🔢 mathematics

Explicit Jordan decompositions for ideal lattices in CM fields

This paper provides explicit formulas for the Jordan decomposition of ideal lattices over CM fields at prime ideals, expressed in terms of the prime ideal factorization of the underlying ideal, by reducing the computation to local behavior following the approach of Erez, Morales, and Perlis.

Original authors: Guilhem Mureau

Published 2026-08-05

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Guilhem Mureau

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

Imagine you are a master architect trying to build a fortress, but instead of bricks and mortar, your building blocks are numbers. In the world of mathematics, specifically a field called number theory, these numbers live in special "neighborhoods" known as number fields. Sometimes, these neighborhoods have a very specific, symmetrical structure called a CM field, which is like a perfectly mirrored city where every street has a twin on the other side.

To understand how these number cities are built, mathematicians use tools called lattices. Think of a lattice as a grid of points, like the dots on a piece of graph paper, but stretched and twisted into complex shapes in higher dimensions. These grids aren't just for drawing; they are the secret sauce behind modern cryptography, the digital locks that keep your bank accounts and private messages safe. To crack a code or prove a system is secure, you need to know exactly how these grids are shaped. One of the most powerful ways to describe a shape is to break it down into its simplest, most fundamental building blocks. In the world of lattices, this process is called a "Jordan decomposition." It's like taking a complex Lego castle apart to see exactly which sizes of bricks were used and how they were stacked.

For a long time, mathematicians knew how to take apart these grids if the city they lived in was simple. But when the city had that special "CM" symmetry, the instructions were missing. This is the puzzle Guilhem Mureau tackles in his paper. He provides a new, explicit recipe for taking apart these complex, mirrored lattices. By doing so, he gives cryptographers and number theorists a precise way to compare two different lattices and instantly know if they are secretly the same shape, just rotated or flipped. This is crucial for understanding which digital locks are truly unique and which might be vulnerable because they are actually identical to a known, weaker design.

The Great Grid Detective: Unpacking the Invisible

Imagine you have a giant, invisible 3D puzzle made of numbers. It's so complex that looking at the whole thing at once makes your brain hurt. Now, imagine you have a magic pair of scissors that can cut this puzzle into smaller, simpler pieces. If you can describe exactly what those smaller pieces are, you can describe the whole puzzle. This is essentially what Guilhem Mureau has done for a specific type of mathematical object called an "ideal lattice" inside a "CM field."

In the paper, Mureau acts like a master detective trying to figure out the "fingerprint" of these number grids. He focuses on a specific question: If you have two different grids built from the same number city, how can you tell if they are actually the same shape underneath? To answer this, he uses a method called Jordan decomposition.

Think of a Jordan decomposition like sorting a messy pile of socks. You don't just throw them in a bin; you group them by size and color. In math, you group parts of the lattice by their "scale" (how stretched out they are) and their "shape" (how they twist). Mureau's paper provides the exact instructions for this sorting process. He shows how to look at the "prime factors" of the ideal (which are like the unique DNA of the lattice) and immediately read off the sizes and shapes of the pieces you'll get when you cut it up.

The Two Worlds: The Easy Way and the Tricky Way

Mureau's discovery splits the problem into two distinct scenarios, much like how a video game might have a "Day Mode" and a "Night Mode" with different rules.

1. The "Non-Dyadic" Day (The Easy Mode)
This happens when the number 2 is not a problem in the local neighborhood. In this world, the rules are surprisingly simple. Mureau found that if you know how the lattice is built from its prime ingredients, you can write down a closed formula—a direct recipe—to tell you exactly how the lattice breaks down.

The Analogy: Imagine you have a bag of Lego bricks. In this "Day" mode, if you tell me how many red bricks and blue bricks you have, I can instantly tell you exactly how many small towers and big walls you can build. There's no guessing. The paper proves that for these cases, the local shape of the lattice is determined entirely by the "valuations" (a fancy word for the count of prime ingredients) of the ideal.
The Result: If two lattices have the same counts of prime ingredients in the right places, they are isometric (identical in shape). Mureau gives a clear checklist: if the lists of counts match, the lattices match.

2. The "Dyadic" Night (The Tricky Mode)
This is where things get messy. This happens when the number 2 is involved in a specific way (when 2 belongs to the prime ideal). In this "Night" mode, the simple rules of the Day mode break down. The "socks" are now sticky and weirdly shaped.

The Analogy: Now, just counting the red and blue bricks isn't enough. You also need to know if the bricks are slightly warped or if they have a secret twist. Mureau explains that in this mode, you need extra "invariants" (extra clues) like the "norm group" and "weight" to tell the shapes apart.
The Catch: Mureau doesn't solve every possible case in this mode. He puts up a sign that says, "Warning: We can only solve this if the lattice doesn't have certain complicated prime ingredients." Specifically, he rules out cases where the prime ideal divides the number $a$ (a specific part of the field's definition) in the dyadic setting. He admits that for these specific, messy cases, the math gets too tangled to give a simple formula right now.
The Result: For the cases he can solve (where the prime doesn't divide $a$ ), he provides a new recipe. He shows that if the "residue degree" (a measure of the neighborhood's size) is odd, you can still break the lattice down into pieces, though one piece remains a bit mysterious and requires a special "anisotropic" block (a piece that refuses to flatten out).

Why This Matters: The Cryptography Connection

Why should a curious teenager care about sorting invisible number socks? Because these lattices are the backbone of post-quantum cryptography.

Imagine a future where supercomputers can break today's internet security. To stop them, scientists are building new locks based on these complex lattices. The security of these locks depends on the fact that two different-looking lattices are actually impossible to turn into each other.

Mureau's paper is like giving the lock-makers a new magnifying glass. Before this, if they wanted to check if two lattices were the same, they might have had to do a massive, slow calculation. Now, thanks to Mureau's formulas, they can look at the "prime factorization" (the ingredient list) and instantly know the answer.

If the lattices are p-separated: This is a fancy way of saying the ingredients are spread out enough that they don't get mixed up. If they are separated, Mureau's rules say you can compare them piece-by-piece. If the ingredient lists match, the locks are identical.
If they aren't separated: The pieces might overlap, making the comparison harder. Mureau acknowledges this limitation but provides the tools to handle the cases where the pieces don't overlap.

The Bottom Line

Guilhem Mureau hasn't solved every single mystery in the universe of number lattices. He hasn't cracked the code for every possible "Night Mode" scenario, and he explicitly states that his formulas work best when the prime ingredients are well-behaved (specifically, when the ideal is "p-separated" and, in the tricky dyadic cases, when the prime doesn't divide a specific number $a$ ).

However, for the vast majority of cases that matter in cryptography, he has provided a proven, explicit formula. He has turned a vague, difficult problem into a clear, step-by-step instruction manual. He showed that by looking at the "DNA" of the lattice (its prime factorization), you can predict its "skeleton" (its Jordan decomposition) with total certainty. This gives mathematicians and cryptographers a powerful new way to test their digital locks, ensuring that the secrets of the future remain safe.

Technical Summary: Explicit Jordan Decompositions for Ideal Lattices in CM Fields

Problem Statement
The paper addresses the problem of determining explicit closed formulas for the Jordan decomposition of ideal lattices defined over CM (Complex Multiplication) number fields. Specifically, let $E$ be a CM number field, $F$ a totally real subfield (e.g., $\mathbb{Q}$ ), and $\mathfrak{a}$ a non-zero fractional ideal of $E$ . The lattice is defined as $L_{\mathfrak{a}, F} = (\mathfrak{a}, h_{E/F})$ , where $h_{E/F}(x, y) = \text{Tr}_{E/F}(xy)$ is the trace form.

While the Jordan decomposition of the ring of integers $(\mathcal{O}_K, \text{Tr}_{K/\mathbb{Q}}(x^2))$ has been studied extensively (e.g., by Erez, Morales, and Perlis), those works typically vary the field $K$ to produce arithmetic invariants of the field itself. In contrast, this paper fixes the ambient field $E$ and varies the lattice through the fractional ideals of $E$ . The primary goal is to express the local Jordan decomposition of $L_{\mathfrak{a}, F}$ at a prime ideal $\mathfrak{p} \subset \mathcal{O}_F$ directly in terms of the prime ideal factorization of $\mathfrak{a}$ in $E$ . This provides local invariants for the isometry relation between ideal lattices, a problem relevant to the lattice isomorphism problem in cryptography.

Methodology
The approach follows the strategy of reducing global computations to local behaviors at prime ideals, adapting methods from [2, 6]. The methodology proceeds in three main stages:

Local Decomposition: The global lattice $L_{\mathfrak{a}, F}$ is decomposed into an orthogonal sum of local lattices $L_{\mathfrak{a}_P, F_P}$ corresponding to the completions at primes $P$ of $\mathcal{O}_E$ lying above $\mathfrak{p}$ . This relies on the tensor product decomposition of completions.
Tower Descent: For each local component, the computation is reduced through a tower of field extensions:
- Over the CM extension ( $E/K$ ): The local lattice is analyzed based on the splitting behavior of the prime $P$ in the quadratic extension $E/K$ (Split, Inert, or Ramified).
- Over the totally ramified extension ( $K/M$ ): The lattice is descended from the ramified extension $K$ to its maximal unramified subextension $M$ . This step involves explicit Gram matrix computations using uniformizers and trace formulas.
- Over the unramified extension ( $M/F$ ): Finally, the lattice is descended from the unramified extension $M$ to the base field $F$ .
Case Distinction: The analysis is split into two distinct cases based on the prime $\mathfrak{p}$ $p$ :
- Non-dyadic case ( $2 \notin \mathfrak{p}$ ): The classification of quadratic lattices is simpler, relying on rank, scale, and discriminant.
- Dyadic case ( $2 \in \mathfrak{p}$ ): The classification requires additional invariants (norm group and weight) due to the non-unit nature of 2. This case imposes stricter assumptions, particularly excluding ramified primes where the ideal $\mathfrak{a}$ is divisible by the prime in question.

Key Contributions and Results

Explicit Formulas: The paper provides explicit formulas for the Jordan blocks of $L_{\mathfrak{a}, F}$ at a prime $\mathfrak{p}$ . These formulas are derived directly from the valuations of $\mathfrak{a}$ at the prime ideals $P$ of $E$ lying above $\mathfrak{p}$ .
- In the non-dyadic case, the Jordan components are described in terms of the Euclidean division of the valuations by the ramification index. Theorem 4.14 and Propositions 4.12–4.13 detail the isometry classes for split, inert, and ramified primes.
- In the dyadic case, the results (Theorem 5.13 and Propositions 5.10–5.11) describe the decomposition using hyperbolic planes and specific anisotropic blocks involving the quadratic defect, under the assumption that $f$ (the residue degree) is odd and $\mathfrak{a}$ is not divisible by ramified primes.
Local Isometry Criteria: The paper establishes criteria for when two ideal lattices are locally isometric at $\mathfrak{p}$ .
- The $p$ -separated condition: To avoid the combinatorial complexity of merging Jordan blocks from different primes $P | \mathfrak{p}$ , the author introduces the concept of a " $p$ -separated" ideal (Definition 4.17), where the sets of scales contributed by distinct primes are disjoint.
- Corollaries 4.21 and 5.14: Under the assumption that $K/F$ is Galois and the ideals are $p$ -separated, the paper provides necessary and sufficient conditions for local isometry. These conditions depend on the multisets of valuations $u_P(\mathfrak{a})$ (defined based on the splitting type of $P$ ) and the parity of the residue degree $f$ .
Handling of Dyadic Primes: The paper explicitly addresses the difficulties of the dyadic case, where standard invariants are insufficient. It utilizes the quadratic defect and specific structural lemmas (e.g., Lemma 5.7 regarding the behavior of the unramified quadratic extension under field extension) to derive explicit decompositions, albeit with the restriction that $f$ must be odd for the descent of certain units to hold.

Significance and Claims
The paper claims to provide a tractable class of ideal lattices for studying which arithmetic information is captured by local quadratic invariants. The results serve two main purposes:

Algorithmic: They provide effective local invariants for comparing ideal lattices, which is a prerequisite for solving the lattice isomorphism problem in the context of module lattices over CM fields.
Arithmetic: The decompositions serve as a starting point for studying spinor genera and local densities of these lattices, extending the work of Mantilla-Soler and Fiori on integral trace forms.

The author notes that while the cryptographic instances of interest often involve higher module ranks, the ideal lattice case provides a natural testing ground. The paper does not claim to solve the general lattice isomorphism problem for all module ranks but rather establishes the foundational local invariants for the ideal case. The results are presented as explicit formulas that allow one to "read" local invariants directly from the prime ideal factorization of the ideal $\mathfrak{a}$ .

The Great Grid Detective: Unpacking the Invisible

The Two Worlds: The Easy Way and the Tricky Way

Why This Matters: The Cryptography Connection

The Bottom Line

More like this