🔢 mathematics

$Ω$ -Results for Exponential Sums Related to Maass Cusp Forms for $\mathrm{SL}_3(\mathbb Z)$

This paper establishes $\Omega$ -results for linear exponential sums weighted by Hecke eigenvalues of Maass cusp forms for $\mathrm{SL}_3(\mathbb{Z})$ with small prime denominators, demonstrating that these lower bounds match conjectural upper limits in specific ranges while also providing improved upper bounds and non-trivial results for short segments.

Original authors: Jesse Jääsaari

Published 2026-02-06

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Jesse Jääsaari

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 trying to listen to a specific, faint melody played on a very complex instrument. The instrument is a "Maass cusp form" for the group $SL_3(\mathbb{Z})$ , and the notes it plays are numbers called Hecke eigenvalues (let's call them $A(m, 1)$ ).

These numbers are mysterious. Sometimes they seem to cancel each other out perfectly (like noise-canceling headphones), and sometimes they build up into a loud, chaotic roar. Mathematicians want to know: How loud can this music get? Specifically, if we add a "twist" to the melody—shifting the notes by a fraction like $1/k$ (where $k$ is a small prime number)—what is the maximum volume we can expect?

This paper, by Jesse Jäätääri, is about proving that the music can get surprisingly loud, matching the loudest volume mathematicians had guessed was possible.

Here is a breakdown of the paper's journey using everyday analogies:

1. The Problem: The "Sharp Cut" vs. The "Smooth Fade"

Imagine you are counting the total volume of this music from the start up to a specific time $x$ .

The Sharp Cut: You suddenly stop the music at exactly time $x$ . This is mathematically "messy" and hard to analyze, like trying to stop a spinning top instantly without it wobbling.
The Smooth Fade: You slowly turn down the volume over a period of time. This is mathematically "clean" and easy to handle.

The author wants to study the Sharp Cut (the real-world scenario), but the math is too difficult to do directly. So, he uses a clever trick.

2. The Trick: The "Riesz Weighted" Ladder

The author builds a ladder of "smoothed" versions of the music.

Step 0: The messy, sharp-cut sum (the one we actually care about).
Step 1, 2, 3...: Smoother, weighted versions of the sum.

Think of this like a transmission system. You can't easily measure the speed of a car's wheels directly (Step 0) because they are spinning too fast and irregularly. But you can easily measure the speed of the engine's flywheel (Step 2 or 3) because it spins smoothly.

The author proves a special relationship (an identity) that connects the smooth flywheel back to the messy wheels. If you know how loud the smooth version is, you can deduce how loud the messy version must be.

3. The Tool: The "Voronoi Mirror"

To analyze the smooth versions, the author uses a mathematical tool called a Voronoi summation formula.

Imagine you have a complicated pattern of dots (the music notes).
The Voronoi formula acts like a magic mirror. It reflects the pattern into a new world (the "dual side").
In this new world, the pattern often looks simpler or behaves in a predictable way.

However, for these specific 3D musical instruments ( $SL_3$ ), the mirror usually produces a lot of "static" (error terms) that drowns out the signal. The author's breakthrough was finding a way to use the mirror on the higher steps of his ladder (Steps 2 and 3) where the static is manageable, and then using the ladder to pull that information back down to the messy Step 0.

4. The Main Discovery: The "Omega" Result

In mathematics, an $\Omega$ -result is a way of saying, "No matter how you try to suppress it, this thing will reach at least this loud." It's a guarantee of a lower bound.

The paper proves two main things:

The Long Sum (The Whole Concert): When you listen to the music from the beginning up to a very long time $x$ , and the twist is a simple fraction (like $1/k$ ), the volume will definitely reach a level proportional to $k^{1/2} x^{1/3}$ .
- Why this matters: This matches the "conjectural upper bound." It means the music is as loud as the smartest mathematicians thought it could possibly be. The author has proven that the "noise" isn't just random; it has a specific, powerful peak.
The Short Sum (The Quick Glance): The author also looked at short segments of the music (from time $x$ to $x + \Delta$ ). Even in these short bursts, the music can get loud, provided the segment isn't too short and the fraction $k$ isn't too big.

5. The "Averaging" Secret

One tricky part of the proof is that the volume depends on which fraction you choose (the numerator $h$ ). Sometimes the music is quiet for $h=1$ but loud for $h=2$ .
To solve this, the author didn't try to find the loudest single fraction immediately. Instead, he averaged the volume over all possible fractions (all numerators $h$ that are coprime to $k$ ).

Analogy: Imagine trying to find the loudest speaker in a room full of 100 speakers. Instead of testing them one by one, you turn them all on at once and measure the total roar. If the total roar is loud, you know at least one of them must be screaming. This averaging technique allowed the author to isolate the "main signal" from the "static."

6. The Bonus: A Better Upper Bound

While proving the music can get loud, the author also improved the best-known estimate for how loud it can't get (the upper bound).

Previously, the "ceiling" for the volume was a bit loose.
The author tightened this ceiling, showing that the music is even more constrained than we thought, but still reaches the peak predicted by the lower bound.

Summary

Jesse Jäätääri solved a puzzle about the "loudness" of a complex mathematical music piece.

He couldn't measure the messy, raw sound directly.
He built a ladder to measure the smooth, clean versions of the sound instead.
He used a "magic mirror" (Voronoi formula) to analyze the clean versions.
He used the ladder to pull those results back down to the messy version.
The Result: He proved that the music definitely reaches a specific, high volume (matching the theoretical maximum) and provided a tighter limit on how quiet it can be.

This confirms that the distribution of these mysterious numbers has a specific, powerful structure, rather than being purely random noise.

Technical Summary: $\Omega$ -Results for Exponential Sums Related to Maass Cusp Forms for $SL_3(\mathbb{Z})$

1. Problem Statement and Context
The paper investigates the extent of cancellation in linear exponential sums weighted by Hecke eigenvalues of Maass cusp forms for the group $SL_3(\mathbb{Z})$ . Specifically, it analyzes sums of the form:
$\sum_{x \le m \le x+\Delta} A(m, 1) e\left(\frac{mh}{k}\right)$
where $A(m, 1)$ are the Fourier coefficients (Hecke eigenvalues) of a fixed Maass cusp form, $h/k$ is a rational twist with a small denominator $k$ , and $\Delta$ represents the length of the summation interval (which may be a long sum where $\Delta \approx x$ or a short sum where $\Delta \ll x$ ).

While significant literature exists for such sums in the classical rank-one setting (modular forms), results for higher-rank forms ( $GL_3$ ) are sparse. The author notes that existing upper bounds for long sums, such as Miller's $O_\varepsilon(x^{3/4+\varepsilon})$ , are not always sharp compared to conjectural bounds derived from moment heuristics and Voronoi summation identities. A primary challenge in the higher-rank setting is that standard truncated Voronoi summation formulas often produce error terms that dominate the expected main terms, rendering them ineffective for deriving precise pointwise bounds or $\Omega$ -results (lower bounds on the magnitude of the sum).

2. Methodology
To circumvent the limitations of truncated Voronoi identities in the $GL_3$ setting, the author employs a multi-step strategy centered on Riesz weighted sums and moment analysis:

Riesz Weighted Sums: Instead of analyzing the sharp cut-off sum directly (corresponding to the 0-th order Riesz mean), the paper introduces smoothed sums of higher order $a \ge 1$ :
$\mathcal{A}_a\left(x; \frac{h}{k}\right) := \frac{1}{a!} \sum_{m \le x}' A(m, 1) e\left(\frac{mh}{k}\right) (x-m)^a - \text{residue terms}.$
The smoothing parameter $a$ is chosen such that the associated Voronoi summation formulas on the dual side converge absolutely, avoiding the large error terms associated with sharp cut-offs.
Voronoi Identities for Riesz Means: The paper derives new Voronoi summation identities for these Riesz weighted sums.
- For $a \ge 2$ , the identities are derived using Perron's formula and the asymptotic analysis of integrals involving quotients of Gamma functions, identified as Meijer $G$ -functions. This approach simplifies the derivation of asymptotics compared to previous methods.
- For $a = 1$ , the derivation is more delicate due to convergence issues. The author establishes the identity by differentiating the $a=2$ case termwise, utilizing specific convergence arguments and discrete Fourier transforms of Kloosterman sums to handle the resulting series.
Moment Relations and Averaging: The core technical innovation lies in relating the moments of sums of different orders. Using the fundamental property $\int_x^{x+H} \mathcal{A}_a(t) dt = \mathcal{A}_{a+1}(x+H) - \mathcal{A}_{a+1}(x)$ , the author constructs inequalities linking the second moments of $\mathcal{A}_0$ , $\mathcal{A}_1$ , and $\mathcal{A}_2$ .
- The second moments of the higher-order sums ( $\mathcal{A}_1$ and $\mathcal{A}_2$ ) are evaluated asymptotically using the derived Voronoi identities.
- Crucially, the author averages over the numerators $h \in (\mathbb{Z}/k\mathbb{Z})^\times$ . This averaging is necessary to extract positive main terms from the diagonal contributions in the second moment calculations, overcoming the difficulty that Kloosterman sums in higher rank do not automatically cancel in the same way as in rank one.
- By establishing a lower bound for the moment of $\mathcal{A}_1$ and an upper bound for $\mathcal{A}_2$ , the author deduces a lower bound for the moment of the original sharp cut-off sum $\mathcal{A}_0$ (which corresponds to the sum in question).

3. Key Contributions and Results
The paper establishes several new results regarding the lower bounds ( $\Omega$ -results) and upper bounds for these exponential sums:

Optimal $\Omega$ -Results for Long Sums:
For a prime denominator $k \ll x^{1/3-\delta}$ , the paper proves:
$\max_{h \in (\mathbb{Z}/k\mathbb{Z})^\times} \left| \sum_{m \le x}' A(m, 1) e\left(\frac{mh}{k}\right) \right| = \Omega\left( k^{1/2} x^{1/3} \right).$
This matches the conjectural upper bound $O_\varepsilon(k^{1/2} x^{1/3+\varepsilon})$ in this range, providing the first $\Omega$ -results for rationally additively twisted sums in a higher-rank setting.
Non-Trivial $\Omega$ -Results for Short Sums:
For short sums with length $\Delta$ satisfying $k^{3/2} x^{1/2+\delta} \ll \Delta \ll k x^{2/3-\delta}$ , the paper proves:
$\max_{h \in (\mathbb{Z}/k\mathbb{Z})^\times} \left| \sum_{x \le m \le x+\Delta}' A(m, 1) e\left(\frac{mh}{k}\right) \right| = \Omega\left( \Delta x^{-1/3} k^{-1/2} \right).$
This result is of the correct order of magnitude relative to the conjectural bound when $\Delta \approx k x^{2/3}$ .
Improved Upper Bound:
As a byproduct of the analysis, the paper improves the known upper bound for long sums with small denominators. For $1 \le k \ll x^{1/3}$ :
$\sum_{m \le x} A(m, 1) e\left(\frac{mh}{k}\right) \ll_\varepsilon k^{3/4} x^{1/2 + \vartheta/2 + \varepsilon},$
where $\vartheta$ is the exponent towards the Ramanujan–Petersson conjecture for $SL_3(\mathbb{Z})$ . This improves upon previous bounds which included larger error terms depending on $k$ .

4. Significance and Claims
The author claims that this work provides the first evidence of $\Omega$ -results for rationally additively twisted sums in the higher-rank setting, demonstrating that the conjectured cancellation rates are indeed achievable (or at least not violated) in the lower bound sense.

The significance of the paper lies in:

Methodological Advancement: It successfully navigates the technical difficulties of higher-rank Voronoi summation by utilizing Riesz means and Meijer $G$ -functions, bypassing the dominance of error terms that plagues sharp cut-off approaches.
Bridging Theory and Conjecture: The results align with the conjectural bounds derived from moment heuristics and the shape of truncated Voronoi identities, offering strong support for these conjectures in the $GL_3$ context.
Practical Utility: The improved upper bound and the techniques for handling short sums with rational twists are presented as practical tools for reducing smoothing errors in other analytic number theory problems involving linear exponential sums.

The paper remains modest regarding the generality of its results, noting that the restriction to prime denominators $k$ is primarily for technical simplicity in handling divisor sums, and that the averaging over $h$ is a necessary component of the current method to extract main terms. It does not claim to resolve the full distribution of these sums for arbitrary parameters but establishes the existence of large values within the expected ranges.