Continued Fractions and Irrationality Measures for Chowla--Selberg Gamma Quotients

Basic Information

• Paper ID: 2510.00215
• Title: Continued Fractions and Irrationality Measures for Chowla--Selberg Gamma Quotients
• Authors: Henri Cohen (Bordeaux) and Wadim Zudilin (Nijmegen)
• Classification: math.NT (Number Theory), math.CA (Classical Analysis)
• Publication Date: November 11, 2025
• Paper Link: https://arxiv.org/abs/2510.00215v3

Abstract

This paper provides 39 rapidly convergent continued fraction representations for Chowla-Selberg gamma quotients and derives good irrationality measures for 20 of them. These results include CS(-3) = (Γ(1/3)/Γ(2/3))³, a^(1/4)CS(-4) (where a=12 and a=1/5), and CS(-7) = Γ(1/7)Γ(2/7)Γ(4/7)/(Γ(3/7)Γ(5/7)Γ(6/7)). These appear to be the first reasonable irrationality measures proven for gamma quotients.

Research Background and Motivation

Core Problem

This paper investigates the irrationality measure problem for Chowla-Selberg gamma quotients CS(D), where D is a negative fundamental discriminant. These quantities are related to the values of the Dedekind eta function at CM points through the Lerch and Chowla-Selberg formulas.

Significance

1. Theoretical Importance: Although Chudnovsky and Nesterenko have proven that CS(D), π, and exp(π√D) are algebraically independent over ℚ (in particular, CS(D) is transcendental), explicit and reasonable irrationality measures are extremely rare.
2. Historical Gap: Previously, only one irrationality measure for CS(-3) based on numerical experiments existed (proposed in 12), and the bounds for μ(Γ(1/3)) and μ(Γ(1/4)) in reference 3 were prohibitively large and impractical.
3. Methodological Innovation: Traditional irrationality measures are typically obtained through integral or hypergeometric constructions (such as Mahler's work on π), whereas this paper employs continued fraction methods, which are extremely rare in this field.

Limitations of Existing Methods

• Traditional integral and series methods are not applicable to gamma quotients
• Known hypergeometric series representations converge too slowly (typically values at z=1)
• No known useful integral representations exist

Core Contributions

1. Constructed 39 rapidly convergent continued fractions of the form: $L = [[0, a_1, A(n-1)], [b_0, -K(Dn-1)(D(n-1)+1)]]$ where D ∈ {2, 3, 4, 6}.
2. Proved irrationality measures for 20 gamma quotients, including the first rigorous bounds for CS(-3), CS(-7), etc., such as:
   • μ(CS(-3)) < 5.548
   • μ(CS(-7)) < 5.283
   • μ(CS(-163)) < 2.477
3. Established a systematic theoretical framework:
   • Constructed continued fractions via Laguerre continued fractions and hypergeometric functions
   • Computed continued fraction limits using modular forms and CM theory
   • Developed Apéry-type arithmetic analysis methods for bounding denominators
4. Complete classification of CM values: Provided a complete table of 44 rational CM values, identifying which ones yield irrationality measures.

Detailed Methods

Task Definition

For an irrational number L, construct sequences of rational numbers u(n) and v(n) such that:

1. |v(n)L - u(n)| is sufficiently small: log|v(n)L - u(n)| ~ -Fn
2. Denominators are controlled: there exists d(n) such that d(n)u(n), d(n)v(n) ∈ ℤ and log(d(n)) < Mn

When F > M, an upper bound on the irrationality measure is obtained: μ(L) ≤ 1 + (F+M)/(F-M)

Core Methodological Architecture

Step 1: Laguerre Continued Fractions (Section 2)

Starting from the classical Laguerre continued fraction: $\left(\frac{1+z}{1-z}\right)^a = [[1, 1-az, 2n-1], [2az, -z^2(n^2-a^2)]]$

The convergence rate is: $\left(\frac{1+z}{1-z}\right)^a - \frac{p(n)}{q(n)} \sim \frac{2\sin(\pi a)((1+z)/(1-z))^a}{((1+\sqrt{1-z^2})/z)^{2n+1}}$

Key Technique: By "half-shifting" (replacing n with n-1/2), new continued fractions are obtained whose limits are gamma quotients.

Step 2: Construction of Continued Fraction Families (Section 3)

Study continued fraction families of the specific form: $C = [[0, a_1, A(n-1)], [b_0, -K(Dn-1)(D(n-1)+1)]]$

Convergence Analysis (Proposition 3.1):

• Let R = (A + √(A² - 4KD²))/2, E = R²/(KD²)
• When A² - 4KD² > 0, the continued fraction converges
• Convergence rate: L - p(n)/q(n) ~ C₁/E^n

Step 3: Limit Computation (Sections 4-5)

Method 1: Hypergeometric Function Representation (Theorem 4.7) $L_0 = -K^{1/2}\frac{T_1}{T_0}\left(\frac{1}{2}, 1-\frac{1}{D}; \frac{1}{2}+\frac{A}{4DK^{1/2}}\right)$

where $T_n(a,b;z) = \frac{\Gamma(a+n)\Gamma(b+n)}{\Gamma(2a+2n)}(1/z)^{a+n} \cdot {}_2F_1(a+n, b+n; 2a+2n; 1/z)$

Method 2: Modular Forms Theory (Section 5)

Utilizing the Hauptmodul of triangle groups (p,p,∞) and modular hypergeometric evaluations:

• (3,3,∞): $_2F_1(1/2, 5/6; 1; 1/(1-R_1(τ))) = (F_{1,-}^{1/2}/F_{1,+}^{1/3})(τ)$
• (4,4,∞): $_2F_1(1/2, 3/4; 1; 1/(1-R_2(τ))) = (F_{2,-}^{1/2}/F_{2,+}^{1/4})(τ)$
• (6,6,∞): $_2F_1(1/2, 2/3; 1; 1/(1-R_3(τ))) = (F_{3,-}^{1/2}/F_{3,+}^{1/6})(τ)$

By choosing CM points τ such that (2R_N(τ)-1)² is rational, combined with the Chowla-Selberg formula to compute the fundamental period Ω(τ), explicit expressions for the continued fraction limits are obtained.

Technical Innovations

1. Systematic Application of Half-Shifting: By replacing n with n-1/2 in the Laguerre continued fraction, continued fractions for gamma quotients are systematically derived from those of algebraic numbers.
2. Apéry-Type Recurrence: The key recurrence has the form $(n+B)v_{n+1} = Znv_n + K(B-n)v_{n-1}$ where coefficients are linear functions of n (rather than general quadratic functions), making denominator analysis possible.
3. Explicit Formula (Theorem 7.1): Provides exact expressions for convergents p(n)/q(n) involving truncated hypergeometric sums: $P_n = \sum_{j=0}^{\lfloor n/2\rfloor} (-1)^j K^{n-j}(Z/K)^{n-2j}\frac{(n-j)!}{(n-2j)!}\sum_{i=0}^j \frac{(-1)^i(n-i)!}{i!^2(j-i)!}\Lambda_i$
4. Optimized Denominator Bounds: Introduces d*_D(n) to replace the naive d_D(n) = lcm(Dj+1), significantly improving denominator estimates through excluding primes p satisfying specific conditions (set P_n).

Experimental Setup

Data Sources

This is a pure theoretical mathematics work; "experiments" primarily refer to:

1. Numerical Verification: Using high-precision computation to verify continued fraction convergence to expected gamma quotients
2. CM Value Enumeration: Based on the complete classification in 2, comprising 44 rational CM values
3. OEIS Identification: Utilizing the Online Encyclopedia of Integer Sequences to identify numerical coefficients

Evaluation Metrics

• Convergence Rate E: log|E|/2 determines convergence speed
• Denominator Growth Rate m*_D: Asymptotically log(d_D(n)) ~ m_D · n
• Irrationality Measure μ(L): When log|E|/2 > m*_D, a measure is obtained

Key Parameters

• m*_2 = 2
• m*_3 = 3 - π/(2√3) ≈ 2.093
• m*_4 = 4 - π/2 ≈ 2.429
• m*_6 = 6 - π√3/2 ≈ 3.279

Experimental Results

Main Results (Theorem 6.2)

Best Irrationality Measures:

• μ(CS(-163)) < 2.477 (closest to optimal value 2)
• μ(CS(-232)) < 2.652
• μ(CS(-267)) < 2.869
• μ(CS(-67)) < 3.002

Important Special Values:

• μ(CS(-3)) < 5.548 (first rigorous proof, improving over numerical result 13.418)
• μ(CS(-7)) < 5.283
• μ(12^(1/4)CS(-4)) < 25.733

Corollaries:

• μ(Γ(1/3)/Γ(2/3)) < 16.644

Continued Fraction Examples

Rapidly Convergent Continued Fraction for CS(-3) (Label 1.4): $CS(-3) = [[0, 31, 1012(n-1)], [240, -(6n-1)(6n-5)]]$

Convergence rate: E = (16 + 5√10)⁴/36 ≈ 28446.4, i.e., $CS(-3) - \frac{p(n)}{q(n)} \sim \frac{3^{3/2}CS(-3)}{(16+5\sqrt{10})^{4n}6^{-2n}}$

Continued Fraction for CS(-7) (Label 1.5): $CS(-7) = [[324, 10773, \frac{10773}{2}(n-1)], [3570, -84(6n-1)(6n-5)]]$

Completeness Analysis

• Among 44 rational CM values, 39 yield convergent continued fractions
• Of these, 20 converge sufficiently rapidly to obtain irrationality measures
• The remaining 19, though convergent, satisfy log|E|/2 ≤ m*_D, preventing measure derivation
• 5 continued fractions diverge (A² - 4KD² < 0)

Case Study: CM Value at z = 128/3

For τ = (-3 + 3√(-3))/2, we have 1 - R₁(τ) = 128/3. Choosing Ω(τ) = e^(iπ/4)η(τ)²:

Computational Results:

• Ω = 3^(-19/12)Γ(1/3)/Γ(2/3)²
• R₁ = -125/3
• $_2F_1(1/2, 5/6; 1; 3/128) = 2^{25/6}3^{-3/2}5^{-1}\frac{\Gamma(1/3)}{\Gamma(2/3)^2}$

By computing $_2F_1(3/2, 11/6; 3; 3/128)$ via Serre derivatives and modularity properties, the continued fraction is ultimately proven to converge to CS(-3).

Related Work

Classical Methods for Irrationality Measures

1. Mahler's Measure of π: Earliest use of integral methods
2. Apéry's ζ(3) and π²: Using continued fractions, later reinterpreted as integrals by Beukers
3. Zeilberger-Zudilin's Measure of π: Current best, using hypergeometric constructions

Related Work on Gamma Functions

• Chudnovsky 5 and Nesterenko 14,17: Proved transcendence of CS(D)
• Bruiltet 3: Prohibitively large bounds for Γ(1/3) and Γ(1/4) (impractical)
• Dougherty-Bliss et al. 12: Numerical measure for CS(-3) (not rigorously proven)

Modular Forms and Hypergeometric Functions

• Fricke Formula: $_2F_1(1/12, 7/12; 1; 1728/(1728-j(τ))) = E_6^{1/6}(τ)$
• Beukers-Cohen 2: Hauptmoduli and CM value classification for arithmetic triangle groups
• Cohen-Guillera 9: Rational hypergeometric Ramanujan identities for 1/π

Relationship to Related Work

• Similar methodology to 19 (recent work by the second author) but applied to different problems
• Related to 10 (another work by the authors) on continued fractions for CS(-3)² and CS(-4)², but more systematic
• First rigorous and reasonable irrationality measures for gamma quotients

Conclusions and Discussion

Main Conclusions

1. First Rigorous Proof: Provides the first batch of rigorous and reasonable irrationality measures for 20 gamma quotients
2. Methodological Innovation: Demonstrates that continued fraction methods can be superior to traditional integral/series methods in certain cases
3. Theoretical Completeness: Provides a complete theoretical framework for continued fractions of specific forms
4. Result Optimality: Numerical evidence suggests denominator bounds are asymptotically optimal, making further improvement of irrationality measures difficult

Limitations

1. Restricted to Special Forms: Method applies only to continued fractions where a(n) is linear and b(n) is quadratic
2. Restriction on D Values: Only handles D ∈ {2, 3, 4, 6}
3. Finite CM Values: The finiteness of rational CM values limits the discriminants that can be handled
4. Missing Values: CS(-4), CS(-8), CS(-88) appear only as algebraic multiples
5. CS(-11) Lacks Measure: Although a convergent continued fraction exists, convergence is insufficient

Future Directions

1. Compact Triangle Groups: Explore compact arithmetic triangle groups (p,q,r), potentially obtaining more examples through automorphic forms on Shimura curves
2. Deeper Continued Fractions: Study continued fractions with higher degrees for a(n) and b(n), such as those for CS(-3)², CS(-4)², CS(-8)² mentioned in the paper: $(2^{1/3}CS(-3))^2 = [[72, 33, 40n^2+2], [648, -9(2n+1)^4]]$
3. Connection to 1/π²: Explore relationships with rational hypergeometric formulas, potentially obtaining continued fractions for 15 CS value squares
4. p-adic Analogues: The Gross-Koblitz formula is the p-adic version of the Chowla-Selberg formula; parallel theory development may be possible
5. Integral Representations: Seek useful integral or series representations for gamma quotients

In-Depth Evaluation

Strengths

1. Major Breakthrough: Achieves a significant breakthrough on the long-standing problem of irrationality measures for gamma quotients, filling an important gap in the field
2. Original Methodology:
   • Application of continued fraction methods to irrationality measures is extremely rare
   • Systematic application of half-shifting technique is innovative
   • The framework combining modular forms theory and arithmetic analysis is elegant
3. Theoretical Completeness:
   • Forms a complete closed loop from continued fraction construction through limit computation to denominator bounding
   • Provides explicit formulas (Theorem 7.1) rather than merely existence results
   • Complete enumeration of all possible CM values (Table 1)
4. Technical Depth:
   • Identification and utilization of Apéry-type recurrences demonstrates profound insight
   • Denominator optimization (d*_D(n)) through refined prime sieving significantly improves results
   • Technical proof of Lemma 7.4 demonstrates sophisticated techniques
5. Clear Exposition:
   • Section 1.2's motivating example excellently guides readers
   • Detailed CS(-3) computation case (Section 6.1) aids understanding
   • Well-organized tables facilitate reference

Weaknesses

1. Limited Applicability:
   • Applies only to continued fractions of specific forms and D values
   • No obvious generalization path to arbitrary gamma quotients
   • Only 20 of 44 CM values yield measures
2. Suboptimal Results in Some Cases:
   • μ(12^(1/4)CS(-4)) < 25.733 is quite large
   • μ(CS(-51)) < 2598.5 is practically useless
   • Still gaps from theoretical optimal value 2
3. Insufficient Theoretical Explanation:
   • Why do these specific CM values produce rapidly convergent continued fractions?
   • Deep reasons for half-shifting not fully clarified
   • Possible connections to Hilbert modular forms only briefly mentioned
4. Computational Details:
   • Some derivations (e.g., Theorem 7.1) claimed "easily verified" but actually complex
   • Precision and methods of numerical verification not detailed
   • Specific values of constants C₁, C₂ not provided
5. p-adic Generalization:
   • Although p-adic direction proposed, specific approaches not provided
   • Application method of Gross-Koblitz formula unclear

Impact

1. Theoretical Contribution:
   • Opens new direction of studying gamma quotients via continued fractions
   • Demonstrates continued fractions can outperform integral methods for certain transcendental problems
   • Provides new application scenario for Apéry-type methods
2. Practical Value:
   • Irrationality measures have applications in Diophantine approximation
   • Continued fraction formulas themselves useful for high-precision numerical computation
   • Methods potentially generalizable to other special function values
3. Reproducibility:
   • Theoretical derivations detailed and verifiable
   • Tables 1 and 2 provide complete data
   • Dependent reference 2, though unpublished, available from authors
4. Follow-up Research:
   • Already inspired follow-up work 19 by the second author
   • Forms a series with 10 (another work by the authors)
   • Provides template for compact triangle group direction

Applicable Scenarios

1. Number Theory Research:
   • Irrationality and algebraic independence in transcendental number theory
   • Diophantine approximation theory
   • Arithmetic properties of special values
2. Modular Forms Theory:
   • Applications of CM theory
   • Evaluation of modular hypergeometric functions
   • Study of arithmetic triangle groups
3. Computational Mathematics:
   • High-precision computation of gamma quotients
   • Implementation of continued fraction algorithms
   • Numerical verification of special function values
4. Related Fields:
   • Potentially insightful for periods of elliptic curves and abelian varieties
   • Possible relevance to Feynman integrals in quantum field theory
   • Potential applications to generating functions in combinatorics

Key References

2 F. Beukers and H. Cohen, Arithmetic Triangle Groups, Hauptmoduln, Complex Multiplication, and Hypergeometric Functions (in preparation) - Provides foundational CM value classification

5 G.V. Chudnovsky (1976) - Proves algebraic independence of CS(D)

10 H. Cohen and W. Zudilin (2025), Variations on a theme of Apéry - Related Apéry-type continued fraction research

12 R. Dougherty-Bliss, Ch. Koutschan and D. Zeilberger (2022) - Numerical measure for CS(-3)

14 Yu.V. Nesterenko (1996) - Modular functions and transcendence problems

19 W. Zudilin (2025), Linear independence measures for Chowla–Selberg periods - Related work by second author

Overall Assessment: This is an outstanding number theory paper achieving a major breakthrough on the long-standing problem of irrationality measures for gamma quotients. The methodology is novel, the theory complete, and the results significant. While applicable scope is limited, within its domain the results are nearly optimal. The paper demonstrates profound synthesis of continued fractions, modular forms, hypergeometric functions, and arithmetic analysis, providing valuable methodological contributions to related fields. Subsequent research should focus on generalization to broader cases and connections with other mathematical branches.