then \(\{-1,m\}\in \nabla ^2\) if and only if
is solvable in \(\mathbb {Z}\); \(\{-1,m(u+\sqrt{d})\}\in \nabla ^2\) if and only if
is solvable in \(\mathbb {Z}.\)
Proof
Since for every odd prime p dividing d, \(\left( \frac{\pm 1}{p}\right) =\left( \frac{\pm 2}{p}\right) =1,\) the result follows from Theorem 3.3 and Lemma 3.1. \(\square \)
4 Proofs of main results
Lemma 4.1
Let \(d = p_1\ldots p_k\) be a product of rational primes congruent to 1 mod 8. Assume that \(d=u^2-2w^2, v=u+w(u>0).\) Then \(v\equiv 3\pmod {4}\) if and only if an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 32y^2.\)
Proof
First we assume that \(k=1.\) There exist two integers u, w such that \(p=u^2-2w^2\). Assume that \(u>0\) and write \(v=u+w\). Since \(u^2-2w^2=(3u+4w)^2-2(2u+3w)^2\) and \(p\equiv 1\pmod {8}\), by a suitable choice of u, w, we can write \(p=u^2-2(4w^\prime )^2\). Hence \(v=u+4w^\prime \equiv u\pmod {4}\). By [1], we have that \(v\equiv 1\pmod {4}\) if and only if \(p=x^2 + 32y^2\), equivalently, \(v\equiv 3\pmod {4}\) if and only if \(p\not =x^2 + 32y^2.\)
We consider now the general case. Assume that \(d_1=u_1^2-2w_1^2,\) \(d_2=u_2^2-2w_2^2\) and \(d_1d_2=u^2-2w^2\). Then \(d_1d_2=(u_1u_2+2w_1w_2)^2-2(u_1w_2+u_2w_1)^2\). If we assume further that \(d_1\equiv d_2\equiv 1\pmod {4}\), then both \(w_1\) and \(w_2\) are even. We have
i.e., \(v=u+w\equiv v_1v_2\pmod {4},\) where \(v_1=u_1+w_1, v_2=u_2+w_2.\) Now suppose that \( p_1\ldots p_k=u^2-2w^2, v=u+w(u>0).\) For any \(1\le i\le k\), assume that \(p_i=u_i^2-2w_i^2~(u_i>0)\) and write \(v_i=u_i+w_i.\) By induction,
Note that \(v_i\equiv 1\) or \(3\pmod {4}\) and \(v_i\equiv 1\pmod {4}\) if and only if \(p_i=x^2 + 32y^2.\) Therefore, \(v\equiv 3\pmod {4}\) if and only if an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 32y^2.\) \(\square \)
Lemma 4.2
Let \(p_1,\ldots , p_k\) be rational primes congruent to 1 mod 8. Let \(d = p_1\ldots p_k\) or \(2p_1\ldots p_k\), let \(E=\mathbb {Q}(\sqrt{d}),\) and let C(E) be the class group of E. Write \(\varepsilon \) for the fundamental unit of \(E=\mathbb {Q}(\sqrt{d})\). Then
- (i):
-
\(N\varepsilon =1\) if and only if the following Diophantine equation is solvable for some \(m\mid d\), \(m\ne \pm 1\) or \(\pm d\),
$$\begin{aligned} mX^2-\frac{d}{m}Y^2=\pm 1. \end{aligned}$$(3) - (ii):
-
if \(N\varepsilon =-1\) then \(_2C(E)\) can be generated by the ideals \(\wp _1,\ldots ,\wp _k\), where for any i, \(\wp _i^2=p_i\).
Proof
-
(i)
We apply the result on the Tate kernel of real quadratic fields to show that the norm of the fundamental unit of \(E=\mathbb {Q}(\sqrt{d})\) is 1 if and only if
$$\begin{aligned} mX^2-\frac{d}{m}Y^2=\pm 1 \end{aligned}$$is solvable for some \(m\mid d,\) \(m\ne \pm 1,\pm d.\) In fact, since \(\{-1, \varepsilon \}\in \ _2K_2(O_E),\) by Theorem 3.2, there exists some \(m\mid d,\) \(m\ne \pm 1,\pm d\) such that \(\{-1, \varepsilon \}=\{-1,m\}\) or \(\{-1, \varepsilon \}=\{-1,m(u+\sqrt{d})\}.\) It is straightforward to see that
$$\begin{aligned} \varepsilon m(u+\sqrt{d})\not \in E^{*2}\cup 2E^{*2}. \end{aligned}$$Clearly, we have \(\varepsilon m\not \in 2E^{*2}\). Hence, \(\varepsilon m\in E^{*2}\). Therefore, \(m^2=\varepsilon m{{\bar{\varepsilon }} m}=(X^2-dY^2)^2\) for some integers X, Y, so we obtain \(\pm m=X^2-dY^2\), i.e., (3) holds.
Conversely, if \(\pm m=X^2-dY^2\), \(m\ne \pm 1,\pm d,\) we put
$$\begin{aligned} \varepsilon =\frac{X+\sqrt{d}Y}{X-\sqrt{d}Y}. \end{aligned}$$It is obvious that \((m,Y)=1.\) Hence, for any finite prime P of E, if \(P\mid m\), then \(v_P(X+\sqrt{d}Y)= v_P(X-\sqrt{d}Y)=1\); if \(P\not \mid m\), then \(v_P(X+\sqrt{d}Y)= v_P(X-\sqrt{d}Y)=0\). Therefore, \(\varepsilon \) is an algebraic integer. Moreover, \(\varepsilon \) is not a square, since \(\varepsilon (X-\sqrt{d}Y)^2=X^2-dY^2=\pm m,\) \(m\ne \pm 1,\pm d,\) is not a square.
-
(ii)
By genus theory, the 2-rank of C(E) is \(k-1\) if \(E=\mathbb {Q}(\sqrt{p_1\ldots p_k}),\) and k if \(E=\mathbb {Q}(\sqrt{2p_1\ldots p_k})\). For any \(m\mid d\), the ideal \((m,\sqrt{d})\) satisfies that \((m,\sqrt{d})^2=(m)\). Moreover, \((m,\sqrt{d})\) is principal in C(E) if and only if
$$\begin{aligned} \pm m=X^2-dY^2 \end{aligned}$$(4)is solvable over \(\mathbb {Z}\).
But the assumption that \(N\varepsilon =-1\) implies that for any \(m\mid d\), \(m\ne \pm 1,\pm d,\) (4) has no integer solution. Therefore, the ideals \(\wp _1,\ldots ,\wp _k\) generate \(_2C(E).\)
\(\square \)
Proof of Theorem 2.1
We apply Theorem 3.3 to compute \(r_4(K_2O_E)\).
We first show that the assumption \(\mathrm{(ii)}\) implies \(\{-1,m(u+\sqrt{d})\}\not \in \nabla ^2.\)
Recall that in our case, all prime factors are congruent to 1 mod 8. By Corollary 3.4 to theorem of Qin, the condition that \(\{-1,m(u+\sqrt{d}\}\in \nabla ^2\) is equivalent to the fact that there exists a positive factor m of d, so that
is solvable in \(\mathbb {Z}\). Since \(d\equiv 1\pmod {8}, m\mid d,\) \(X^2+dY^2\equiv 0,1,2\pmod {4}.\) Now the assumption \(\mathrm{(ii)}\) and Lemma 4.1 imply that \(v\equiv 3\pmod {4}.\) Hence, \(vmZ^2\equiv 0,3\pmod {4}\) and so (5) has no solution with \((X,Y,Z)=1\).
Therefore, in \(_2K_2O_E\) only \(\{-1,m\}\) could belong to \(\nabla ^2.\) By sending \(\{-1,m\}\) to the ideal \(\mathfrak {M} (\mathfrak {M}^2=m),\) we obtain a bijection from the set \(\{\{-1,m\},m\mid d\} \subseteq \) \( _2K_2O_E\) to \(_2C(E).\)
By [7], we see that \(\mathfrak {M}\) is a square in C(E) if and only if
is solvable in \(\mathbb {Z}\). Furthermore, \(\{-1,m\}\) is a square in \(K_2O_E\), i.e., \(\{-1,m\}\in \nabla ^2\) if and only if
is solvable in \(\mathbb {Z}.\) As (6) and (7) have the same solvability,
holds. \(\square \)
To prove Theorem 2.2, we need the following Lemma 4.5, which is an analog of Lemma 4.1. First we have
Lemma 4.3
Let \(\alpha =2^{\frac{1}{4}}\) and \(L=\mathbb {Q}(\alpha )\). Then \(\mathbb {Z}[\alpha ]\) is the ring of algebraic integers of L. The class number of L is 1.
Proof
This can be done by the GP calculator. \(\square \)
Lemma 4.4
Let \(p\equiv 1\pmod {8}\) be a prime. Assume that \(p=u^2-2w^2\) with \(u>0.\) Then \(p=x^2+64y^2\) if and only if \(u\equiv 1,3\pmod {8}\).
Proof
We need the following result: Let \(p\equiv 1\pmod {8}\) be a prime. Then \(\gamma ^4 \equiv 2\pmod {p}\) is solvable if and only if \(p=x^2+64y^2\). See, for example, Theorem 7.5.2 in [2]. Hence, for a prime \(p=x^2+64y^2\), \(\gamma ^4 \equiv 2\pmod {p}\) has four distinct solutions. It follows that p splits completely in L. Hence, there are integers a, b, c, d such that
Since \(p\equiv 1\pmod {8}\), we see that a is odd and \(2\mid b\). Hence we have \(u=a^2+2c^2-4bd\equiv 1,3\pmod {8}.\)
Conversely, if \(p=u^2-2w^2\) with \(|u|\equiv 1,3\pmod {8}\) and w even, then \(p=x^2+64y^2.\) In fact, we may write \(u=t^2u^\prime , w=2^es^2w^\prime \), where \(u^\prime ,w^\prime \) are square-free odd and positive integers. For any prime \(l\mid w^\prime \), \((\frac{p}{l})=1\), hence, \(w\equiv \alpha ^2\pmod {p}\) since 2 is a square \(\pmod {p}\). On the other hand, for any prime \(l\mid u^\prime \), \((\frac{-2p}{l})=1\), hence, \(|u|\equiv 1,3\pmod {8}\) implies that \(u \equiv \alpha ^2\pmod {p}.\) Therefore, \(\gamma ^4 \equiv 2\pmod {p}\) is solvable and so \(p=x^2+64y^2\). \(\square \)
Lemma 4.5
Let \(d = 2p_1\ldots p_k\) with \(p_1\equiv \cdots \equiv p_k \equiv 1\pmod {8}.\) Assume that \(p_i=u_i^2-2w_i^2, v_i=u_i+w_i(u_i>0).\) If we write \(p_1\ldots p_k=u^2-2w^2,\) and \(2p_1\ldots p_k=U^2-2W^2, U>0,\) then \(u\equiv u_1\ldots u_k\pmod {8}\) and \(U+W\equiv u\) or \(3u\pmod {8}\). Moreover, \(U+W\equiv 5,\) or \(7\pmod {8}\) if and only if an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented over \(\mathbb {Z}\) by the quadratic form \(x^2 + 64y^2.\)
Proof
Assume that \(a=u^2-2w^2\) and \(b=\mu ^2-2\omega ^2\). Then we have
Note that if both \(a,b\equiv 1\pmod {8}\), it is easy to see w and \(\omega \) are even, so \(u\mu +2w\omega \equiv u\mu \pmod {8}\). By induction, we have \(u\equiv u_1\ldots u_k\pmod {8}.\) Observe that
Since \((2u+2w)+(-u-2w)=u\) and \((2u+2w)+(u+2w)=3u+4w\equiv 3u\pmod {8},\) \(U+W\equiv u\) or \(3u\pmod {8}.\)
If \(k=1\) and \(d=2 p_1=U^2-2W^2,\) by Lemma 4.4, \(p_1\ne x^2 + 64y^2\) if and only if \(U+W\equiv 5,7\pmod {8}.\) Observe that if \(u_1\equiv 1,3\pmod {8},u_2\equiv 1,3\pmod {8}\) or \(u_1\equiv 5,7\pmod {8},u_2\equiv 5,7\pmod {8},\) then \(u_1u_2\equiv 1,3\pmod {8}\), and if \(u_1\equiv 1,3\pmod {8},u_2\equiv 5,7\pmod {8},\) then \(u_1u_2\equiv 5,7\pmod {8}.\) Therefore, by (8) and (9), if an odd number of the primes \(p_1,\ldots , p_k\) fail to be represented by \(x^2 + 64y^2,\) then \(U+W\equiv 5,\) or \(7\pmod {8},\) and vice versa. \(\square \)
Proof of Theorem 2.2
As in the proof of Theorem 2.1, we apply again Theorem 3.3 to compute \(r_4(K_2O_F)\).
Write \(2d=u^2-2w^2, ~u>0,~v=u+w\). We claim that for any \(m\mid 2d\), \(\{-1,m(u+\sqrt{2d})\}\not \in \nabla ^2.\)
Indeed, by Theorem 3.3 and Lemma 3.1, if for some positive factor m of 2d, \(\{-1,m(u+\sqrt{2d})\}\in \nabla ^2\), then
is solvable in \(\mathbb {Z}\). By Lemma 3.1, we can assume that m is odd. Since \(2d\equiv 2\pmod {8},\) one has \(X^2+2dY^2\equiv 1,2,3,6\pmod {8}\) if \((X,Y)=1\). It follows from Lemma 4.5 that the assumption \(\mathrm{(ii)}\) implies that \(v\equiv 5,7\pmod {4}.\) Hence, \(vmZ^2\equiv 0,4,5,7\pmod {8}\) and so (10) has no solution with \((X,Y,Z)=1\).
Note that \(\{-1,2m\}=\{-1,m\}\). Therefore, the same discussion as in the proof of Theorem 2.1 shows that
\(\square \)
Proof of Theorem 2.3
The assumption that the norm of the fundamental unit of F is \(-1\) implies that there is an ideal \(\wp \) of F, which is non-principal with \(\wp ^2=(2)\). But in \(K_2O_F\), \(\{-1,2\}=1\). Then the proof is analogous to that of Theorem 2.2. \(\square \)
We turn to prove Theorems 2.4 and 2.5. Let \(m\mid d\). We introduce the following notation.
where
It is easy to check that \(\sqcap _{i=1}^{k}\epsilon _i=1\). Similarly, we put
where
By Theorem 3.3, \(\{-1,m\}\in \nabla ^2\) if and only if \(V(m,p_1,\ldots ,p_k)=(1,\ldots ,1)\). If \(m,n\mid d\) and \((m,n)=t\), we write \(V(mn,p_1,\ldots ,p_k)=V(mn/t^2,p_1,\ldots ,p_k)\). Moreover, if \(V(m,p_1,\ldots ,p_k)=(\epsilon _1,\ldots ,\epsilon _k)\) and \(V(n,p_1,\ldots ,p_k)=(\delta _1,\ldots ,\delta _k),\) then \(V(mn,p_1,\ldots ,p_k)=(\epsilon _1\delta _1,\ldots ,\epsilon _k\delta _k).\) Similarly, \(V(m(u+\sqrt{d}),p_1,\ldots ,p_k)=(\epsilon _1,\ldots ,\epsilon _k),\) and \(V(n,p_1,\ldots ,p_k)=(\delta _1,\ldots ,\delta _k),\) then \(V(mn(u+\sqrt{d}),p_1,\ldots ,p_k)=(\epsilon _1\delta _1,\ldots ,\epsilon _k\delta _k).\)
Proof of Theorem 2.4
If both (i) and (ii) are satisfied, then by Theorem 2.1, we have \(r_4(K_2(\mathcal {O}_E))=0\), i.e., the 2-primary part of \(K_2(\mathcal {O}_E)\) is elementary abelian.
Conversely, if the 2-primary part of \(K_2(\mathcal {O}_E)\) is elementary abelian, then by Lemma 4.2, Lemma 3.1 and Theorem 3.3, we see that the norm of the fundamental unit of E is \(-1\). We show that \(\{-1,m(u+\sqrt{d})\}\not \in \nabla ^2\) implies condition \(\mathrm{(ii)}\), which in turn shows that \(v\equiv 3\pmod {4}\) as we can see from Lemma 4.1.
Since \(r_4(K_2O_E)=0\), for any \(m\mid d\),
and equivalently,
has no integer solutions. Hence, we have proved that \(r_4(C(E))=0\).
Since \(r_4(K_2O_E)=0\), for any \(m\mid d\), \(m\ne \pm 1,\pm d\), \(V(m,p_1,\ldots ,p_k)\not =(1,\ldots ,1)\). By a suitable choice of integers \(m_i\) dividing d, we may assume that \(\{-1,-1\},\{-1,m_1\},\) \(\ldots ,\{-1,m_{k-1}\},\{-1,u+\sqrt{d}\},\{-1,u_{-1}+\sqrt{d}\}\) generate \(_2K_2O_E\), where \(u_{-1}^2-d=-w_{-1}^2\), and the \(m_j's\) are chosen in such a way that, for \(1\le i\le k-1\), \(\epsilon _k=\epsilon _i=-1\) and for \(j\not =i,k,\epsilon _j=1\).
Suppose that \(v\equiv 1\pmod {4}\). We have \((u+w)(u-w)-w^2=d\) since \(u^2-2w^2=d\), hence \(\left( \frac{-d}{v}\right) =1\) and so \(\left( \frac{v}{d}\right) =1.\) On the other hand, if \(m\mid d\) with \(m\equiv 1\pmod {8},\) then \(\left( \frac{\frac{d}{m}}{m}\right) \left( \frac{m}{\frac{d}{m}}\right) =1.\) For any odd and positive \(m\mid d\), assuming that \(V(m(u+\sqrt{d}),p_1,\ldots ,p_k)=(\epsilon _1,\ldots ,\epsilon _k),\) we have \(\sqcap _{i=1}^{k}\epsilon _i=\left( \frac{\frac{vd}{m}}{m}\right) \left( \frac{vm}{\frac{d}{m}}\right) =\left( \frac{v}{d}\right) \left( \frac{\frac{d}{m}}{m}\right) \left( \frac{m}{\frac{d}{m}}\right) =\left( \frac{v}{d}\right) =1\). Therefore, there exists \(n\mid d\) such that \(V(n(u+\sqrt{d}),p_1,\ldots ,p_k)=(1,\ldots ,1).\) This contradicts the assumption that \(r_4(K_2O_E)=0\). Hence, \(v\equiv 3\pmod {4}\), and so we have (ii). \(\square \)
Proof of Theorem 2.5
The proof is analogous to that of Theorem 2.4.
Applying Theorem 2.2, we see that (i) and (ii) imply that \(r_4(K_2(\mathcal {O}_F))=0\).
Conversely, the assumption that \(r_4(K_2(\mathcal {O}_F))=0\) implies that \(r_4(C(E))=0\) and the norm of the fundamental unit of E is \(-1\), i.e., (i) holds. To prove (ii), by Lemma 4.5, it is sufficient for us to show that \(V=U+W\equiv 5,~7\pmod {8}.\)
Since \(U^2-2W^2=2d\), \(\left( \frac{-2d}{V}\right) =1\). Hence, \(\left( \frac{V}{d}\right) =1\) if and only if \(V\equiv 1,~3\pmod {8}.\) If \(V\equiv 1,~3\pmod {8},\) then for any odd and positive \(m\mid d\), assuming that \(V(m(U+\sqrt{d}),p_1,\ldots ,p_k)=(\epsilon _1,\ldots ,\epsilon _k),\) we have \(\sqcap _{i=1}^{k}\epsilon _i=\left( \frac{V}{d}\right) =1\). Therefore, one can find \(n\mid d\) such that \(V(n(U+\sqrt{d}),p_1,\ldots ,p_k)=(1,\ldots ,1),\) i.e., \(r_4(K_2(\mathcal {O}_F))>0\), a contradiction!
This completes the proof. \(\square \)
References
Barrucand, P., Cohn, H.: Note on primes of type \(x^2+ 32y^2\), class number, and residuacity. J. Reine Angew. Math. 238, 67–70 (1969)
Berndt, B.C., Evans, R.J., Williams, K.S.: Gauss and Jacobi Sums. Wiley, New York (1998)
Browkin, J., Schinzel, A.: On Sylow \(2\)-subgroups of \(K_2\cal{O}_F\) for quadratic number fields \(F\). J. Reine Angew. Math. 331, 104–113 (1982)
Conner, P.E., Hurrelbrink, J.: On elementary abelian \(2\)-Sylow \(K_2\) of rings of integers of certain quadratic number fields. Acta Arith. 73, 59–65 (1995)
Ireland, K., Rosen, M.: A Classical Introduction To Modern number Theory. Graduate Texts in Mathematics, vol. 84. Springer, New York (1972)
Keune, F.: On the structure of \(K_2\) of the ring of integers in a number field. K-theory 2, 625–645 (1989)
Kolster, M.: The 2-part of the narrow class group of a quadratic number field. Ann. Sci. Math. Que. 29, 73–96 (2005)
Qin, H.R.: 2-Sylow subgroup of \(K_2 \cal{O}_F\) for real quadratic fields \(F\). Sci. China 37, 1302–1313 (1994)
Qin, H.R.: The \(2\)-Sylow subgroups of the tame kernel of imaginary quadratic fields. Acta Arith. 69, 153–169 (1995)
Qin, H.R.: The \(4\)-rank of \(K_2(\cal{O})\) for real quadratic fields \(F\). Acta Arith. 72, 323–333 (1995)
Qin, H.R.: Tame kernels and Tate kernels of quadratic number fields. J. Reine Angew. Math. 530, 105–144 (2001)
Qin, H.R.: The structure of the tame kernels of quadratic number fields(I). Acta Arith. 113, 203–240 (2004)
Tate, J.: Relations between \(K_2\) and Galois Cohomology. Invent. Math. 36, 257–274 (1976)
Vazzana, A.: On the \(2\)-primary part of \(K_2\) of rings of integers in certain quadratic number fields. Acta Arith. 80, 225–235 (1997)
Vazzana, A.: Elementary abelian \(2\)-primary parts of \(K_2\cal{O}\) and related graphs in certain quadratic number fields. Acta Arith. 81, 253–264 (1997)
Acknowledgements
The author would like to thank the referee for insightful and useful suggestions that improved the presentation of this paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by the National Natural Science Foundation of China (Grant No. 11971102).
Rights and permissions
About this article
Cite this article
Wu, X. The 2-Sylow subgroup of \(K_2O_F\) for certain quadratic number fields. Ramanujan J 54, 463–473 (2021). https://doi.org/10.1007/s11139-020-00251-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11139-020-00251-4