Conjecture 1.1 (, Conjecture)
T6DLFor any prime number p and any positive integer m, there is an infinite family of m + 1 successive real (or imaginary) quadratic fields,
with d\in \mathbb{Z} whose class numbers are all divisible by p.
In this paper, we construct infinitely many quadruples of real quadratic fields whose class numbers are all divisible by 3. To the best of our knowledge, this is the first result towards the divisibility of the class numbers of certain tuples of real quadratic fields. At the end, we give an application of this result to produce some elliptic curves having a 3-torsion subgroup.
For any prime number p and any positive integer m, there is an infinite family of m + 1 successive real (or imaginary) quadratic fields,
with d\in \mathbb{Z} whose class numbers are all divisible by p.
There are infinitely many quadruples of real quadratic fields of the form,
with D\in \mathbb{N} whose class numbers are all divisible by 3.
For any two integers u and v, let
If
(a) u and v are relatively prime;
(b) F_{u, v}(Z) is irreducible over \mathbb{Q};
(c) the discriminant D_{F_{u, v}} of F_{u, v}(Z) is not a perfect square in \mathbb{Z};
(d) one of the following conditions holds:
(d.1) 3\nmid v,
(d.2) 3\mid v,\hspace*{3mm} uv\not\equiv 3\pmod 9, \hspace*{3mm} u\equiv v\pm 1 \pmod 9 ,
(d.3) 3\mid v, \hspace*{3mm} uv\equiv 3 \pmod 9, \hspace*{3mm} u\equiv v\pm 1 \pmod { 27} ,
then the normal closure of \mathbb{Q}(\alpha), where \alpha is a root of F_{u,v}(Z), is a cyclic, cubic, unramified extension of \mathbb{K}=\mathbb{Q}(\sqrt{D_{F_{u,v}}}); in particular, \mathbb{K} has class number divisible by 3. Conversely, every quadratic number field \mathbb{K} with class number divisible by 3 and every unramified, cyclic and cubic extension of \mathbb{K} is given by suitable choices of integers u and v.
Let \mathbb{K}=\mathbb{Q}(\sqrt{d}). Suppose \alpha=\frac{a+b\sqrt{d}}{2} \in {\mathcal{O}}_\mathbb{K} with N_{\mathbb{K}/\mathbb{Q}}(\alpha) is a cube in \mathbb{Z}. Then P_{\alpha} (X) is reducible over \mathbb{Q} if and only if \alpha is a cube in \mathbb{K}.
Let \alpha \in R^*_D (resp. \alpha \in R^*_d). Then the splitting field, S_{\mathbb{Q}}(P_\alpha) of P_\alpha(X) over \mathbb{Q} is an S_3-field containing \mathbb{K}=\mathbb{Q}(\sqrt{d}) (resp. \mathbb{L}=\mathbb{Q}(\sqrt{D})) which is a cyclic cubic extension of \mathbb{K} (resp. \mathbb{L}) unramified outside 3 and contains a cubic subfield \mathbb{K}' with \mathit{v}_3(\Delta_{\mathbb{K}'})\ne 5. Conversely, every S_3-field containing \mathbb{K} (resp. \mathbb{L}) which is unramified outside 3 over \mathbb{K} (resp. \mathbb{L}) and contains a cubic subfield \mathbb{K}' satisfying \mathit{v}_3(\Delta_{\mathbb{K}'})\ne 5 is given by S_{\mathbb{Q}}(P_\alpha) with \alpha \in R^*_D (resp. \alpha \in R^*_d).
Suppose that
is irreducible over \mathbb{Q} and that either \mathit{v}_3(a)<2 or \mathit{v}_3(b)<3 holds. Let \theta be a root of g(X). Then 3 is totally ramified in \mathbb{Q}(\theta)/\mathbb{Q} if and only if one of the following conditions holds:
(i) 1\leq \mathit{v}_3(b)\leq \mathit{v}_3(a),
(ii) 3\mid a, \ a\not\equiv3 \pmod 9, \ 3\nmid b \ and \ b^2\not\equiv a+1 \pmod 9,
(iii) a \equiv3 \pmod 9, \ 3\nmid b \ and \ b^2\not\equiv a+1 \pmod {27}.
For any positive integer x, the class number of the real quadratic field \mathbb{Q}(\sqrt{216000x^3+457200x^2+322580x+75866}) is divisible by 3.
Let us choose u=4 and v=3(180x+127). Then \gcd(u, v)=1. We set:
Now, reading F_{u,v}(Z) under modulo 5, we get F_{u,v}(Z)\equiv Z^3-4Z-1\pmod 5. It is easy to check that F_{u,v}(Z)\pmod 5 is irreducible, and thus F_{u,v}(Z) is irreducible as a polynomial with integer coefficients as well.
The discriminant of F_{u,v}(Z) is
This can be simplified as \Delta_{F_{u, v}}=12^4D, where D=216000x^3+457200x^2+322580x+75866. Since D\equiv 2\pmod 4, D is not a square in \mathbb{Z} and so is \Delta_{F_{u,v}}.
We see that 3\mid v, and uv\equiv 3\pmod 9. Further, v+1\equiv 4\pmod {27}. Therefore, F_{u,v}(Z) satisfies the conditions (a)-(c) and (d.3) of Theorem A. This completes the proof by Theorem A.
For any positive integer y, the class number of the real quadratic field \mathbb{Q}(\sqrt{432y^3+1080y^2+900y+223}) is divisible by 3.
The proof of this proposition is very similar to that of Proposition 3.1. Nevertheless, we give the proof in brief for the sake of completeness.
We put u=2 and v=6y+5. Then \gcd(u, v)=1, and we define,
The discriminant of F_{u,v}(Z) is
which can be simplified as 4^2d with d=432y^3+1080y^2+900y+223. As d\equiv 2\pmod 3, d is not a square in \mathbb{Z} and so is \Delta_{F_{u, v}}. It is easy to see that F_{u,v}(Z)\pmod 3 is irreducible, and thus F_{u,v}(Z) is irreducible over \mathbb{Z} too. Since 3\nmid v, therefore by Theorem A, we conclude the proof.
Let \mathbb{K} be a number field with its group of fractional ideals and class group respectively, \mathcal{I}_\mathbb{K} and Cl(\mathbb{K}). If H(\mathbb{K}) is the Hilbert class field of \mathbb{K}, then the Artin map
is surjective, and its kernel is exactly the subgroup P_\mathbb{K} of principal fractional ideals. Thus this map induces the following deep correspondence
For any positive integer k, the class number of the real quadratic field \mathbb{Q}(\sqrt{40500k^3+89100k^2+65340k+16215}) is divisible by 3.
Assume that d=-(13500k^3+29700k^2+21780k+5405). Then d\not \equiv 0\pmod 3, and thus we set D:=-3d=40500k^3+89100k^2+65340k+16215. As d\equiv 2\pmod 3 and D\equiv 3\pmod 4, so that both d and D are not perfect squares.
Let \mathbb{K}=\mathbb{Q}(\sqrt{-d}), and define \alpha \in \mathbb{K} as
Then T_{\mathbb{K}/\mathbb{Q}}(\alpha)=9 and N_{\mathbb{K}/\mathbb{Q}}(\alpha)=-(15k+11)^3, and thus \gcd(T_{\mathbb{K}/\mathbb{Q}}(\alpha), N_{\mathbb{K}/\mathbb{Q}}(\alpha))=1. Therefore with this \alpha, we define
which is
If Z=p/q is a root of f_\alpha(Z) in \mathbb{Q}, then p\mid 9 and q\mid 1, which further imply that
However, f_\alpha(\pm1)\ne 0, f_\alpha(\pm3)\ne 0 and f_\alpha(\pm9)\ne 0. This confirms that f_\alpha(Z) is irreducible over \mathbb{Q}. Therefore by Lemma 2.1, \alpha is not a cube in \mathbb{K}, and hence \alpha \in R_d. As \gcd(T_{\mathbb{K}/\mathbb{Q}}(\alpha), N_{\mathbb{K}/\mathbb{Q}}(\alpha))=1, so that \alpha\in R^*_d. Therefore by Theorem B, S_\mathbb{Q}(f_\alpha) is a cyclic cubic extension of \mathbb{L} which is unramified outside 3.
It remains to check the unramification of S_\mathbb{Q}(f_\alpha) at 3. To see this, we will apply Lemma 2.2. By the assumptions, we see that v_3(3(15k+11))=1 and v_3(b)=v_3(9)=2. Thus (i) of Lemma 2.2 does not hold. Rest of two conditions of Lemma 2.2 do not hold since b=9\equiv 0\pmod 3. Thus by Lemma 2.2, we can conclude that S_\mathbb{Q}(f_\alpha) is unramified over \mathbb{L} at 3 as well. Therefore, we complete the proof by Theorem C.
Assume that S is a finite set such that \{\infty\}\subset S\subset V_\mathbb{Q}. Let f(x)\in \mathbb{Q}[x] be a polynomial of degree at least 3 with distinct roots in \mathbb{C}. Then the set
is finite, where R_S=\{x\in \mathbb{Q}\mid v_p(x)\geq 0 \text{ for all } p\in V_\mathbb{Q}\setminus S\}; so-called the ring of S-integers of \mathbb{Q}.
The set
Assume that E_1, E_2 and E_3 are as defined above. Then there is a 3-torsion in the {\mathbb Q}-rational points on each of E_1, E_2 and E_3.
The text on this page is from arXiv:2512.11346v1 by Kalyan Banerjee, Ankurjyoti Chutia and Azizul Hoque, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.
The record of each item is in analysis/objects/2512/2512.11346.jsonl.