Sums related to Euler's totient function

This paper establishes an upper bound for the sum of powers of the ratio between positive integers and their Euler's totient function values under specific constraints, and applies this result to derive an upper bound for the count of integers in a given range where this ratio exceeds a specified threshold.

The Gist

Imagine you are a mathematician trying to understand the "personality" of numbers. Specifically, you are looking at a famous function called Euler's Totient Function (let's call it $\phi$).

Think of $\phi(n)$ as a "popularity score" for a number $n$. It counts how many smaller numbers are "friends" with $n$ (meaning they share no common factors other than 1).

• If $n$ is a prime number, it's very popular; almost everyone is its friend. So $\phi(n)$ is high, and the ratio $n/\phi(n)$ is close to 1.
• If $n$ is a number with many small prime factors (like 12, which is $2 \times 2 \times 3$), it has fewer friends.$\phi(n)$is low, and the ratio$n/\phi(n)$ gets very big.

The Big Question:
How often does this ratio get really huge? And if we add up these ratios for a bunch of numbers, how big does the total sum get?

This paper by Artyom Radomskii is like a detective report that answers these questions with some very tight, precise limits. Here is the breakdown using simple analogies.

---

1. The Main Discovery: The "Crowded Party" Analogy

Imagine a party where every guest is a number. Some guests are "normal" (primes), and some are "chaotic" (numbers with many small factors).

• The ratio $n/\phi(n)$ is like a "chaos meter." A low meter means the guest is calm; a high meter means they are causing a scene.
• The author wants to know: If we take a list of numbers (maybe generated by a polynomial like $n^2+1$), how "chaotic" can the whole group get?

The Result:
The paper proves that even if you pick a huge list of numbers, the total "chaos" (the sum of these ratios raised to a power) cannot explode out of control. It grows, but it grows very slowly—much slower than you might expect.

It's like saying: "Even if you invite a million people to a party, the total noise level won't reach the sound of a jet engine; it will stay within a manageable, predictable range."

2. The "Rare Event" Guarantee

The paper also looks at the extreme cases. What if we ask, "How many numbers have a chaos meter so high that it's practically impossible?"

The author proves that these "super-chaotic" numbers are incredibly rare.

• The Analogy: Imagine looking for a person who is 10 feet tall. You might find one in a million. But if you look for someone who is 100 feet tall, you will find none.
• The Math: The paper shows that the number of integers where the ratio is huge drops off so fast it's almost zero. It uses a "double exponential" drop-off (like $e^{-e^t}$), which is a fancy way of saying "it disappears almost instantly."

3. The Tools: The "Sieve" and the "Filter"

To prove this, the author uses a technique called the Sieve Method.

• The Analogy: Imagine you have a bag of mixed nuts (numbers). You want to count how many are "bad" (have many small factors). You use a sieve (a filter) to separate them.
• The author builds a very specific, custom-made sieve. Instead of just filtering out primes, this sieve filters based on how the numbers interact with polynomials (like $n^2+1$) and linear functions.
• The paper refines an old sieve (used by a mathematician named Maynard) to make it sharper, allowing for higher powers ($s$) and more complex number patterns.

4. The Applications: Polynomials and Primes

The paper isn't just about random numbers; it applies to specific patterns:

• Polynomials: If you take a formula like $f(n) = n^2 + 1$ and plug in numbers $1, 2, 3...$, the results are usually "well-behaved." The paper proves that even for these complex formulas, the "chaos" stays under control.
• Primes: It also looks at what happens if you only feed prime numbers into these formulas. The result is the same: the chaos is predictable and rare.

5. Why Does This Matter?

You might ask, "Who cares about the sum of these ratios?"

• In the Real World: This isn't just abstract math. These sums appear in cryptography (how we keep internet data safe) and in the study of prime number distribution.
• The "Maynard" Connection: The paper improves upon work by James Maynard (a Fields Medalist). Maynard found a way to prove things about prime numbers, but his method had a "loose end" (a slightly weaker bound). Radomskii tightened that loose end, making the mathematical "net" stronger and more precise.

Summary in One Sentence

This paper proves that for almost any sequence of numbers you generate (using polynomials or primes), the "chaos" caused by numbers with many small factors is strictly limited, and extreme outliers are so rare they are practically non-existent.

The Takeaway:
Mathematics often deals with the unpredictable, but this paper shows that even in the wild world of number theory, there are strict, predictable rules governing how "messy" things can get.

Technical Summary

1. Problem Statement

The paper investigates the distribution and upper bounds of sums involving the ratio $\frac{n}{\phi(n)}$, where $\phi$ is Euler's totient function. Specifically, the author seeks to bound the sum:
$$\sum_{n \le N} \left( \frac{a_n}{\phi(a_n)} \right)^s$$
where $a_1, \dots, a_N$ are positive integers (not necessarily distinct), $s \in \mathbb{N}$ is an exponent, and the sequence $a_n$ satisfies specific distributional constraints modulo integers.

The core difficulty lies in the erratic behavior of $\frac{n}{\phi(n)}$, which is large when $n$ has many small prime factors. The paper aims to establish sharp upper bounds for these sums and derive tail estimates for the number of integers $n$ where $\frac{a_n}{\phi(a_n)}$ exceeds a threshold $t$.

2. Methodology

The author employs a combination of analytic number theory techniques, specifically:

• Sieve Methods: The proofs rely heavily on sieve theory to estimate the number of integers in a sequence divisible by certain moduli.
• Multiplicative Function Theory: The ratio $\frac{n}{\phi(n)}$ is expressed as a product over prime divisors. The author utilizes properties of multiplicative functions to decompose the sum.
• Lemma 3.1 (Key Inequality): A crucial tool is a lemma (cited from previous work) stating that for $\alpha \in (0, 1]$:
  $$\frac{n}{\phi(n)} \le \frac{c}{\alpha} \prod_{p|n, p \le (\log n)^\alpha} \left(1 + \frac{1}{p}\right)$$
  This allows the author to truncate the product to small primes, effectively reducing the problem to sums over square-free integers with small prime factors.
• Combinatorial Decomposition: The sum is expanded into a sum over divisors $d_1, \dots, d_s$. By grouping terms based on their least common multiple (LCM), the author defines a multiplicative function $f(n)$ and utilizes the condition $\omega(n) \le K g(n)$, where $\omega(n)$ counts the number of $a_n$ divisible by $n$.
• Polynomial and Linear Sieves: For specific applications (Theorems 1.3, 1.4, 1.5), the author applies the Brun-Titchmarsh inequality and bounds on the number of roots of polynomials modulo $m$ (Lemma 4.1) to estimate $\omega(n)$ for sequences generated by polynomials and linear forms.

3. Key Contributions and Results

General Theorems (1.1 & 1.2)

• Theorem 1.1: Establishes a general upper bound for $\sum (a_n/\phi(a_n))^s$ under the assumption that the count of $a_n$ divisible by $n$ (denoted $\omega(n)$) is bounded by $K g(n)$, where $g$ is a multiplicative function. The bound involves a product over primes up to a parameter $y$.
• Theorem 1.2: Refines Theorem 1.1 for the case where $K = \gamma N$ and $g(p)$ is bounded. It provides:
  1. A moment bound: $\sum (a_n/\phi(a_n))^s \le \exp(s \log \log(s+2) + Cs) N$.
  2. A tail bound: The number of $n$ such that $a_n/\phi(a_n) > t$ is bounded by $c_1 \exp(-\exp(c_2 t)) N$.
  • Significance: The tail bound shows that the ratio $a_n/\phi(a_n)$ is extremely unlikely to be very large, decaying doubly exponentially.

Applications to Polynomials (Theorem 1.3)

• Result: For a polynomial $f(n)$ with integer coefficients, the sum $\sum_{n \le x} (f(n)/\phi(f(n)))^s$ is bounded by $\exp(s \log \log(s+2) + Cs)x$.
• Improvement: This improves upon a previous result (Corollary 1.1 in [4]) which had a weaker bound of $\exp(s \log s + Cs)x$. The new bound $\log \log(s+2)$ is significantly tighter for large $s$.
• Tail Estimate: The paper also provides the first tail estimate for polynomial sequences, showing the count of $n$ where $f(n)/\phi(f(n)) > t$ is negligible.

Applications to Linear Forms and Sieves (Theorem 1.4)

• Context: This addresses sums arising in modern sieve methods, specifically sums of the form $\sum \frac{\Delta_L}{\phi(\Delta_L)}$, where $\Delta_L$ relates to a set of linear forms.
• Result: The author proves bounds for sums involving $\left(\frac{a^{k+1}|f(b)|}{\phi(a^{k+1}|f(b)|)}\right)^s$ over a range of $b$.
• Extension: This extends a result by Maynard (Lemma 8.1 in [3]) from $s=1$ to arbitrary $s \in \mathbb{N}$ and improves upon bounds in [4] by removing factorial terms ($s!$) from the upper bound, replacing them with logarithmic terms.

Applications to Primes (Theorem 1.5)

• Result: Theorems 1.3 and 1.2 are adapted to the case where the sequence is restricted to prime numbers ($p \le x$).
• Bound: $\sum_{p \le x} (f(p)/\phi(f(p)))^s \le \exp(s \log \log(s+2) + Cs)\pi(x)$.
• Significance: This confirms that the distributional properties of the totient ratio hold even when the input sequence is the set of primes, a non-trivial result requiring the Brun-Titchmarsh inequality.

4. Significance

• Optimization of Bounds: The paper achieves a significant improvement in the dependence on the exponent $s$. By replacing the $\log s$ term in the exponent with $\log \log s$, the bounds become much sharper for large moments.
• Tail Estimates: The derivation of the double-exponential decay bound ($\exp(-\exp(c_2 t))$) is a major contribution, providing a rigorous quantification of how "rare" it is for the ratio $n/\phi(n)$ to be large in these specific sequences.
• Unification: The framework unifies the treatment of general sequences, polynomial sequences, and prime sequences under a single set of sieve-theoretic conditions.
• Sieve Theory Applications: The results directly support modern sieve methods (e.g., in the study of prime gaps or bounded gaps between primes) by providing precise control over the error terms involving the totient function, which often appear in the denominators of sieve weights.

In summary, Radomskii provides a robust analytical framework for bounding sums of powers of the totient ratio, offering improved asymptotic bounds and new tail estimates that are applicable to a wide range of problems in analytic number theory.