🔢 mathematics

Kuroda's Class Number Formula

The paper presents a non-analytic proof of the class number formula specifically for biquadratic extensions of number fields.

Original authors: Franz Lemmermeyer

Published 2026-07-30

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Franz Lemmermeyer

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Great Number Detective Story

Imagine you are a detective trying to solve a mystery about how numbers behave when they travel between different worlds. In the world of mathematics, these "worlds" are called number fields. Think of a number field as a special neighborhood where you can do math with a specific set of numbers, like the whole numbers, or numbers involving square roots. Every neighborhood has its own "population count" of special patterns called ideal classes. These aren't just random groups; they are the fingerprints of the neighborhood's structure. If you know how many fingerprints a small neighborhood has, you might think you can guess how many fingerprints a bigger, more complex neighborhood will have if you build it by combining the smaller ones.

This is the heart of a famous puzzle in number theory called the class number formula. For a long time, mathematicians knew a simple rule for a specific type of neighborhood expansion: if you take a base world and build a bigger one with four distinct "sub-worlds" inside it (a structure mathematicians call a $V_4$ -extension), there is a way to calculate the total number of fingerprints in the big world using the counts from the small ones. However, there's a catch. Just like building a house, sometimes the materials you bring from the small neighborhoods don't fit perfectly, or they get squished together in unexpected ways. This "squishing" is measured by something called the unit index, which tracks how the "building blocks" (units) of the small worlds combine to form the big one.

The question that has kept mathematicians up at night is: Does the old, simple rule always work, or are there hidden traps? Specifically, does the rule break down if the big world contains very special, complex numbers like the 8th roots of unity (numbers that, when multiplied by themselves eight times, equal 1)? This paper dives deep into that question, using algebraic tools to prove exactly when the old rules hold and when they need a serious upgrade.

The Paper: Unraveling the Kuroda Formula

This paper, written by Franz Lemmermeyer, is a masterclass in detective work. It revisits a famous formula discovered by Kuroda, which acts like a recipe for calculating the "class number" (the population of ideal classes) of a large number field $K$ based on its three smaller sub-fields ( $k_1, k_2, k_3$ ). The recipe involves multiplying the class numbers of the small fields together and adjusting the result with a few correction factors, including the unit index ( $q(K)$ ), which measures how the units (the "multiplicative building blocks") of the sub-fields fit together in the big field.

The author's main goal is to generalize an algebraic proof originally created by Kubota. While previous proofs relied on heavy calculus-like methods (analytic methods), Lemmermeyer shows how to do this entirely with algebra, using the logic of class field theory. Think of this as solving a puzzle using only the shapes of the pieces rather than measuring their weight.

The Big Finding:
The paper confirms that Kuroda's formula is generally correct for these four-part extensions, but it provides a rigorous, step-by-step algebraic proof that clarifies exactly why it works. The formula essentially says:
$h(K) = \frac{2^{d-\kappa-2-v} \cdot q(K) \cdot h_1 h_2 h_3}{h_k^2}$
Here, $h(K)$ is the class number of the big field, $h_1, h_2, h_3$ are the class numbers of the three sub-fields, and $h_k$ is the base field. The terms $d$ , $\kappa$ , and $v$ are specific counters for how many places "ramify" (get messy or split) and how the units behave.

The "Gotcha" and the Correction:
The paper explicitly rules out a popular alternative formula proposed by Walter. Walter tried to create a "one-size-fits-all" version of this recipe. Lemmermeyer proves that Walter's formula is not always correct. Specifically, if the big field $K$ contains the 8th roots of unity, Walter's formula can give the wrong answer.

To prove this, the author constructs a vivid counter-example. Imagine a specific number field $K$ built from the rational numbers and some square roots.

Walter's Formula predicts: The class number should be 2.
The Reality (and Kuroda's Formula): The class number is actually 1.

Why the difference? In this specific case, the field contains the 8th roots of unity. This creates a situation where the "building blocks" (units) from the sub-fields combine in a way that Walter's formula didn't account for. It turns out that a root of unity (a special number like $i$ ) can be written as a product of units from the sub-fields, but it isn't just a simple product of roots of unity from those sub-fields. This subtle mismatch breaks Walter's logic but is perfectly handled by Kuroda's more detailed approach.

How Sure Are We?
The paper doesn't just suggest or simulate these results; it proves them. The author uses a two-part proof:

The "Capitulation" Check: First, the paper measures how many ideal classes from the small fields "capitulate" (become trivial or principal) when they move into the big field. This is done using the "ideal-theoretic" version of class field theory.
The Index Calculation: The second part is a lengthy but precise calculation of indices (ratios of group sizes). The author carefully tracks every unit, every root of unity, and every ramified prime to ensure the numbers add up exactly.

The conclusion is definitive: Kuroda's formula, with the specific corrections for the unit index and the behavior of roots of unity, is the correct tool. Walter's formula is a useful shortcut that works most of the time, but it fails in the specific, tricky case where the field contains the 8th roots of unity. The paper doesn't just say "it's wrong"; it shows exactly where the logic breaks and provides the corrected algebraic machinery to fix it.

In the end, this paper is a triumph of clarity. It takes a complex, analytic problem and solves it with pure algebra, ensuring that mathematicians have a reliable, proven method to count the hidden fingerprints of these number worlds, even when the 8th roots of unity are lurking in the shadows.

Technical Summary: Kuroda's Class Number Formula

Problem Statement
The paper addresses the computation of the class number $h(K)$ for a number field $K$ that is a $V_4$ -extension (a normal extension with Galois group isomorphic to Klein's four-group) over a base number field $k$ . While the odd part of the ideal class group behaves predictably in such extensions, the 2-part presents significant difficulties due to capitulation (ideal classes becoming principal) and the coincidence of ideal classes from subfields. The specific challenge lies in generalizing Kuroda's class number formula, which relates $h(K)$ to the class numbers of the three intermediate fields $k_1, k_2, k_3$ and the unit index $q(K) = [E_K : E_1 E_2 E_3]$ . Previous attempts, notably by Walter, provided general formulas but were shown to yield incorrect results in specific cases where $K$ contains the 8th roots of unity.

Methodology
The author, Franz Lemmermeyer, provides an algebraic proof generalizing a method originally developed by Kubota for the case where $k = \mathbb{Q}$ . The proof is divided into two distinct parts:

Ideal-Theoretic Class Field Theory: The first part utilizes the ideal-theoretic formulation of class field theory (referencing Hasse and Garbanati) to measure the extent to which the class group $\text{Cl}(K)$ is generated by the images of the class groups of the intermediate fields $\text{Cl}(k_i)$ . This involves defining a homomorphism $j: \text{Cl}(k_1) \times \text{Cl}(k_2) \times \text{Cl}(k_3) \to \text{Cl}(K)$ and analyzing its kernel and cokernel. The author establishes a correspondence between the image of a restricted homomorphism and the genus class field $K_{\text{gen}}$ of $K$ with respect to $k$ .
Index Computations: The second part involves a detailed and technical computation of the orders of the kernel and cokernel of the relevant homomorphisms. This requires:
- Analyzing the structure of ambiguous ideal classes in the intermediate fields.
- Investigating the group of units $E_K$ and its relationship with the units of the subfields and the group of roots of unity.
- Proving a specific proposition (Prop. 3) regarding norm residues in $V_4$ -extensions, linking global norms to local norm residues.
- Calculating indices involving the unit groups, specifically the index $(E : E^*)$ and the relationship between the unit index $q(K)$ and the structure of roots of unity in $K$ .

Key Contributions and Results
The primary contribution is the rigorous derivation of Kuroda's class number formula for arbitrary base fields $k$ (not just $\mathbb{Q}$ ) using purely algebraic methods, avoiding the analytic techniques used in earlier proofs.

The main result is Theorem 1, which states that for a $V_4$ -extension $K/k$ :
$h(K) = 2^{d-\kappa-2-v} q(K) \frac{h_1 h_2 h_3}{h_k^2}$
where:

$d$ is the number of infinite places of $k$ ramified in $K$ .
$\kappa$ is the $\mathbb{Z}$ -rank of the unit group $E_k$ .
$v$ is an integer determined by the structure of the units (specifically, $v=1$ if $K = k(\sqrt{\varepsilon}, \sqrt{\eta})$ with $\varepsilon, \eta \in E_k$ , and $v=0$ otherwise).
$q(K) = [E_K : E_1 E_2 E_3]$ is the unit index.

The paper explicitly derives the formula for specific cases, such as when $k=\mathbb{Q}$ and $K$ is real or complex, and when $k$ is a complex quadratic extension.

Furthermore, the paper critically examines Walter's formula. It demonstrates that Walter's generalization does not always yield correct results when $K$ contains the 8th roots of unity. The author provides a concrete counterexample using $k = \mathbb{Q}(\sqrt{3})$ and $K = \mathbb{Q}(\zeta_{24})$ , where Walter's formula predicts a class number of 2, while the actual class number is 1 (as $K$ is Euclidean). The discrepancy arises from the failure of the condition that every root of unity in $K$ expressible as a product of units from subfields must be a product of roots of unity from those subfields.

Significance
The paper claims significance primarily in providing a non-analytic, algebraic proof of Kuroda's formula for general base fields, filling a gap where such proofs were previously known only for very special cases or for $k=\mathbb{Q}$ . By generalizing Kubota's approach, the author clarifies the structural reasons behind the formula's validity and identifies the precise conditions under which alternative generalizations (like Walter's) fail. The work serves to correct the literature regarding the applicability of class number formulas in extensions containing higher roots of unity, ensuring the validity of subsequent work that relies on these formulas (such as that of Parry and Castela) by clarifying the necessary constraints.

The Great Number Detective Story

The Paper: Unraveling the Kuroda Formula

More like this