🔢 mathematics

Congruences and ramified primes in fields of coefficients of newforms

This paper investigates the splitting behavior of a prime $\ell$ in the coefficient field of a newform $f$ that is congruent modulo $\ell$ to another newform $g$ of lower level, demonstrating that the maximal real subfield of the $\ell$ -th cyclotomic field is necessarily contained within the coefficient field of $f$ .

Original authors: Nuno Freitas, Filip Gawron

Published 2026-04-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Nuno Freitas, Filip Gawron

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 detective trying to solve a mystery about a very special kind of mathematical object called a Newform.

Think of a Newform as a complex, multi-layered musical score. This score has a specific rhythm (called the Level, $N$ ) and a specific melody (the Coefficients). The "field of coefficients" is like the unique language or dialect in which this melody is written. Usually, this language is simple, but sometimes it gets incredibly complicated, containing hidden secrets.

The paper by Freitas and Gawron is about finding a specific clue that tells us when this language must become complicated.

The Main Mystery: "The Splitting of Primes"

In the world of numbers, primes (like 2, 3, 5, 7, 11...) are the building blocks. When you take a complicated number system (like the language of our Newform) and look at how these building blocks behave inside it, they can either stay whole or "split" into smaller pieces.

The authors are asking: "If our musical score (the Newform) has a very specific, messy rhythm (Level $N$ ), does that force the language (the Coefficient Field) to contain a specific, complex ingredient?"

The Clue: The "Congruence" Connection

The authors discovered a powerful trick. Imagine you have two musical scores:

Score A (The Big One): A complex piece with a very high, messy level $N$ .
Score B (The Simple One): A much simpler piece with a lower level.

The authors found that if Score A and Score B sound almost identical when you listen to them through a specific, low-quality filter (a "modulo $\ell$ " filter), then Score A is forced to contain a very specific, complex ingredient in its language.

This ingredient is a Root of Unity.

Analogy: Imagine the language of the Newform is a house. The authors prove that if the house has a "messy" foundation (high level $N$ ) and looks like a simpler house next door, then the house must have a secret room containing a specific, beautiful crystal chandelier (the root of unity). You can't have the messy foundation and the similarity without the chandelier.

The "Messy Foundation" (Ramification)

The paper focuses on a specific type of messiness called Ramification.

Normal Foundation: The level $N$ is divisible by a prime $p$ just once ( $p \times \dots$ ).
Messy Foundation: The level $N$ is divisible by $p$ squared or cubed ( $p^2 \times \dots$ or $p^3 \times \dots$ ).

The authors prove that if the foundation is "messy" (divisible by $p^2$ ) but the "simplified version" of the music (the Galois representation) suddenly becomes cleaner (loses some of that messiness), then the language of the Newform must contain the full set of these crystal chandeliers.

The Two Main Discoveries

The paper splits its findings into two scenarios based on how "messy" the foundation is:

The "Double Mess" ( $p^2$ divides $N$ ):
If the level is divisible by $p^2$ , the language of the Newform must contain a "half-chandelier" (specifically, the real part of the root of unity).
- Real-world analogy: If you build a house with a double-thick, unstable wall, the interior design must include a specific type of mirror.
The "Triple Mess" ( $p^3$ or more divides $N$ ):
If the level is even messier (divisible by $p^3$ ), the language must contain the entire chandelier (the full cyclotomic field).
- Real-world analogy: If the wall is triple-thick and unstable, the interior design must include the entire crystal chandelier, not just the mirror.

Why Does This Matter? (The "Sherlock Holmes" Twist)

The authors don't just prove this exists; they show you how to build these complex Newforms on purpose.

They use a method called Level Raising.

The Trick: You start with a simple, well-known song (a Newform with a small level).
The Magic: You find a prime number $p$ and a filter $\ell$ such that if you "raise the level" of the song (make the foundation messier), the new song will sound exactly like the old one through the filter.
The Result: Because of the theorem, you know for a fact that this new, messy song has a very complex language with highly "ramified" primes (primes that split in a very specific, deep way).

The "Crystal Ball" (Examples)

In the last part of the paper, they act like architects. They say, "Let's build a house with a triple-thick wall using prime 23 and filter 11."
They run the math, and sure enough, they generate a Newform where the number 11 splits in a very dramatic way inside the language. They provide a table of these "built" houses, showing that their theory works perfectly in practice.

Summary in One Sentence

If a complex mathematical song (Newform) has a very "bumpy" rhythm (high level) but sounds like a simpler song when viewed through a specific lens, then the language of that song is guaranteed to contain a specific, complex mathematical structure (roots of unity), and the authors show exactly how to construct such songs to prove it.

1. Problem Statement

The paper investigates the arithmetic properties of the coefficient field $Q_f$ of a newform $f \in S_k(\Gamma_0(N), \varepsilon)$ . Specifically, it addresses Question 1.1: What can be said about the splitting behavior of primes in $Q_f$ in terms of the level $N$ and the nebentypus $\varepsilon$ ?

While it is known that $Q_f$ contains values of the nebentypus and certain cyclotomic fields based on the divisibility of $N$ (e.g., Brumer's results), the authors focus on a specific phenomenon: ramification loss.

Let $\rho_{f,\lambda}$ be the Galois representation associated with $f$ at a prime $\lambda \mid \ell$ in $Q_f$ .
Let $N'$ be the prime-to- $\ell$ part of the Artin conductor of the semisimplification $\rho_{f,\lambda}^{ss}$ .
The paper considers the scenario where $f$ has a level $N$ divisible by $p^2$ (for a prime $p \neq \ell$ ), but the associated Galois representation "loses" ramification at $p$ (i.e., $v_p(N') < v_p(N)$ ).
Core Question: If the Galois representation loses ramification at a prime $p$ dividing the level, does the prime $\ell$ necessarily ramify in the coefficient field $Q_f$ ? If so, to what extent?

2. Methodology

The authors employ a combination of Galois representation theory, local class field theory, and congruence properties of modular forms.

Carayol's Classification: The proof relies heavily on Carayol's classification of possible Weil-Deligne types for representations with conductor exponent $a(\rho) \geq 2$ where the semisimplification has a strictly smaller conductor ( $a(\rho^{ss}) < a(\rho)$ ). This restricts the inertial type of $f$ at $p$ to three cases: Principal Series, Special (Steinberg), or Supercuspidal.
Inertial Type Analysis: By analyzing the restriction of the Galois representation to the inertia subgroup $I_p$ , the authors determine the specific roots of unity that must appear in the coefficient field based on the character values of the inertial type.
Torsion-Free Kernel Argument: To establish lower bounds on the ramification degree $e(\lambda \mid \ell)$ , the authors utilize a lemma regarding the kernel of the reduction map $GL_2(\mathcal{O}) \to GL_2(\mathbb{F}_\ell)$ . They argue that if the ramification degree is small ( $e < \ell - 1$ ), this kernel is torsion-free. Since the inertial image is a finite non-trivial subgroup of this kernel (due to the ramification loss), a contradiction arises unless the ramification degree is sufficiently large.
Level Raising (Diamond-Taylor): To generate explicit examples, the authors use the level raising theorem of Diamond and Taylor. This allows them to construct newforms $f$ of higher level ( $p^\alpha N'$ ) that are congruent to a lower-level form $g$ , thereby forcing the specific ramification loss conditions required by their main theorem.

3. Key Contributions and Results

Theorem 1.2 (Main Result)

Let $f$ be a newform of weight $k \geq 2$ and level $N$ . Let $p$ be a prime such that $p^2 \mid N$ , and let $\ell \neq p$ be an odd prime. Let $\lambda$ be a prime of $Q_f$ above $\ell$ , and let $N'$ be the prime-to- $\ell$ part of the Artin conductor of $\rho_{f,\lambda}^{ss}$ .
Assumption: $v_p(N') < v_p(N)$ (Ramification loss at $p$ ).

The theorem establishes the following containments and ramification bounds:

Case $v_p(N) = 2$ :
- The maximal real subfield of the $\ell$ -th cyclotomic field is contained in $Q_f$ :
  $\mathbb{Q}(\zeta_\ell + \zeta_\ell^{-1}) \subset Q_f$
- Consequently, for any prime $\lambda'$ of $Q_f$ above $\ell$ , the ramification degree satisfies:
  $\frac{\ell-1}{2} \mid e(\lambda' \mid \ell)$
- Stronger Result: If $v_p(N') = 0$ (total loss of ramification at $p$ ) and $\rho_{f,\lambda}$ is irreducible, then there exists a prime $\lambda$ above $\ell$ such that:
  $e(\lambda \mid \ell) \geq \ell - 1$
Case $v_p(N) > 2$ :
- The full cyclotomic field is contained in $Q_f$ :
  $\mathbb{Q}(\zeta_\ell) \subset Q_f$

Significance of the Result

Generalization: This generalizes previous results (e.g., by Freitas and others) by allowing for non-trivial nebentypus and providing stronger conclusions regarding the specific cyclotomic subfields contained in $Q_f$ .
Ramification Degree: It provides a concrete lower bound on the ramification index of primes above $\ell$ in $Q_f$ , linking the geometric property of the modular form (level) directly to the arithmetic property of its coefficient field.
Necessity of Conditions: Example 4.2 demonstrates that the condition $p^2 \mid N$ is necessary; if $p \parallel N$ (only simple ramification), ramification loss can occur without $\ell$ ramifying in $Q_f$ .

4. Explicit Examples and Computational Verification

The authors provide a systematic method to generate examples satisfying the theorem's hypotheses using Proposition 4.3 (a level-raising result).

Method: Start with a newform $g$ of level $N'$ (where $\ell \nmid N'$ ) and a prime $p$ . If specific congruence conditions are met (e.g., $\ell \mid p \pm 1$ ), one can find a newform $f$ of level $p^\alpha N'$ congruent to $g$ .
Tables 1 & 2: The paper presents extensive tables of examples for various pairs $(\ell, p)$ $(ℓ, p)$ and levels $N'$ $N^{'}$ .
- Table 1 covers the case $v_p(N)=2, v_p(N')=0$ .
- Table 2 covers the case $v_p(N)=2, v_p(N')=1$ .
Observations: In all computed examples, the ramification degree of primes above $\ell$ is divisible by $(\ell-1)/2$ . Furthermore, there is always at least one prime $\lambda$ where $e(\lambda \mid \ell) \geq \ell - 1$ , confirming the theoretical lower bound.

5. Significance and Future Directions

Theoretical Impact: The work deepens the understanding of the relationship between the level of a modular form and the arithmetic of its coefficient field. It confirms that "hidden" ramification in the Galois representation (loss of conductor) forces significant ramification in the field of coefficients.
Computational Utility: The paper offers a constructive algorithm to generate newforms with highly ramified coefficient fields. This is useful for testing conjectures in the Langlands program and studying the distribution of ramification in families of modular forms.
Hilbert Modular Forms: Remark 1.3 notes that the results and proofs extend naturally to Hilbert newforms over totally real fields, suggesting broad applicability beyond classical modular forms.

In summary, Freitas and Gawron establish a rigorous link between the local behavior of Galois representations at primes dividing the level and the global ramification structure of the coefficient fields of newforms, providing both theoretical bounds and explicit computational evidence.