Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling

DOI: 10.54254/2753-8818/53/20240201

Riemann Zeta Function and Its Applications

Yuxuan Shi1,3,*, Zhanyu Lin2,4

1
School of Mathematics and Physics, Xi’an Jiangtong-Liverpool University, Suzhou,
215123, China
2
School of Mathematics and Applied mathematics, Chongqing University, Chongqing,
400044, China

3
[email protected]
4
[email protected]
*corresponding author

Abstract. This paper focuses on the introduction and proof of the fundamental properties of 𝜻(𝒛),
i.e. Riemann zeta function and explores its applications in algebra. We begin with a systematic
derivation and proof of the basic characteristics of the zeta function. Following this, we examine
its application in algebra, including the use of the Dirichlet L-function to prove Dirichlet’s
theorem. Furthermore, we show the classical result for the subgroup growth rate of 𝓙-groups and
the enumeration of n-dimensional irreducible representations of Heisenberg groups.

Keywords: Riemann’s zeta function, number theory, group zeta function.

1. Introduction
The discovery of zeta function can be traced back to around 1350, French mathematician Nicole Oresme
1
identified the divergence of harmonic series.∑∞𝑘=1 𝑘
Several methods can be employed to prove this result:
11
Riemann sum: Transform the series into ∫0 𝑥 𝑑𝑥
Scaling:
∞
1 1 1 1 1 1 1 1
∑ =1+ +( + )+( + + + )+⋯
𝑛 2 3 4 5 6 7 8
𝑛=1
1 1 1 1 1 1 1
≥1+ +( + )+( + + + )+⋯
2 4 4 8 8 8 8
1 1
= 1 + + + ⋯ = +∞
2 2
Scaling again:

© 2024 The Authors. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0
(https://creativecommons.org/licenses/by/4.0/).

191
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

∞
1 1 1 1 1 1
∑ = (1 + + ⋯ + ) + ( + + ⋯+ ) + ⋯
𝑛 2 9 10 11 99
𝑛=1
1 1 1 1 1 1
>(+ + ⋯+ ) + ( + +⋯+ )+⋯
10 10 10 100 100 100
9 9
= + + ⋯ = +∞
10 10
Using the inequality 𝑥 > ln(𝑥 + 1):
1
In 1644, Pietro Mengoli posed the Basel problem: calculating the sum ∑∞𝑛=1 𝑛2 .It was first addressed
by Euler in 1735 using identity
∞
𝑠𝑖𝑛𝑥 𝑥2 𝑥4 𝑥2
= 1 − + + ⋯ = ∏ [1 − ]
𝑥 3! 5! (𝑛𝜋)2
𝑛=1
Compare the coefficients of the term 𝑥 2 , it follows that
∞
1 𝜋2
∑ 2=
𝑛 6
𝑛=1
Moreover, Parseval’s identity can be applied to solve this problem:
∞
1 𝜋
∑ |𝐶𝑛 |2 = ∫ |𝑓(𝑥)|2 𝑑𝑥
2𝜋 −𝜋
𝑛=−∞
1 𝜋
where 𝐶𝑛 = ∫ 𝑓 (𝑥)𝑒 −𝑖𝑛𝑥 𝑑𝑥
2𝜋 −𝜋
denotes the Fourier coefficient.
∞
sin𝑘𝑥
𝑓(𝑥) = 𝑥 = 2 ∑(−1)𝑘+1
𝑘
𝑘=1
1 1 1 𝜋 2𝜋2 1 𝜋2
We obtain 4 (1 + + + ⋯ ) = ∫leading to the solution ∑∞
2
𝑥 𝑑𝑥 = , 𝑛=1 𝑛2 =
22 32 𝜋 −𝜋 3 6
The Basel problem can also be addressed using complex analysis. Let
𝜋
2
𝐼 = ∫ ln(2cosx) 𝑑𝑥
0
Since 2cos𝑥 = 𝑒 𝑖𝑥 + 𝑒 −𝑖𝑥 ,
𝜋 𝜋 𝜋
2
𝑖𝑥 −𝑖𝑥
2
𝑖𝑥 −2(𝑖𝑥)
𝑖𝜋 2 2
𝐼 = ∫ ln (𝑒 +𝑒 ) = ∫ ln [𝑒 (1 + 𝑒 )] = + ∫ ln (1 + 𝑒 −2(𝑖𝑥) )𝑑𝑥
0 0 8 0
Expanding ln(1 + 𝑒 −2𝑖𝑥 ) into a Taylor series, we have
𝜋 ∞
2
−2𝑖𝑥
1 1 3
∫ ln (1 + 𝑒 )= ∑ 2
= 𝜁(2)
0 𝑖 (2𝑘 + 1) 4𝑖
𝑘=0
Consequently, we find
𝜋2 3
𝐼 = 𝑖 ( − 𝜁(2))
8 4
Since 𝐼 is a real number, it follows that
𝜋2 3 𝜋2
( − 𝜁(2)) = 0 ⇒ 𝜁(2) =
8 4 6
Building on the divergent harmonic series and the classical Bessel problem, we now turn our attention
to the research of zeta function:
Definition 1.1. 𝜁(𝑧) = ∑∞ 𝑛=1 𝑛
−𝑧

The zeta function not only generalizes the concept of series but also encapsulates profound
connections between various mathematical fields, including number theory and algebra. In the following

192
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

chapters, we will first prove the preliminary properties of the zeta function, which were initially
proposed by Riemann and have since been rigorously formalized by generations of mathematicians. We
will also explore the applications of the zeta function in series calculation and algebra.

2. Basic Property of Riemann Zeta function

When discussing the 𝜁(𝑧), an essential characteristic is the Euler Product Formula. Euler initially
discovered that 𝜁(𝑧) can be represented as a product of factors corresponding to prime numbers.
1 −1
Proposition 2.1. For 𝑧 > 1, 𝜁(𝑧) = ∏p is prime (1 − 𝑝𝑧)
Proof.
1 −1 1 −1 1 −1
∏p is prime (1 − 𝑧) = (1 − 2𝑧) (1 − 3𝑧) ⋯
𝑝
1 1 1 1
= (1 + 2𝑧 + 22𝑧 + ⋯ ) (1 + 3𝑧 + 32𝑧 + ⋯ ) ⋯
1
= ∏p is prime ∑∞
𝑖=0 𝑝𝑖𝑧
𝑎 𝑎 1 1 1 1
∀𝑁 ∈ ℤ, let 𝑁 = 𝑝1 1 ⋯ 𝑝𝑛 𝑛 . Then 𝑁𝑧 = ∏𝑛𝑖=1 𝑎 , ∑𝑁≥1 𝑁𝑧
= ∏p is prime ∑∞
𝑖=0 𝑝𝑖𝑧
𝑝𝑖 𝑖
In fact, this formula can be generalized to apply to multiplicative functions:
Theorem 2.2. (Smith, 1983) Consider 𝐿(𝑥) ∈ ℓ1 that respects the property 𝐿(𝑥𝑦) = 𝐿(𝑥)𝐿(𝑦), then
∑∞ 2
𝑥=1 𝐿 (𝑥) = ∏p is prime (1 + 𝐿(𝑝) + 𝐿(𝑝 ) + ⋯ )
Proposition 2.1 gives an approximate characterization for the roots of 𝜁(𝑧). If 𝜌 is a root for 𝜁(𝑧),
1 −1
then for some prime 𝑞, (1 − ) = 0, which implies that Re(𝑧) ≤ 1. In other words, the function
𝑞−𝑧
𝜁(𝑧) is free of zeron in the half-plane where Re(𝑧) > 1.
Conversely, by taking the logarithm of both sides, we derive an alternate expression of the Euler
Product Formula:
ln𝑝, 𝑛 = 𝑝𝑘 1 𝜁 ′ (𝑧)
Corollary 2.3. Let 𝛬(𝑛) = { , then∑∞
𝑛=1 𝛬 (𝑛) 𝑛𝑧 = − 𝜁(𝑧)
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Proof. Applying a logarithmic transformation followed by differentiation, it follows that
𝜁 ′ (𝑧) ln𝑝 1
− 𝜁(𝑧)
= ∑p is prime 𝑝𝑧 −1
= ∑p is prime ∑∞
𝑗=1(ln𝑝) 𝑝𝑗𝑧
1
= ∑∞
𝑗=1 ∑p is prime (ln𝑝) 𝑝𝑗𝑧
1
= ∑∞
𝑗=1 𝛬 (𝑗) 𝑗𝑧
Understanding 𝜁(𝑧) near 𝑧 = 1 is crucial, as 𝜁(1) diverges to infinity. Additionally,
𝜁(𝑧)
Proposition 2.4. lim (𝑧−1)−1 = 1
𝑧→1
Proof. Since the function 𝑡 −𝑧 decreases monotonically as 𝑡 increases for a fixed z, it follows that
1 𝑛+1 1 1 ∞ 1
(𝑛+1)𝑧
< ∫𝑛 𝑡 𝑧 𝑑𝑡 < 𝑛𝑧 By summing this inequality over 1 to ∞, 𝜁(𝑧) − 1 < ∫1 𝑡 𝑧 𝑑𝑡 < 𝜁(𝑧) Since
∞ 1 𝜁(𝑧)
∫1 𝑡𝑧
𝑑𝑡 = (𝑧 − 1)−1 ⇒ 1 < (𝑧−1)−1 < 𝑧. As 𝑧 approaches 1, we find that lim𝑧→1 (𝑧 − 1)𝜁(𝑧) = 1.
Through some transformations and calculations, we can obtain another form of this proposition:
ln𝜁(𝑧)
Corollary 2.5. lim −ln(𝑧−1) = 1
𝑧→1
Typically, the Riemann zeta function is not used with 𝑧 as a real number where 𝑧 > 1, but rather by
treating 𝑧 as a complex number and considering the region where 𝑅𝑒(𝑧) ≤ 1. Consequently, extending
𝜁(𝑧) analytically to cover the entire complex plane is crucial, as highlighted in:
Theorem 2.6. 𝜁(𝑧) can be extended to the entire ℂ. Its only singularity is at 𝑧 = 1, where the residue
is 1

193
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

The elementary techniques employed in this proof are partial summation and the Stieltjes integral.
The former transforms sums into more manageable sums or integrals, while the latter does the opposite
by converting sums into integrals. One can find the proof in Book (lvic, 1949)
The analytic extension of 𝜁(𝑧) is crucial for exploring the distribution of its zeros. J. Hadamard once
proved that 𝜁(𝑧) does not vanish on {𝑧 ∈ ℂ|Re(𝑧) = 1}:
Proposition 2.7. ∀𝑦 ∈ ℝ, 𝜁(1 + 𝑖𝑦) ≠ 0
Proof.
Assuming that 1 + 𝑖𝑦 is a m-order zero. Theorem 7 implies that 1 + 𝑖𝑦 is a first-order pole with
𝜁 ′ (𝑧)
residue 𝑚 ≥ 1 for the function . Consequently, for 𝛼 > 1 but sufficiently close to 1, we obtain
𝜁(𝑧)
𝜁 ′ (𝛼+𝑖𝑦) 𝑚 1 −1 −1 𝜁 ′ (𝛼)
𝜁(𝛼+𝑖𝑦)
= 𝛼−1 + 𝑜 (𝛼−1) Given that 1 is a pole of 𝜁(𝑧) with residue 1, we get = 𝛼−1 + 𝑜 (𝛼−1) 𝜁(𝛼)
𝜁 ′ (𝛼+2𝑖𝑦) 𝑘 𝑘
Let = + 𝑜 ( ) Where 𝑘 = 0 if 𝜁(1 + 2𝑖𝑦) ≠ 0, 𝑘 ≥ 1 if 𝜁(1 + 2𝑖𝑦) = 0. Therefore,
𝜁(𝛼+2𝑖𝑦) 𝛼−1 𝛼−1
𝜁 ′ (𝛼+2𝑖𝑦) 3𝜁 ′ (𝛼) 4𝜁 ′ (𝛼+𝑖𝑦) 𝑘−3+4𝑚 𝑘−3+4𝑚
𝑅𝑒 ( 𝜁(𝛼+2𝑖𝑦) + 𝜁(𝛼) + 𝜁(𝛼+𝑖𝑦) ) = 𝛼−1 + 𝑜 ( 𝛼−1 ) > 0 On the other hand, by 4,
𝜁 ′ (𝛼+2𝑖𝑦) 3𝜁 ′ (𝛼) 4𝜁 ′ (𝛼+𝑖𝑦) 1
𝑅𝑒 ( 𝜁(𝛼+2𝑖𝑦) + 𝜁(𝛼) + 𝜁(𝛼+𝑖𝑦) ) = − ∑∞ 𝑗=1 𝛬 (𝑗) 𝑗 𝛼 (cos(2𝑡ln𝑗) + 3 + 4cos(𝑡ln𝑗))
1
= − ∑∞ 2
𝑗=1 2 𝛬(𝑗) 𝑗 𝛼 (1 + cos(𝑡ln𝑗) ) ≤ 0
This is contradiction. Thus 𝜁(1 + 𝑖𝑦) ≠ 0 for all 𝑦 ∈ ℝ.
In the next part, we will introduce a symmetric form of the zeta function based on its functional
equations. These are fundamental results in zeta function theory, originally proven by Riemann. The
three equivalent functional equations are given by:
1 𝑧
Proposition 2.8. (cf. (Smith, 1983), Theorem 1.6) For all complex 𝑧, let 𝑍(𝑧) = 𝑧/2 𝛤 ( ) 𝜁(𝑧), then
𝜋 2
𝑍(𝑧) = 𝑍(1 − 𝑧)
This equation for the zeta function implies that ∀𝑠 ∈ ℂRe(𝑠)<0 , 𝑠 is not a root for 𝜁(𝑧). Because if 𝜌
is a root of 𝜁(𝑧), then 1 − 𝜌 is also a root. In summary, we conclude that the strip 0 ≤ Re(𝜌) < 1
contain every nontrivial zero of 𝜁(𝑧)
Next, we will explore specific values of 𝜁(𝑧) and examine how to compute series associated with it.
The zeta function’s values at positive even integers can be determined using the recurrence relation
provided below:
(−1)𝑘+1 𝜋2𝑘 𝜁(2𝑛−2𝑘)
Proposition 2.9. 𝜁(2𝑛) = ∑𝑛𝑘=1 (2𝑘+1)!
The validity of this formula can be easily demonstrated with mathematical induction.Or we can just
sin𝑥
compare the coefficients of the terms in the Taylor series and product expansion of 𝑥 and obtain it by
inference.
Zeta function’s value at negative integers can be obtained using the following expression of zeta
function(One can prove it readily by proving that the definite integral in the formula equals 𝜁(𝑧)𝛤(𝑧)):
1 ∞ 𝑥 𝑧−1
Proposition 2.10. 𝜁(𝑧) = ∫ 𝑑𝑥
𝛤(𝑧) 0 𝑒 𝑥 −1
𝐵
Corollary 2.11. 𝜁(−𝐾) = (−1)𝐾 𝐾+1
∈ ℤ+ 𝐵𝐾 denotes the Bernoulli number
𝐾+1
, ∀𝐾
Ramanujan also offered a simple method to calculate 𝜁(−𝐾),that is: Let 𝐴 = 1𝐾 + 2𝐾 + ⋯ and 𝐵 =
1 − 2𝐾 + ⋯, then A adds B and B adds itself by dislocation for a few times,but in contrast, the
𝐾

aforementioned method is much more persuasive.

The following results from Aries in Zhihu shows the series related to zeta function:
1 𝜋
Proposition 2.12. ∑∞𝑘=1 𝑥
2𝑘
𝜁(2𝑘) = − 𝑥cot𝜋𝑥
2 2
Proof.
∞ ∞
sin𝜋𝑥 𝑥2 𝑥𝑛
= ∏ (1 − 2 ) ⇒ lnsin𝜋𝑥 − ln𝜋𝑥 = ∑ ln (1 − 2 )
𝜋𝑥 𝑛 𝑛
𝑛=1 𝑛=1

194
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

Take the derivative, we have

∞ ∞
2𝑘
1 𝜋
𝜋𝑥cot𝜋𝑥 = 1 − 2 ∑ 𝑥 𝜁(2𝑘) ⇒ ∑ 𝑥 2𝑘 𝜁(2𝑘) = − 𝑥cot𝜋𝑥
2 2
𝑘=1 𝑘=1
𝜁(2𝑘)
1 ∞ 𝜁(2𝑘) 1 𝜋
Corollary 2.13. ∑∞
𝑘=1 ∑
= 2 , 𝑘=1 16𝑘 = 2 − 8
4𝑘
1 1
Proof. Just take x = 2 and x = 4 in Proposition 2.13, we will arrive at the answer.
𝜁(4𝑘) 𝜋 𝜋 𝜋
Corollary 2.14. ∑∞ 𝑘=1 4 𝑘 = 4√2 (cot √2 + coth √2)
Proof. Substitute 𝑖𝑥 for 𝑥 in Proposition 13, we have
∞
1 𝜋
∑(−1)𝑘 𝑥 2𝑘 𝜁(2𝑘) = − 𝑥coth𝜋𝑥
2 2
𝑘=1
1
Adding 2.12 and [4], then taking 𝑥 = , we have
√2
∞
𝜁(4𝑘) 𝜋 𝜋 𝜋
∑ 𝑘
= (cot + coth )
4 4√2 √2 √2
𝑘=1
3
Corollary 2.15. ∑∞𝑘=1[𝜁(2𝑘) − 1] = 4 (a)
Proof. Using the identity
∞
𝑥2
∑ 𝑥 2𝑘 =
1 − 𝑥2
𝑘=1
Then By Proposition 13 and take the limit 𝑥 → 1, we can prove the result. ◻
1
Corollary 2.16. ∑∞
𝑘=1[𝜁(2𝑘) − 𝜁(2𝑘 + 1)] = 2
1
∑∞
𝑘=1[𝜁(2𝑘 + 1) − 1] =
4
∑∞
𝑘=1[𝜁(𝑘) − 1] = 1
Proof. Denote the three series with (b),(c) and (d) from top to bottom. (b) is quite simple, we just
1
calculate it directly with 𝜁(𝑧) = ∑∞ 𝑛=1 𝑛𝑧 (c)=(a)-(b) and (d)=(a)+(c)
Following are the amazing conclusions which demonstrate that zeta function is also related to
trigonometric functions.
cos𝑛 𝜋 𝑖 𝜁(2𝑘−2𝑎)
∑∞ 𝑘 ∞
𝑛=1 𝑛2𝑘 = (−1) 2(2𝑘−1)! + ∑𝑎=0(−1) (2𝑎)!
Proposition 2.17. sin𝑛 𝜋 𝜁(2𝑘−2𝑎)
∑∞𝑛=1 2𝑘+1 = (−1)
𝑘
+ ∑∞ 𝑖
𝑎=0(−1) (2𝑎+1)!
𝑛 2(2𝑘)!
Obviously, the formula of Fourier Series is a nice proof for the aforementioned two series. We can
also take several fixed positive integer k and it will be quite easy to conclude the formula by inference.

3. Dirichlet Theorem
This section aims to prove the Dirichlet Theorem by the technique of zeta function:
Theorem 3.1. Assume 𝑎 and 𝑚 are coprime integers. Define 𝑃(𝑎, 𝑚) as the set of prime numbers 𝑝
that satisfy 𝑝 ≡ 𝑎 𝑚𝑜𝑑 𝑚. Then 𝑃(𝑎, 𝑚) contains infinitely many elements.
Based on the generalization of the zeta function, we can approach the proof. In general, for an
arithmetic function, the Dirichlet L-function associated with 𝜆 is defined by
∞

𝐿(𝑧, 𝜆) = ∑ 𝜆 (𝑛)𝑛−𝑧
𝑛=1
Proposition 2.1 holds for 𝐿(𝑧, 𝜆) if 𝜆 is multiplicative, we have
Proposition 3.2. Let 𝐿(𝑧, 𝜆) be the Dirichlet L-function with respect to 𝜆, then 𝐿(𝑧, 𝜆) = ∏p is prime
1 −1
(1−𝜆(𝑝)𝑝−𝑧 )

195
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

More importantly, we investigate the properties of ln𝐿(𝑧, 𝜆). By taking the logarithm of the above
expression, we obtain:
1
Proposition 3.3. ln𝐿(𝑧, 𝜆) = ∑p is prime 𝜆 (𝑝) 𝑝𝑧 + 𝑅(𝑧) Where 𝑅(𝑧) remains bounded as 𝑧 → 1
1
Proof. Only need to prove ln𝜁(𝑧) = ∑p is prime 𝑝𝑧
+ 𝑅(𝑧) By proposition 2, Let 𝜁(𝑧) = ∏𝑝≤𝑀
1 −1
(1 − 𝑝𝑧) 𝜆𝑀 (𝑧), 𝜆𝑀 (𝑧) → 1, 𝑀 → ∞
1 1
So ln𝜁(𝑧) = ∑𝑝≤𝑀 − ln (1 − 𝑝𝑧) + ln𝜆𝑀 (𝑧) Since 𝑝−𝑧 < 1 , we have ln (1 − 𝑝𝑧) =
1
− ∑∞
𝑛=1 Thus as 𝑀 approaches infinity,
𝑛𝑝−𝑛𝑧
1 1 1 1 1
ln𝜁(𝑧) = ∑𝑝 ∑∞
𝑘=1 𝑘 𝑝𝑘𝑧 = ∑𝑝 ∑∞
𝑘=1 𝑘𝑝𝑘𝑧 = ∑𝑝 + ∑𝑝 ∑∞
𝑘=2 𝑘𝑝𝑘𝑧
𝑝
1 1 −1 1 𝜋2
Therefore, ln𝜁(𝑧) ∼ ∑𝑝
= ∑𝑝 (1 − 𝑧) ≤ (1 − 2−𝑧 )−1 ∑𝑝 ≤
𝑝2𝑧 𝑝 𝑝2𝑧 3
1
𝑝𝑧
Since 𝜆 is a character of ℤ/𝑚ℤ, for 𝑝 ∣ 𝑚, 𝜆(𝑝) = 0. Therefore the above product is equals to

ln𝐿(𝑧, 𝜆) = ∑ 𝜆 (𝑝)𝑝−𝑧 + 𝑅(𝑧)

𝑝∤𝑚
From Proposition 3.2 and Corollary 2.5, we can define a meaningful quantity to characterize the
density of a specific set of prime numbers:
∑𝑝∈𝒫 𝑝−𝑧
Definition 3.4. Assume 𝒫 is a set of positive primes. If the limit 𝑑(𝒫) : = 𝑙𝑖𝑚 exists, we
𝑧→1 −𝑙𝑛(𝑧−1)
called 𝒫 has Dirichlet density.
The basic fact of Dirichlet density is
0, 𝒫 is limited
𝑑(𝒫) = {
1, 𝒫 consists of all but finitely many positive primes
In that cases, Theorem 3.1 is equivalent to 𝑑(𝒫(𝑎, 𝑚)) > 0.
Now we can prove Theorem 19:
Proof. Let 𝐿(𝑧, 𝜆) be the Dirichlet L-function with respect to 𝜆, where 𝜆 is a character of ℤ/𝑚ℤ. Since
|𝜆(𝑛)| = 1 ⇒ |𝜆(𝑛)𝑛−𝑠 | ≤ 𝑛−𝑠 , 𝐿(𝑧, 𝜆) converges and continuous for 𝑧 > 1. Now define 𝐺(𝑧, 𝜆) =
1 1
∑q is prime ∑𝑘=1 ( ) 𝜆(𝑞 𝑘 ) 𝑘𝑧 It also converges and continuous for 𝑧 > 1 . Additionally, for any
𝑘 𝑞
1 1
complex number 𝑠 with |𝑠| < 1, we have exp (∑∞ 𝑘
𝑘=1 𝑘 𝑠 ) = (1 − 𝑠)
−1
Substituting 𝑠 = 𝜆(𝑞) 𝑞𝑧 , we
1 1 1 −1
obtain exp (∑∞ 𝑘
𝑘=1 𝑘 𝜆(𝑞 ) 𝑞𝑘𝑧 ) = (1 − 𝜆(𝑞) 𝑞𝑧 ) Therefore, exp𝐺(𝑧, 𝜆) = 𝐿(𝑧, 𝜆). By Proposition 21,
1
𝐺(𝑧, 𝜆) = ∑𝑞∤𝑚 𝜆 (𝑞) 𝑞𝑧 + 𝑅𝜆 (𝑧) where lim𝑧→1 𝑅𝜆 (𝑧) < ∞ . Through some transformations and
1
calculations, we have ∑𝜆 𝜆(𝑎) 𝐺(𝑧, 𝜆) = ∑𝑞∤𝑚 ∑ 𝜆(𝑎) 𝜆(𝑞) + ∑𝜆 𝜆(𝑎) 𝑅𝜆 (𝑧)
𝑞𝑠 𝜆
By utilizing the
orthogonality of the characters(cf. (Kenneth Ireland, 2000) Proposition 16.3.2) ∑𝜆 𝜆(𝑎) 𝐺(𝑧, 𝜆) =
𝑅𝜆,𝑎 (𝑧) 𝐺(𝑧,𝜆0 )
𝜓(𝑚) ∑𝑞≡𝑎 mod 𝑚 𝑞 −𝑧 + 𝑅𝜆,𝑎 (𝑧) Dividing −ln(𝑧 − 1) on both sides, ln(𝑧−1) − ln(𝑧−1) −
∑𝜆≠𝜆0 𝜆(𝑎)𝐺(𝑧,𝜆) 𝜆,𝑎 𝑅 (𝑧)
ln(𝑧−1)
= 𝜓(𝑚)𝑑(𝒫(𝑎, 𝑚)) Since 𝑅𝜆,𝑎 (𝑧) is bounded as 𝑧 → 1 , lim ln(𝑧−1) = 0 Notice that
𝑧→1
𝐺(𝑧,𝜆 )
𝐺(𝑧, 𝜆0 ) = ∑𝜆∣𝑚 ln (1 − 𝜆−𝑧 ) + ln𝜁(𝑧). By Proposition 5, lim ln(𝑧−1)
0
= −1 It remains to prove that
𝑧→1
𝐺(𝑧, 𝜆) remains bounded as 𝑧 → 1 forall nontrival character 𝜆. It’s equivalent to prove 𝐿(𝑧, 𝜆) can be
extended to half complex plane and 𝐿(1, 𝜆) ≠ 0 forall nontrival character. The first assertion is similar
to Theorem 7. For the second assertion, summing over 𝜆 on both sides of 𝐺(𝑧, 𝜆) = ∑q is prime ∑𝑘=1
1 1
( ) 𝜆(𝑞 𝑘 )𝑞−𝑘𝑠 we have ∑𝜆 𝐺 (𝑧, 𝜆) = 𝜓(𝑚) ∑𝑞𝑘 ≡1 mod 𝑚 𝑞 −𝑘𝑧 Since RHS ≥ 0 , we take the
𝑘 𝑘
exponential of both sides: 𝐹(𝑧) : = ∏𝜆 𝐿 (𝑧, 𝜆) ≥ 1, ∀𝑧 > 1 . Now Firstly we prove that if 𝜆 is a

196
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

nontrivial complex character of ℤ/𝑚ℤ , then 𝐿(1, 𝜆) ≠ 0 :Assumed 𝐿(1, 𝜆) = 0 , then 𝐿(𝑧, 𝜆) =
𝐿(𝑧, 𝜆)implies 𝐿(1, 𝜆) = 0. Therefore 𝐹(1) = 0, contradiction!. Next we prove the nontrivial real
character 𝜆 also satisfies 𝐿(1, 𝜆) ≠ 0 : Since |𝜆(𝑛)| = 1, 𝜆(𝑛) ∈ ℝ ⇒ 𝜆(𝑛) = ±1, ∀𝑛 ∈ ℤ . Assume
𝐿(𝑧,𝜆)𝐿(𝑧,𝜆 ) 1
𝐿(1, 𝜆) = 0, We construct a function as follows:𝜙(𝑧) = 𝐿(2𝑧,𝜆 ) 0 𝜙(𝑧) is analytic on 𝑅𝑒 𝑧 > 2 and
0
1
𝜙(𝑧) → 0 as 𝑧 → 2 . Supposed 𝑧 is real and 𝑧>1 , then by Proposition 2,
(1−𝑞−2𝑧 ) 1+𝑞−𝑧
𝜙(𝑧) = ∏𝑞∤𝑚 (1−𝑞−𝑧)(1−𝜆(𝑞)𝑞−𝑧) = ∏𝜆(𝑞)=1 1−𝑞−𝑧 = (1 + 𝑞 −𝑧 )(∑∞
𝑘=0 𝑞
−𝑘𝑧
)
Using Theorem 3,
= 1 + 2𝑞 −𝑧 + 2𝑞 −2𝑧 + ⋯ +
𝜙(𝑧) = ∑∞
𝑛=1 𝑎𝑛 𝑛
−𝑧
converges for 𝑧 > 1 . Expanding 𝜓(𝑧) into power series at 𝑧 = 2 , we have
∞ 1
𝜙(𝑧) = ∑𝑚=0 𝑏𝑚 (𝑧 − 2)𝑚 Since 𝜙(𝑧) is analytic for 𝑅𝑒 𝑧 > , the radius of convergence of it is at least
2
3 𝜙(𝑚) (2)
2
.Since 𝑏𝑚 = 𝑚!
differentiate 𝜙(𝑠) m times it follows that 𝜙 (𝑚) (2) = ∑∞
, 𝑚 −2
𝑛=1 𝑎𝑛 (−ln𝑛) 𝑛
: = (−1)𝑚 𝑐𝑚 , 𝑐𝑚 ≥ 0 Thus 𝜙(𝑧) = ∑∞ 𝑚
𝑛=0 𝑐𝑚 (2 − 𝑧) with the coefficient 𝑐𝑚 nonnegative and 𝑐0 =
1
𝜙(2) = ∑∞ 𝑛=1 𝑎𝑛 𝑛
−2
≥ 𝑎1 = 1 It follows that for < 𝑧 < 2, 𝜙(𝑧) ≥ 1. This contradicts 𝜙(𝑧) → 0 as
2
1
𝑧→ 2
! Therefore, forall the nontrival character 𝜆, we have 𝐿(1, 𝜆) ≠ 0, so that 𝐺(𝑧, 𝜆) remains bounded
1
as 𝑧 → 1. Taking 𝑧 → 1 in [1], we obtain 1 = 𝜓(𝑚)𝑑(𝒫(𝑎, 𝑚)) ⇒ 𝑑(𝒫(𝑎, 𝑚)) = 𝜓(𝑚) > 0 Where
𝜓(𝑛) is the Euler’s totient function. Thus we finally prove the Theorem 19.

4. Counting Problem For Groups

Zeta-function is suitable for some counting questions: Let 𝛤 be a mathematical object, 𝑎𝛤 (𝑛) is a non-
negative integers sequence which encoded with some information about 𝛤. Thus 𝑛 ↦ 𝑎𝛤 (𝑛) is an
arithmetic functions. It’s natural to consider the Dirichlet L-function of 𝑎𝛤 (𝑛), which gives rise to the
following definition:
Definition 4.1. The zeta function of (𝛤, 𝑎𝛤 (𝑛)) is the Dirichlet generating series 𝜁𝛤 (𝑧) =
∞
∑ 𝑎𝛤 (𝑛)𝑛−𝑧
𝑛=1
Firstly let’s consider the number of subgroup with finite index of an infinite group. Let 𝛤 be a infinite
group. For a given 𝑛 ∈ ℤ, denoted 𝑎𝛤 (𝑛) = #{𝐻 ≤ 𝛤| |𝛤: 𝐻| = 𝑛}. Therefore the arithmetic function
𝑛 ↦ 𝑎𝛤 (𝑛) is called the subgroup number function. The partial sum 𝑆𝛤 (𝑛) = ∑𝑚≤𝑛 𝑎𝛤 (𝑛) is called the
subgroup growth functions. The subgroup zeta function is given by
∞

𝜁𝛤 (𝑧) = ∑ 𝑎𝛤 (𝑛)𝑛−𝑧 = ∑ |𝛤: 𝐻|−𝑧

𝑛=1 𝐻≤𝑓 𝛤
where the notation 𝐻 ≤𝑓 𝛤 indicates that the summation is over all subgroups with finite index. By
finding an explicit expression for this function, we can determine the properties of 𝑎𝛤 (𝑛) through
coefficient comparison or asymptotic analysis.
It’s better that 𝑛 ↦ 𝑎𝛤 (𝑛) has the following properties:
1. Polynomial growth: 𝑆𝑛 (𝛤) = 𝑂(𝑛𝑎 ) for some 𝑎 ∈ ℝ.
𝑒 𝑒
2. Multiplicativity: If 𝑛 = ∏𝑖 𝑝𝑖 𝑖 , then 𝑎𝛤 (𝑛) = ∏𝑖 𝑎𝛤 (𝑝𝑖 𝑖 ).
Property 1 ensures the convergence of 𝜁𝛤 (𝑧). Generally, the abscissa of convergence is a useful tools
to characterize the region of convergence, defined as
ln𝑆𝑛 (𝛤)
𝛼(𝑎𝛤 (𝑛)) = limsup
𝑛→∞ ln𝑛
Marcus du Sautoy Shows the relation between abscissa of convergence and region of convergnece:(cf.
(Sautoy & Grunewald, 2000))

197
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

Theorem 4.2. Consider a 𝒥 −group 𝛤. The abscissa of convergence 𝛼(𝑎𝛤 (𝑛)) ∈ ℚ. And 𝜁𝛤 (𝑧) is
analytically continued to the half plane 𝑅𝑒 (𝑧) > 𝛼(𝑎𝛤 (𝑛)) . Moreover, the line {𝑧 ∈ ℂ|𝑅𝑒(𝑧) =
𝛼(𝑎𝛤 (𝑛)) can have at most one pole of 𝜁𝛤 (𝑧) located at 𝑧 = 𝛼(𝑎𝛤 (𝑛))
Property 2 shows that Proposition 2 can be generalized to 𝜁𝛤 (𝑧):
∞

𝜁𝛤 (𝑧) = ∏ ∑ 𝑎𝛤 (𝑝𝑖 )𝑝−𝑖𝑧

𝑝 prime 𝑖=0
The factor 𝜁𝛤,𝑝 (𝑧) : = ∑∞ 𝑖 −𝑖𝑧
𝑖=0 𝑎𝛤 (𝑝 )𝑝 is called the local factor of 𝜁𝛤 (𝑧) at the prime 𝑝.
The next question is to identify the types of groups that satisfy these properties. Grunewald, Fritz J
and Scharlau, Rudolf has proved in (Grunewald & Scharlau, 1979) that torsion-free, finitely generated
nilpotent group, denoted as 𝒥 -group, exhibits polynomial subgroup growth and multiplicativity.
Consequently, the focus shifts to finding a general formula for the zeta function of −𝐽 group or
calculating specific cases.
Example 4.3. 𝜁ℤ𝑛 (𝑧) = ∏𝑛−1 𝑖=0 𝜁 (𝑧 − 𝑖)
Proof.
Lemma 26. Let 𝐺 be a group, 𝐺 = 𝐴 × 𝐵, 𝑄 ≤ 𝐵 ≤ 𝐿 ≤ 𝐺 , then there is 𝐻, 𝐻 ≤ 𝐺, 𝑠. 𝑡. 𝐻𝐵 =
𝐿, 𝐻 ∩ 𝐵 = 𝑄
Proof. Let 𝐻 = 𝐿 ∩ 𝑄𝐴, then
𝐻 ∩ 𝐵 = (𝐿 ∩ 𝑄𝐴) ∩ 𝐵 = 𝑄𝐴 ∩ 𝐵 = 𝑄(𝐴 ∩ 𝐵) = 𝑄
𝐻𝐵 = (𝐿 ∩ 𝑄𝐴)𝐵 = 𝐿 ∩ 𝑄𝐴𝐵 = 𝐿 ∩ 𝐺 = 𝐿
Lemma 27. Subject to the conditions of Lemma 26 and further supposes that 𝑄 ⊴ 𝐵, then number of
distinct choices for 𝐻 is |𝐷𝑒𝑟(𝐿/𝐵, 𝐵/𝑄)|
Proof. In 𝐿/𝑄 we have the classical configuration for which the result is well konwn. Note that in
order to define derivations we need to prescribe an action of 𝐿/𝐵 on 𝐵/𝑄. This is done by distinguishing
an 𝐻 arbitrarily (𝐻/𝑄) ≅ 𝐿/𝐵), and defining the action as conjugation by elements of 𝐻/𝑄. ◻
Let 𝐺 ≅ ℤ𝑛 ≅ ℤ ⊕ ℤ. Denote the first component o this direct sum by 𝐵. Suppose 𝐻 ≤ 𝐺, |𝐺: 𝐻| <
∞. 𝐻 give rise to 𝐿 = 𝐻𝐵 and 𝑄 = 𝐻 ∩ 𝐵, and |𝐺: 𝐿| ⋅ |𝐵: 𝑄| = |𝐺: 𝐻| < ∞. Conversely, if we are
given 𝑄 and 𝐿, 𝑄 ≤𝑓 𝐵 ≤ 𝐿 ≤𝑓 𝐺, then by Lemma 26, there exists 𝐻 ≤𝑓 𝐺 such that 𝐻𝐵 = 𝐿 and 𝐻 ∩
𝐵 = 𝑄.Suppose |𝐺: 𝐿| = 𝑛/𝑑 and |𝐵: 𝑄| = 𝑑 then |𝐺: 𝐻| = 𝑛. Lemma 27 enables us to count there are
|𝐷𝑒𝑟(𝐿/𝐵, 𝐵/𝑄) of 𝐻 associated with the pair (𝐿, 𝑄). Now 𝐺 is abelian so 𝐿/𝐵 acts trivially on 𝐵/𝑄.
Thus we have |𝐷𝑒𝑟(𝐿/𝐵, 𝐵/𝑄)| = |𝐻𝑜𝑚(ℤ𝑛−1 , 𝑐𝑑 )| = 𝑑𝑛−1. Let
∞ ∞

𝜁 (𝑧) = ∑ 𝑎𝑛 𝑛 , 𝜁ℤ𝑛−1 (𝑧) = ∑ 𝑏𝑛 𝑛−𝑧

ℤ𝑛
−𝑧

𝑛=1 𝑛=1
𝐵 has a unique subgroup of each specific index, so
𝑎𝑛 = ∑ 𝑏𝑛/𝑑 𝑑𝑛−1
𝑑∣𝑛
Therefore
𝜁ℤ𝑛 (𝑧) = 𝜁ℤ𝑛−1 𝜁(𝑧 − 𝑛 + 1)
Finally, by induction we can prove this result. ◻
𝑛 𝜋2
Corollary 4.6. Define 𝑎𝑛 = #{𝐻 ≤ ℤ2 |⬚|ℤ2 : 𝐻| = 𝑛}, then ∑ 𝑎𝑖 ∼ 12 𝑛2
𝑖=1
Proof. :
∞
Lemma 4.7. (Tauberian Theorem (Marcus du Sautoy, 1925)): Let 𝑓(𝑧) = ∑ 𝑎𝑛 𝑛−𝑧 be a Dirichlet
𝑛=1
series, 𝑎𝑛 ≥ 0 ∀𝑛 ∈ ℤ. It converges for 𝑅𝑒(𝑧) > 𝛼 > 0, If within the domain of convergence 𝑓(𝑧)
satisfies 𝑓(𝑧) = 𝑎(𝑧)(𝑧 − 𝛼)−𝑤 + 𝑏(𝑧), where 𝑎(𝑧), 𝑏(𝑧) are analytic on ℂ𝑅𝑒(𝑧)≥𝛼 , 𝑎(𝛼) ≠ 0, 𝑤 > 0,
𝑎(𝛼)
then 𝑙𝑖𝑚 ∑𝑛≤𝑥 𝑎𝑛 = (𝛼𝛤(𝑤)+𝑜(1)) 𝑥 𝛼 (𝑙𝑛𝑥)𝑤−1
𝑥→∞

198
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

𝜋2
Since 𝜁ℤ2 (𝑧) = 𝜁(𝑧)𝜁(𝑧 − 1), it converges at 𝑅𝑒(𝑧) > 2, and 𝜁(2) = 6
. Expanding 𝜁(𝑧 − 1) at
𝑧 = 2:
1
𝜁(𝑧 − 1) = + 𝛾 + 𝑜(𝑧 − 2)
𝑧−2
By Tauberian theorem, take 𝛼 = 2, 𝑤 = 1, 𝑔(𝑥) = 1, then
𝜋 2 𝑥 2 𝜋 2 𝑛2
lim ∑ 𝑎𝑛 = ∼
𝑥→∞ 12 12
𝑛≤𝑥
Example 4.8. Let 𝐹𝑛𝑐 be the free nilpotent group of rank 𝑛 and class 𝑐 , then 𝜁𝐹22 (𝑧) =
𝜁(𝑧)𝜁(𝑧−1)𝜁(2𝑧−2)𝜁(2𝑧−3)
𝜁(3𝑧−3)
Proof. It only suffices to prove that
𝜁 𝑝 (𝑧)𝜁 𝑝 (𝑧 − 1)𝜁 𝑝 (2𝑧 − 2)𝜁 𝑝 (2𝑧 − 3)
𝜁𝐹22 ,𝑝 (𝑧) =
𝜁 𝑝 (3𝑧 − 3)
Let 𝑆 = 𝛾2 (𝐺) ≅ ℤ, then 𝐺/𝑆 ≅ ℤ 𝑎𝑛𝑑𝑆 = 𝑍(𝐺). Suppose 𝑄 ≤ 𝐺, |𝐺: 𝑄| = 𝑝𝑛 . Let |𝐺: 𝑄𝑆| = 𝑝𝑎
2

and |𝑆: 𝑄 ∩ 𝑆| = 𝑝𝑏 , we have 𝑛 = 𝑎 + 𝑏 . 𝑄/𝑄 ∩ 𝑆 ≅ 𝑄𝑆/𝑆 is abelian and 𝑆/𝑄 ∩ 𝑆 is central in

𝑆𝑄/𝑄 ∩ 𝑆 so 𝑄𝑆/𝑄 ∩ 𝑆 is abelian.Conversely, suppose that 𝐾/𝐿 is abelian, then since 𝐾/𝑆 is torsion
free and 𝑆/𝐿 is finite, 𝜏(𝐾/𝐿) = 𝑆/𝐿. We choose a complement to this torsion subgroup of 𝐾/𝐿,
denoteds as 𝑄/𝐿. Then 𝑄𝑆 = 𝐾 and 𝑄 ∩ 𝑆 = 𝐿.Moreover, from Lemma 27 applied to the abelain group
𝐾/𝐿 , the number of distinct choices for 𝑄 is exactly |𝐻𝑜𝑚(𝐾/𝑆, 𝑆/𝐿)| = |𝐻𝑜𝑚(ℤ2 , 𝑐𝑝𝑏 )| = 𝑝2𝑏 .
Suppose (𝑥, 𝑦) are free generators of 𝐺, and 𝑧 = [𝑥, 𝑦]. Then 𝐾 = ⟨𝑥 𝛼 , 𝑥 𝛾 𝑦 𝛽 , 𝑧: 𝛼 > 0,0 < 𝛾 < 𝛼, 𝛽 ≥
𝑎 𝑛
0⟩ uniquely. Now, 𝑝𝑎 = 𝛼𝛽, 𝛾2 (𝐾) = ⟨𝑧 𝑝 ⟩, so 𝐾/𝐿 is abelian ⇔ 𝑏 ≤ 𝑎 ⇔ 𝑏 ≤ [2 ]Suppose
∞ ∞

𝜁𝐹22 ,𝑝 (𝑧) = ∑ 𝑔𝑖 𝑝 −𝑖𝑧

, 𝜁ℤ2 ,𝑝 (𝑧) = ∑ 𝑔𝑖′ 𝑝−𝑖𝑧
𝑖=0 𝑖=0
From the previous discussion, we deduce
[𝑛/2]
′
𝑔𝑛 = ∑ 𝑔𝑛−𝑖 𝑝2𝑖
𝑖=0
Since
𝜁ℤ2 ,𝑝 (𝑧) = 𝜁 𝑝 (𝑧)𝜁 𝑝 (𝑧 − 1) = (1−𝑧 + 𝑝−𝑧 + 𝑝−2𝑧 + ⋯ +)(1−𝑧 + 𝑝 ⋅ 𝑝−𝑧 + 𝑝2 ⋅ 𝑝−2𝑧 + ⋯ +)
= 1−𝑧 + (1 + 𝑝)𝑝−𝑧 + (1 + 𝑝 + 𝑝2 )𝑝−2𝑧 + ⋯
Thus
𝑝𝑖+1 − 1
𝑔𝑖′ = , ∀𝑖 ∈ ℤ
𝑝−1
𝑝𝑛+1 − 1 𝑝𝑛 − 1 2 𝑝𝑛+1−[𝑛/2] − 1 2[𝑛/2]
𝑔𝑛 = + 𝑝 + ⋯+ ⋅𝑝
𝑝−1 𝑝−1 𝑝−1
[𝑛/2] [𝑛/2]

(𝑝 − 1)𝑔𝑛 = 𝑝𝑛+1 ∑ 𝑝𝑖 − ∑ 𝑝2𝑖

𝑖=0 𝑖=0
Let
[𝑛/2] [𝑛/2]

𝑎𝑛 = ∑ 𝑝𝑖 , 𝑏𝑛 = ∑ 𝑝2𝑖
𝑖=0 𝑖=0
then

199
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

𝑎𝑛 = ∑ 𝑓1 (𝑝𝑖 ),where 𝑓1 (𝑝2𝑖 ) = 𝑝𝑖 , 𝑓2 (𝑝2𝑖+1 ) = 0

𝑖=0
𝑛

𝑏𝑛 = ∑ 𝑓2 (𝑝𝑖 ),where 𝑓2 (𝑝2𝑖 ) = 𝑝2𝑖 , 𝑓2 (𝑝2𝑖+1 ) = 0

𝑖=0
From the multiplication of p-Dirichlet series, we deduce
∞

∑ 𝑎𝑛 𝑝−𝑛𝑧 = 𝜁 𝑝 (𝑧)(1 + 𝑝 ⋅ 𝑝−2𝑧 + 𝑝2 ⋅ 𝑝−4𝑧 + ⋯ ) = 𝜁 𝑝 (𝑧)𝜁 𝑝 (2𝑧 − 1),

𝑛=0
∞

∑ 𝑏𝑛 𝑝−𝑛𝑧 = 𝜁 𝑝 (𝑧)(1 + 𝑝2 ⋅ 𝑝−2𝑧 + 𝑝4 ⋅ 𝑝−4𝑧 + ⋯ ) = 𝜁 𝑝 (𝑧)𝜁 𝑝 (2𝑧 − 2).

𝑛=0
Now, from [2] we know
∞ ∞
𝑝 𝑖+1 −𝑖𝑧
(𝑝 − 1)𝜁𝐹2 (𝑧) = ∑𝑝 𝑎𝑛 𝑝 − ∑ 𝑏𝑖 𝑝−𝑖𝑧
2
𝑖=0 𝑖=0
∞ ∞

= 𝑝 ∑ 𝑎𝑛 𝑝−𝑖(𝑧−1) − ∑ 𝑏𝑖 𝑝 −𝑖𝑧
𝑖=0 𝑖=0
= 𝑝𝜁 𝑝 (𝑧 − 1)𝜁 𝑝 (2𝑧 − 3) − 𝜁 𝑝 (𝑧)𝜁 𝑝 (2𝑧 − 2)
𝑝 1
= −𝑧+1 −2𝑧+3
−
(1 − 𝑝 )(1 − 𝑝 ) (1 − 𝑝 )(1 − 𝑝−2𝑧+3 )
−𝑧

𝑝(1 − 𝑝−𝑧 − 𝑝−2𝑧+2 + 𝑝−3𝑧+2 ) − (1 − 𝑝−𝑧+1 − 𝑝−2𝑧+3 + 𝑝−3𝑧+4 )

=
(1 − 𝑝−𝑧+1 )(1 − 𝑝−2𝑧+3 )(1 − 𝑝−𝑧 )(1 − 𝑝−2𝑧+2 )
= (𝑝 − 1)(1 − 𝑝−3𝑧+3 )𝜁 𝑝 (𝑧)𝜁 𝑝 (𝑧 − 1)𝜁 𝑝 (2𝑧 − 2)𝜁 𝑝 (2𝑧 − 3).
Therefore,
𝑝 𝜁 𝑝 (𝑧)𝜁 𝑝 (𝑧 − 1)𝜁 𝑝 (2𝑧 − 2)𝜁 𝑝 (2𝑧 − 3)
𝜁𝐹2 (𝑧)
=
2 𝜁 𝑝 (3𝑧 − 3)
1 𝑝 𝑞
2
The Heisenberg group 𝐻3 (ℤ) = {(0 1 𝑟 ) : 𝑝, 𝑞, 𝑟 ∈ ℤ} is a 𝐹2 groups. So using the
0 0 1
same techinique from Corollary 28 we obtain:
𝑛 𝜁(2)2 𝑛2 ln𝑛
Corollary 4.9. Let 𝑎𝑛 = #{𝐻 ≤ 𝐻3 (ℤ): |𝐻3 (ℤ): 𝐻| = 𝑛}, then ∑ 𝑎𝑛 ∼ 2𝜁(3)
𝑖=1
Grunewald F J provides a concrete formula for subgroup zeta function of 𝒥-group with Hirsch length
3(cf. (Grunewald F J, 1988)). Here are some notations might be used: for a fixed prime 𝑝, let
𝑋𝑏𝑎 = 𝑝𝑏−𝑎𝑧 , 𝑃𝑏𝑎 = (1 − 𝑋𝑏𝑎 )−1 , 𝑍𝑛 = 𝑃01 𝑃11 ⋯ 𝑃𝑛−1
1

Theorem 4.10. Let 𝐽 be a 𝒥-group with Hirsch length 3, number 𝑚 satisfies 𝑝𝑚 ∣ |𝑍(𝐽): 𝛾2 (𝐽)|, then
𝑚 )𝑍 𝑚 3 2 2
𝜁𝐺,𝑝 (𝑧) = (1 − 𝑋2𝑚 3 + 𝑋2𝑚 (1 − 𝑋3 )𝑃2 𝑃3 𝑍2
They also prove that forall 𝒥-group 𝛤, the 𝑝 −local factor of 𝜁𝛤 (𝑧) is a rational function of 𝑝−𝑧 :
Theorem 4.11. Consider a 𝒥-group 𝛤. For any prime number 𝑝, there exists polynomials 𝜙𝑝 (𝑥) and
𝜙 (𝑝−𝑧 )
𝜓𝑝 (𝑥) over ℤ, s.t.𝜁𝛤,𝑝 (𝑧) = 𝜓𝑝 (𝑝−𝑧)Additionally, 𝑑𝑒𝑔𝜙𝑝 , 𝑑𝑒𝑔𝜓𝑝 < ∞
𝑝
Using this theorem we can have a formal generalization of corollary 28 annd corollary 31, which is
proved by Grunewald and duSautoy in (Sautoy & Grunewald, 2000):
Theorem 4.12. Let 𝛤 be a 𝒥 -group, then ∃𝑏(𝛤) ∈ ℤ, 𝑏(𝛤) ≥ 0, ∃𝑐, 𝑐 ′ ∈ ℝ , such that
𝑆𝛤 (𝑛) = ∑𝑚≤𝑛 𝛼𝛤 (𝑚) ∼ 𝑐𝑛𝛼(𝛤) (𝑙𝑛𝑛)𝑏(𝛤) ,
𝑆𝛤 (𝑛)𝛼(𝛤) ∼ 𝑐 ′ (𝑙𝑛𝑛)𝑏(𝛤)+1 .

200
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

Studying the abscissa of convergence is practical and useful, especially when the zeta function is not
explicitly known. The following results from (Smith, 1983) show that the abscissa of convergence is
bounded by the Hirsch length and is invariant under group commensurability:
Theorem 4.13. Let 𝛤 be a finitely generated nilpotent group, then 𝛼𝛤 ≤ ℎ(𝛤)
Particularly, for 𝒥-group of class 2, we can find the exact upper and lower for the abscissas of
convergence:
3𝑛
Theorem 4.14. Let 𝐺 be a 𝒥-group of rank n and class 2. Denoted ℎ = ℎ(𝐺),then − 1 ≤ 𝛼𝐺 ≤
2
ℎ−1
Secondly, we can enumerate the 𝑛-dimensional irreducible complex representations of a group 𝛤,
considering isomorphism classes, which we will denote as 𝐼𝑟𝑟𝛤 (𝑛). The number of such representations
is denoted by 𝑟𝛤 (𝑛) : = #𝐼𝑟𝑟𝛤 (𝑛). If 𝛤 is rigid, i.e. ∀𝑛 ∈ ℕ, 𝑟𝛤 (𝑛) is finite, then we can define the
representation zeta function:
∞

𝜁𝛤𝑖𝑟𝑟 (𝑧) = ∑ 𝑟𝛤 (𝑛)𝑛−𝑧

𝑛=1
Our main purpose is to study the cases for 𝒥-group. The problem is unless 𝛤 is trivial, 𝛤 has infinitely
many one-dimensional representations. Thus it’s valuable to consider the twists equivalence of finite-
dimensional representations
Definition 4.15. Let 𝜌1 , 𝜌2 ∈ 𝐼𝑟𝑟𝑛 (𝛤), if ∃𝜆 ∈ 𝐼𝑟𝑟1 (𝛤), s.t. 𝜌1 ≅ 𝜌2 ⊗ 𝜆,then 𝜌1 and 𝜌2 are twist-
equivalent.
We denoted 𝑟̃𝛤 (𝑛) to be the number of n-dimensional representations up to twists equivalent.
Alexander Lubotzky, Andy R Magid, and Andy Roy Magid has proved that the ∀𝑛 ∈ ℤ, 𝑟̃𝛤 (𝑛) is
finite(cf. (Lubotzky et al., 1985)). Thus
∞

𝜁̃𝛤 (𝑧) = ∑ 𝑟̃𝛤 (𝑛)𝑛−𝑧

𝑛=1
is well defined
Ehud Hrushovski demonstrated in (Hrushovski et al., 2018) that the representation zeta function
shares many properties with the subgroup zeta function.
̃
𝑖𝑟𝑟 𝜁(𝑧−1)
Example 4.16. For 𝐻3 (ℤ), 𝜁̃𝐻(ℤ) (𝑧) =
𝜁(𝑧)
Since
∞
𝜁(𝑧 − 1)
∑ 𝜓 (𝑛)𝑛−𝑧 =
𝜁(𝑧)
𝑛=1
We have
∞ ∞
̃
𝑖𝑟𝑟 𝜁(𝑧 − 1)
𝜁𝐻(ℤ) (𝑧) = ∑ 𝑟̃𝑛 (𝛤)𝑛−𝑧 = = ∑ 𝜓 (𝑛)𝑛−𝑧
𝜁(𝑧)
𝑛=1 𝑛=1
So that the number of irreducible representations of 𝛤 up to twists equivalence is 𝜓(𝑛)

5. Conclusion
This paper investigates various important aspects of the zeta function and its applications in number
theory and group theory. We started by proving that the nontrivial zeros of 𝜁(𝑧) lie within the critical
strip, specifically between 𝑅𝑒 𝑧 = 0 and 𝑅𝑒 𝑧 = 1 on ℂ. This result, closely related to the Riemann
Hypothesis, is a central issue in analytic number theory."
Furthermore, we utilized the asymptotic behavior of 𝜁(𝑧) and its generalization, the Dirichlet L-
function, as 𝑧 approaches 1 to provide a proof of Dirichlet’s theorem. This demonstration highlights the
application of analytic techniques to address problems in number theory, illustrating how such methods
can be leveraged to uncover deep connections and solve complex issues within the field.

201
 Proceedings of the 2nd International Conference on Applied Physics and Mathematical Modeling
DOI: 10.54254/2753-8818/53/20240201

Finally, we extended our study to the realm of group theory by introducing the concept of the zeta
function of a group. Using 𝐻3 (ℤ) as a case study, we calculated both its subgroup zeta function and its
representation zeta function. These computations illustrate the utility of zeta functions in understanding
the structure and representation theory of infinite groups.

Acknowledgement
Yuxuan Shi and Zhanyu Lin contributed equally to this work and should be considered co-first authors.

References
[1] Ezzat, S. (2014). Counting irreducible representations of the heisenberg group over the integers
of a quadratic number field. Journal of Algebra, 397, 609–624.
[2] Grunewald F J, S. G. C., Segal D. (1988). Subgroups of finite index in nilpotent groups.
Inventiones Mathematicae, 93, 185–223.
[3] Grunewald, F. J., & Scharlau, R. (1979). A note on finitely generated torsion-free nilpotent groups
of class 2. Journal of Algebra, 58(1), 162–175.
[4] Hrushovski, E., Martin, B., Rideau, S., & Cluckers, R. (2018). Definable equivalence relations
and zeta functions of groups. J. Eur. Math. Soc.(JEMS), 20(10), 2467–2537.
[5] Kenneth Ireland, M. R. (2000). A classical introduction to modern number theory. Springer.
[6] Lubotzky, A., Magid, A. R., & Magid, A. R. (1985). Varieties of representations of finitely
generated groups (Vol. 336). American Mathematical Soc.
[7] lvic, A. (1949). THE RIEMANN ZETA-FUNCTION: THE THEORY OF THE RIEMANN
ZETA-FUNCTION WITH APPLICATIONS. A Wlley-Intersclence Publication.
[8] Marcus du Sautoy, L. W. (1925). Zeta functions of groups and rings. Spinger.
[9] Sautoy, M. du, & Grunewald, F. (2000). Analytic properties of zeta functions and subgroup
growth. Annals of Mathematics, 152(3), 793–833.
[10] Smith, G. C. (1983). Zeta-functions of torsion-free finitely generated nilpoten groups. The
University of Manchester (United Kingdom).
[11] Sury, B. (n.d.). Subgroup growth and zeta functions.

202