🔢 mathematics

Metric Poissonian pair correlationa and additive energy

This paper establishes that a strictly increasing sequence of natural numbers exhibits Poissonian pair correlation for almost all real numbers $\alpha$ if its additive energy is bounded by $N^3/(\log N)^C$ for some constant $C \geq 14.71$ , thereby providing a specific lower bound for the exponent in the additive energy condition previously studied by Bloom and Walker.

Original authors: Tanmoy Bera, E. Malavika

Published 2026-07-14

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Tanmoy Bera, E. Malavika

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 have a long line of natural numbers, like a string of beads: 1, 2, 3, 4, and so on. Now, imagine you take this string and wrap it around a circle, but you do it in a weird, stretched-out way determined by a secret number, let's call it $\alpha$ . This creates a new pattern of dots on the circle.

Mathematicians love to ask: "How evenly are these dots spread out?" If they are perfectly scattered, like rain on a window, we say they are "uniformly distributed." But there's a stricter, more magical test called Poissonian Pair Correlation (PPC). Think of this as checking if the dots are not just scattered, but scattered with a specific, random-like "friendship" distance. If you look at any two dots, the chance they are a certain distance apart follows a very precise, predictable rule, just like how molecules in a gas behave.

For a long time, mathematicians knew that if your sequence of numbers grows "fast enough" or has a specific structure, it passes this test for almost all secret numbers $\alpha$ . But what if the sequence is messy? How messy can it be before it fails the test?

This is where Additive Energy comes in. Think of additive energy as a measure of "clumping" or "repetition" in your number sequence. If you take your numbers and start adding them together ( $a+b=c+d$ ), a high energy means you have a lot of matching sums—a lot of hidden patterns and clumps. A low energy means the numbers are behaving more like a chaotic, random mess.

In this paper, authors Tanmoy Bera and E. Malavika act like detectives trying to find the exact tipping point. They want to know: How low does the additive energy have to be to guarantee that the sequence passes the PPC test?

Previous detectives (Bloom and Walker) had set a rule: "If the energy is less than $N^3$ divided by $(\log N)^C$ , then you're good." But they didn't know exactly how big the number $C$ needed to be. They guessed it might be just barely bigger than 1, but they couldn't prove it.

Bera and Malavika stepped in to tighten the net. They didn't just guess; they did the heavy lifting of the math to find a concrete, safe lower bound. They proved that if the additive energy is less than $N^3 / (\log N)^C$ where $C$ is at least 14.71, then the sequence definitely has Poissonian pair correlation for almost all $\alpha$ .

Here is the catch: The paper doesn't say the magic number is exactly 14.71 and that's the absolute limit of the universe. Instead, they show that 14.71 is the smallest number their specific method can currently prove works. They explain that their math relies on a specific "moment lemma" (a tool for measuring randomness) that involves a constant called $\beta$ , which is approximately 1.7032. Because of how this tool works, the math forces $C$ to be at least 14.71 (which is roughly $\beta + 13$ ).

The authors are very honest about the limitations of their own map. They suggest that while 14.71 is the best they can do with this particular tool, the true answer might be much lower (perhaps close to 1, as Bloom and Walker conjectured). They admit that to get closer to that lower number, someone would need a completely different approach, not just a tweak of their current method. They believe there is very little room to squeeze the number down further using the techniques they employed.

So, the main takeaway is this: We now have a solid, proven guarantee that if a sequence's "clumping" (additive energy) is low enough—specifically, if it stays below the threshold defined by $C \ge 14.71$ —then the sequence behaves with perfect, random-like spacing for almost all secret numbers. It's a big step forward, even if the ultimate "perfect" threshold might still be hiding in the shadows, waiting for a new kind of mathematical flashlight.

As a bonus, they show that this rule applies to sequences like $[n(\log n)^K]$ (where you take a number, multiply it by the log of itself, and raise it to a power) as long as that power $K$ is at least 15.71. This confirms that these specific, slightly messy sequences are indeed "random enough" to pass the test.

Technical Summary: Metric Poissonian Pair Correlation and Additive Energy

Problem Statement
The paper addresses the relationship between the additive energy of a strictly increasing sequence of natural numbers $(a_n)$ and the metric Poissonian pair correlation (MPPC) property of the sequence of fractional parts $({a_n \alpha})$ . A sequence $(x_n)$ has Poissonian pair correlation (PPC) if its pair correlation function $R_2(s, (x_n), N)$ converges to $2s$ as $N \to \infty$ for any $s > 0$ . The property is termed "metric" if it holds for almost all $\alpha \in \mathbb{R}$ .

Previous work by Aistleitner, Larcher, and Lewko established that if the additive energy $E(A_N)$ of the first $N$ terms satisfies $E(A_N) \ll N^{3-\delta}$ for some $\delta > 0$ , the sequence has MPPC. Bloom and Walker later relaxed this condition, showing that $E(A_N) \le N^3 / (\log N)^C$ suffices for MPPC provided $C$ is sufficiently large. However, Bloom and Walker did not specify a concrete lower bound for $C$ , merely conjecturing that any $C > 1$ might suffice. This paper aims to provide an explicit lower bound for the exponent $C$ .

Methodology
The authors employ a probabilistic approach involving random models and GCD sums to derive explicit constants. The core methodology involves:

Random Zeta Functions: Utilizing a random model introduced by Lewko and Radziwiłł, the authors define a random zeta function $\zeta_X(\sigma)$ based on independent random variables uniformly distributed on the unit circle.
Moment Bounds (Lemma 2.1): A critical component is the derivation of an explicit upper bound for the $2\ell$ -th moment of the random zeta function. The authors prove that for $1/2 < \sigma < 1/2 + 1/4$ and $\ell \ge 4$ :
$\log \mathbb{E}[|\zeta_X(\sigma)|^{2\ell}] \le (\ell^2 + \beta \ell) \log((\sigma - 1/2)^{-1})$
where $\beta = -2\sqrt{3} - 6 \log(1 - 1/\sqrt{3}) \approx 1.7032$ . This improves upon previous non-quantitative bounds.
GCD Sum Estimates (Theorem 3.1 & 3.3): The authors relate the variance of the pair correlation function to GCD sums. By applying the moment lemma to a random variable constructed from the representation function of the set $A$ , they derive an explicit upper bound for the GCD sum in terms of the additive energy and logarithmic factors.
Variance Analysis: Using Lemma 4.1 (from Aistleitner et al.), the variance of the pair correlation function is bounded by the GCD sum. The authors show that if the variance decays sufficiently fast (specifically, faster than $1/(\log N)^{1+\delta'}$ ), the MPPC property holds for almost all $\alpha$ via the Borel–Cantelli Lemma.

Key Contributions and Results

Explicit Lower Bound for $C$ : The main result (Theorem 1.1) states that if the additive energy satisfies $E(A_N) \le N^3 / (\log N)^C$ with $C \ge 14.71$ , then the sequence $(a_n)$ has MPPC.
Derivation of the Constant: The value $14.71$ is derived from the inequality:
$C - \beta - 4 \left\lceil \frac{\sqrt{2C + 1 + 10^{-18}}}{2} \right\rceil > 1$
where $\beta \approx 1.7032$ . The authors note that $14.71$ is the smallest value satisfying this specific inequality derived from their method.
Refinement of Bloom and Walker's Result: While Bloom and Walker conjectured $C > 1$ , this paper provides the first concrete threshold ( $C \ge 14.71$ ) proven via the additive energy framework.
Application to Specific Sequences: As a corollary, the authors show that the sequence $([n(\log n)^K])$ has MPPC for any $K \ge 15.71$ , based on the known additive energy bounds for such sequences.

Significance and Claims
The paper claims to provide the first explicit lower bound for the exponent $C$ in the additive energy condition for MPPC. The authors are modest regarding the optimality of this bound. They explicitly state that their method relies on the specific value of $\beta$ derived in Lemma 2.1. Even if the constant $\beta$ could be improved to 1, the resulting lower bound for $C$ would only decrease by approximately $0.7032$.

Consequently, the authors argue that the current bound of $14.71$ cannot be significantly improved using their specific approach. They suggest that bridging the gap between $14.71$ and the conjectured value of $1 + \epsilon$ would likely require a fundamentally different approach rather than refinements of the current moment and GCD sum techniques. The work serves to quantify the limits of the current methodology in relating additive energy to metric pair correlation.

Technical Summary: Metric Poissonian Pair Correlation and Additive Energy

More like this